id stringlengths 16 29 | text stringlengths 86 3.49k | source stringlengths 14 112 |
|---|---|---|
arxiv_dataset-186002306.06852 | Rethink DARTS Search Space and Renovate a New Benchmark
cs.LG
DARTS search space (DSS) has become a canonical benchmark for NAS whereas
some emerging works pointed out the issue of narrow accuracy range and claimed
it would hurt the method ranking. We observe some recent studies already suffer
from this issue that overshadows the meaning of scores. In this work, we first
propose and orchestrate a suite of improvements to frame a larger and harder
DSS, termed LHD, while retaining high efficiency in search. We step forward to
renovate a LHD-based new benchmark, taking care of both discernibility and
accessibility. Specifically, we re-implement twelve baselines and evaluate them
across twelve conditions by combining two underexpolored influential factors:
transductive robustness and discretization policy, to reasonably construct a
benchmark upon multi-condition evaluation. Considering that the tabular
benchmarks are always insufficient to adequately evaluate the methods of neural
architecture search (NAS), our work can serve as a crucial basis for the future
progress of NAS. https://github.com/chaoji90/LHD
| arxiv topic:cs.LG |
arxiv_dataset-186012306.06952 | Numerically stable neural network for simulating Kardar-Parisi-Zhang
growth in the presence of uncorrelated and correlated noises
physics.comp-ph cond-mat.stat-mech
Numerical simulations are essential tools for exploring the dynamic scaling
properties of the nonlinear Kadar-Parisi-Zhang (KPZ) equation. Yet the inherent
nonlinearity frequently causes numerical divergence within the strong-coupling
regime using conventional simulation methods. To sustain the numerical
stability, previous works either utilized discrete growth models belonging to
the KPZ universality class or modified the original nonlinear term by the
designed specified operators. However, recent studies revealed that these
strategies could cause abnormal results. Motivated by the above-mentioned
facts, we propose a convolutional neural network-based method to simulate the
KPZ equation driven by uncorrelated and correlated noises, aiming to overcome
the challenge of numerical divergence, and obtaining reliable scaling
exponents. We first train the neural network to represent the determinant terms
of the KPZ equation in a data-driven manner. Then, we perform simulations for
the KPZ equation with various types of temporally and spatially correlated
noises. The experimental results demonstrate that our neural network could
effectively estimate the scaling exponents eliminating numerical divergence.
| arxiv topic:physics.comp-ph cond-mat.stat-mech |
arxiv_dataset-186022306.07052 | Gradient Ascent Post-training Enhances Language Model Generalization
cs.CL cs.AI
In this work, we empirically show that updating pretrained LMs (350M, 1.3B,
2.7B) with just a few steps of Gradient Ascent Post-training (GAP) on random,
unlabeled text corpora enhances its zero-shot generalization capabilities
across diverse NLP tasks. Specifically, we show that GAP can allow LMs to
become comparable to 2-3x times larger LMs across 12 different NLP tasks. We
also show that applying GAP on out-of-distribution corpora leads to the most
reliable performance improvements. Our findings indicate that GAP can be a
promising method for improving the generalization capability of LMs without any
task-specific fine-tuning.
| arxiv topic:cs.CL cs.AI |
arxiv_dataset-186032306.07152 | Measuring Sentiment Bias in Machine Translation
cs.CL
Biases induced to text by generative models have become an increasingly large
topic in recent years. In this paper we explore how machine translation might
introduce a bias in sentiments as classified by sentiment analysis models. For
this, we compare three open access machine translation models for five
different languages on two parallel corpora to test if the translation process
causes a shift in sentiment classes recognized in the texts. Though our
statistic test indicate shifts in the label probability distributions, we find
none that appears consistent enough to assume a bias induced by the translation
process.
| arxiv topic:cs.CL |
arxiv_dataset-186042306.07252 | On the Validity of Conformal Prediction for Network Data Under
Non-Uniform Sampling
math.ST stat.ME stat.ML stat.TH
We study the properties of conformal prediction for network data under
various sampling mechanisms that commonly arise in practice but often result in
a non-representative sample of nodes. We interpret these sampling mechanisms as
selection rules applied to a superpopulation and study the validity of
conformal prediction conditional on an appropriate selection event. We show
that the sampled subarray is exchangeable conditional on the selection event if
the selection rule satisfies a permutation invariance property and a joint
exchangeability condition holds for the superpopulation. Our result implies the
finite-sample validity of conformal prediction for certain selection events
related to ego networks and snowball sampling. We also show that when data are
sampled via a random walk on a graph, a variant of weighted conformal
prediction yields asymptotically valid prediction sets for an independently
selected node from the population.
| arxiv topic:math.ST stat.ME stat.ML stat.TH |
arxiv_dataset-186052306.07352 | Multi-Platform Budget Management in Ad Markets with Non-IC Auctions
cs.GT cs.LG math.OC stat.ML
In online advertising markets, budget-constrained advertisers acquire ad
placements through repeated bidding in auctions on various platforms. We
present a strategy for bidding optimally in a set of auctions that may or may
not be incentive-compatible under the presence of budget constraints. Our
strategy maximizes the expected total utility across auctions while satisfying
the advertiser's budget constraints in expectation. Additionally, we
investigate the online setting where the advertiser must submit bids across
platforms while learning about other bidders' bids over time. Our algorithm has
$O(T^{3/4})$ regret under the full-information setting. Finally, we demonstrate
that our algorithms have superior cumulative regret on both synthetic and
real-world datasets of ad placement auctions, compared to existing adaptive
pacing algorithms.
| arxiv topic:cs.GT cs.LG math.OC stat.ML |
arxiv_dataset-186062306.07452 | A remark on Oka's lemma and a geometric property of pseudoconvex domains
math.CV
A direct proof of Oka's lemma on the relation of holomorphic convexity and
the properties of the distance to the boundary function is provided. Some
related problems for subharmonicity properties of this function are also
studied. A new geometric property of pseudoconvex domains is described.
| arxiv topic:math.CV |
arxiv_dataset-186072306.07552 | Galactic: Scaling End-to-End Reinforcement Learning for Rearrangement at
100k Steps-Per-Second
cs.LG cs.AI cs.RO
We present Galactic, a large-scale simulation and reinforcement-learning (RL)
framework for robotic mobile manipulation in indoor environments. Specifically,
a Fetch robot (equipped with a mobile base, 7DoF arm, RGBD camera, egomotion,
and onboard sensing) is spawned in a home environment and asked to rearrange
objects - by navigating to an object, picking it up, navigating to a target
location, and then placing the object at the target location.
Galactic is fast. In terms of simulation speed (rendering + physics),
Galactic achieves over 421,000 steps-per-second (SPS) on an 8-GPU node, which
is 54x faster than Habitat 2.0 (7699 SPS). More importantly, Galactic was
designed to optimize the entire rendering + physics + RL interplay since any
bottleneck in the interplay slows down training. In terms of simulation+RL
speed (rendering + physics + inference + learning), Galactic achieves over
108,000 SPS, which 88x faster than Habitat 2.0 (1243 SPS).
These massive speed-ups not only drastically cut the wall-clock training time
of existing experiments, but also unlock an unprecedented scale of new
experiments. First, Galactic can train a mobile pick skill to >80% accuracy in
under 16 minutes, a 100x speedup compared to the over 24 hours it takes to
train the same skill in Habitat 2.0. Second, we use Galactic to perform the
largest-scale experiment to date for rearrangement using 5B steps of experience
in 46 hours, which is equivalent to 20 years of robot experience. This scaling
results in a single neural network composed of task-agnostic components
achieving 85% success in GeometricGoal rearrangement, compared to 0% success
reported in Habitat 2.0 for the same approach. The code is available at
github.com/facebookresearch/galactic.
| arxiv topic:cs.LG cs.AI cs.RO |
arxiv_dataset-186082306.07652 | Inactivated COVID-19 Vaccination did not affect In vitro fertilization
(IVF) / Intra-Cytoplasmic Sperm Injection (ICSI) cycle outcomes
stat.AP q-bio.TO
Background: The objective of this study is to evaluate the impact of COVID-19
inactivated vaccine administration on the outcomes of in vitro fertilization
(IVF) and intracytoplasmic sperm injection (ICSI) cycles in infertile couples
in China. Methods: We collected data from the CYART prospective cohort, which
included couples undergoing IVF treatment from January 2021 to September 2022
at Sichuan Jinxin Xinan Women & Children's Hospital. Based on whether they
received vaccination before ovarian stimulation, the couples were divided into
the vaccination group and the non-vaccination group. We compared the laboratory
parameters and pregnancy outcomes between the two groups. Findings: After
performing propensity score matching (PSM), the analysis demonstrated similar
clinical pregnancy rates, biochemical pregnancy and ongoing pregnancy rates
between vaccinated and unvaccinated women. No significant disparities were
found in terms of embryo development and laboratory parameters among the
groups. Moreover, male vaccination had no impact on patient performance or
pregnancy outcomes in assisted reproductive technology treatments.
Additionally, there were no significant differences observed in the effects of
vaccination on embryo development and pregnancy outcomes among couples
undergoing ART. Interpretation: The findings suggest that COVID-19 vaccination
did not have a significant effect on patients undergoing IVF/ICSI with fresh
embryo transfer. Therefore, it is recommended that couples should receive
COVID-19 vaccination as scheduled to help mitigate the COVID-19 pandemic.
| arxiv topic:stat.AP q-bio.TO |
arxiv_dataset-186092306.07752 | The Haldane Model with Chiral Edge States using a Synthetic Dimension
cond-mat.quant-gas quant-ph
We explicitly show that the differences, with respect to the appearance of
topological phases, between the traditional Haldane model, which utilises a
honeycomb lattice structure, to that of the Haldane model imbued onto a
brick-wall lattice geometry, are inconsequential. A proposal is then put
forward to realise the Haldane model by exploiting the internal degrees of
freedom of atoms as a synthetic dimension. This leads to a convenient platform
for the investigation of chiral edge states due to the hard boundaries provided
by the hyperfine manifold. We make some cursory comments on the effects of
interactions in the system.
| arxiv topic:cond-mat.quant-gas quant-ph |
arxiv_dataset-186102306.07852 | Globally convergent homotopies for discrete-time optimal control
math.OC cs.SY eess.SY
Homotopy methods are attractive due to their capability of solving difficult
optimisation and optimal control problems. The underlying idea is to construct
a homotopy, which may be considered as a continuous (zero) curve between the
difficult original problem and a related, comparatively easy one. Then, the
solution of the easier one is continuously perturbed along the zero curve
towards the sought-after solution of the original problem. We propose a
methodology for the systematic construction of such zero curves for
discrete-time optimal control problems drawing upon the theory of globally
convergent homotopies for nonlinear programs. The proposed framework ensures
that for almost every initial guess at a solution there exists a suitable
homotopy path that is, in addition, numerically convenient to track. We
demonstrate the results by solving optimal path planning problems for a linear
system and the nonlinear nonholonomic car (Dubins' vehicle).
| arxiv topic:math.OC cs.SY eess.SY |
arxiv_dataset-186112306.07952 | MOFI: Learning Image Representations from Noisy Entity Annotated Images
cs.CV cs.CL cs.LG
We present MOFI, Manifold OF Images, a new vision foundation model designed
to learn image representations from noisy entity annotated images. MOFI differs
from previous work in two key aspects: (i) pre-training data, and (ii) training
recipe. Regarding data, we introduce a new approach to automatically assign
entity labels to images from noisy image-text pairs. Our approach involves
employing a named entity recognition model to extract entities from the
alt-text, and then using a CLIP model to select the correct entities as labels
of the paired image. It's a simple, cost-effective method that can scale to
handle billions of web-mined image-text pairs. Through this method, we have
created Image-to-Entities (I2E), a new dataset with 1 billion images and 2
million distinct entities, covering rich visual concepts in the wild. Building
upon the I2E dataset, we study different training recipes like supervised
pre-training, contrastive pre-training, and multi-task learning. For
contrastive pre-training, we treat entity names as free-form text, and further
enrich them with entity descriptions. Experiments show that supervised
pre-training with large-scale fine-grained entity labels is highly effective
for image retrieval tasks, and multi-task training further improves the
performance. The final MOFI model achieves 86.66% mAP on the challenging
GPR1200 dataset, surpassing the previous state-of-the-art performance of 72.19%
from OpenAI's CLIP model. Further experiments on zero-shot and linear probe
image classification also show that MOFI outperforms a CLIP model trained on
the original image-text data, demonstrating the effectiveness of the I2E
dataset in learning strong image representations. We release our code and model
weights at https://github.com/apple/ml-mofi.
| arxiv topic:cs.CV cs.CL cs.LG |
arxiv_dataset-186122306.08052 | On coloring parameters of triangle-free planar $(n,m)$-graphs
math.CO cs.DM
An $(n,m)$-graph is a graph with $n$ types of arcs and $m$ types of edges. A
homomorphism of an $(n,m)$-graph $G$ to another $(n,m)$-graph $H$ is a vertex
mapping that preserves the adjacencies along with their types and directions.
The order of a smallest (with respect to the number of vertices) such $H$ is
the $(n,m)$-chromatic number of $G$.Moreover, an $(n,m)$-relative clique $R$ of
an $(n,m)$-graph $G$ is a vertex subset of $G$ for which no two distinct
vertices of $R$ get identified under any homomorphism of $G$. The
$(n,m)$-relative clique number of $G$, denoted by $\omega_{r(n,m)}(G)$, is the
maximum $|R|$ such that $R$ is an $(n,m)$-relative clique of $G$. In practice,
$(n,m)$-relative cliques are often used for establishing lower bounds of
$(n,m)$-chromatic number of graph families.
Generalizing an open problem posed by Sopena [Discrete Mathematics 2016] in
his latest survey on oriented coloring, Chakroborty, Das, Nandi, Roy and Sen
[Discrete Applied Mathematics 2022] conjectured that $\omega_{r(n,m)}(G) \leq 2
(2n+m)^2 + 2$ for any triangle-free planar $(n,m)$-graph $G$ and that this
bound is tight for all $(n,m) \neq (0,1)$.In this article, we positively settle
this conjecture by improving the previous upper bound of $\omega_{r(n,m)}(G)
\leq 14 (2n+m)^2 + 2$ to $\omega_{r(n,m)}(G) \leq 2 (2n+m)^2 + 2$, and by
finding examples of triangle-free planar graphs that achieve this bound. As a
consequence of the tightness proof, we also establish a new lower bound of $2
(2n+m)^2 + 2$ for the $(n,m)$-chromatic number for the family of triangle-free
planar graphs.
| arxiv topic:math.CO cs.DM |
arxiv_dataset-186132306.08152 | QFactor: A Domain-Specific Optimizer for Quantum Circuit Instantiation
quant-ph
We introduce a domain-specific algorithm for numerical optimization
operations used by quantum circuit instantiation, synthesis, and compilation
methods. QFactor uses a tensor network formulation together with analytic
methods and an iterative local optimization algorithm to reduce the number of
problem parameters. Besides tailoring the optimization process, the formulation
is amenable to portable parallelization across CPU and GPU architectures, which
is usually challenging in general purpose optimizers (GPO). Compared with
several GPOs, our algorithm achieves exponential memory and performance savings
with similar optimization success rates. While GPOs can handle directly
circuits of up to six qubits, QFactor can process circuits with more than 12
qubits. Within the BQSKit optimization framework, we enable optimizations of
100+ qubit circuits using gate deletion algorithms to scale out linearly with
the hardware resources allocated for compilation in GPU environments.
| arxiv topic:quant-ph |
arxiv_dataset-186142306.08252 | GraphVine: A Data Structure to Optimize Dynamic Graph Processing on GPUs
cs.DS cs.DC
Graph processing on GPUs is gaining momentum due to the high throughputs
observed compared to traditional CPUs, attributed to the vast number of
processing cores on GPUs that can exploit parallelism in graph analytics. This
paper discusses a graph data structure for dynamic graph processing on GPUs.
Unlike static graphs, dynamic graphs mutate over their lifetime through vertex
and/or edge batch updates. The proposed work aims to provide fast batch updates
and graph querying without consuming too much GPU memory. Experimental results
show improved initialization timings by 1968-1269024%, improved batch edge
insert timings by 30-30047%, and improved batch edge delete timings by
50-25262% while consuming less memory when the batch size is large.
| arxiv topic:cs.DS cs.DC |
arxiv_dataset-186152306.08352 | Bayesian Non-linear Latent Variable Modeling via Random Fourier Features
stat.ML cs.AI cs.LG
The Gaussian process latent variable model (GPLVM) is a popular probabilistic
method used for nonlinear dimension reduction, matrix factorization, and
state-space modeling. Inference for GPLVMs is computationally tractable only
when the data likelihood is Gaussian. Moreover, inference for GPLVMs has
typically been restricted to obtaining maximum a posteriori point estimates,
which can lead to overfitting, or variational approximations, which
mischaracterize the posterior uncertainty. Here, we present a method to perform
Markov chain Monte Carlo (MCMC) inference for generalized Bayesian nonlinear
latent variable modeling. The crucial insight necessary to generalize GPLVMs to
arbitrary observation models is that we approximate the kernel function in the
Gaussian process mappings with random Fourier features; this allows us to
compute the gradient of the posterior in closed form with respect to the latent
variables. We show that we can generalize GPLVMs to non-Gaussian observations,
such as Poisson, negative binomial, and multinomial distributions, using our
random feature latent variable model (RFLVM). Our generalized RFLVMs perform on
par with state-of-the-art latent variable models on a wide range of
applications, including motion capture, images, and text data for the purpose
of estimating the latent structure and imputing the missing data of these
complex data sets.
| arxiv topic:stat.ML cs.AI cs.LG |
arxiv_dataset-186162306.08452 | Perfect plasticity versus damage: an unstable interaction between
irreversibility and $\Gamma$-convergence through variational evolutions
math.AP
This paper addresses the question of the interplay between relaxation and
irreversibility through quasi-static evolutions in damage mechanics, by
inquiring the following question: can the quasi-static evolution of an elastic
material undergoing a rate-independent process of plastic deformation be
derived as the limit model of a sequence of quasi-static brittle damage
evolutions? This question is motivated by the static analysis performed by
Babadjian-Iurlano-Rindler, where they have shown how the brittle damage model
introduced by Francfort and Marigo can lead to a model of Hencky perfect
plasticity. Problems of damage mechanics being rather described through
evolution processes, it is natural to extend this analysis to quasi-static
evolutions, where the inertia is neglected. We consider the case where the
medium is subjected to time-dependent boundary conditions, in the
one-dimensional setting. The idea is to combine the scaling law considered by
Babadjian-Iurlano-Rindler with the quasi-static brittle damage evolution
introduced by Francfort and Garroni, and try to understand how the
irreversibility of the damage process will be expressed in the limit evolution.
Surprisingly, the interplay between relaxation and irreversibility is not
stable through time evolutions. Indeed, depending on the choice of the
prescribed Dirichlet boundary condition, the effective quasi-static damage
evolution obtained may not be of perfect plasticity type.
| arxiv topic:math.AP |
arxiv_dataset-186172306.08552 | Transversal transport and topological properties of binary
heterostructures of topological insulators
cond-mat.mes-hall
This paper discusses the topological and transport properties of binary
heterostructures of different topological materials. The creation of multilayer
devices is an alternative to building synthetic topological materials. By
adjusting the pattern of layers, we control the global topological properties
that favor tunneling and optimize the conductance of the edge state. Using a
one-dimensional model and the method of Green's functions, we characterize each
layer's edge states and make the chains couplings to generate the
heterostructure. To study the bulk properties, we calculate the topological
invariant from the Zak phase to build phase diagrams, and we obtain an
analytical result for the separation line between different phases that depends
on the hopping parameters of the heterostructure. We calculate the differential
conductance with the non-equilibrium Green function technique showing the
tunneling of the edge states and discussing its possible design and
experimental application.
| arxiv topic:cond-mat.mes-hall |
arxiv_dataset-186182306.08652 | Staircasing effect for minimizers of the one-dimensional discrete
Perona-Malik functional
math.AP math.FA math.OC
We consider the one-dimensional Perona-Malik functional, that is the energy
associated to the celebrated forward-backward equation introduced by P. Perona
and J. Malik in the context of image processing, with the addition of a forcing
term. We discretize the functional by restricting its domain to a finite
dimensional space of piecewise constant functions, and by replacing the
derivative with a difference quotient.
We investigate the asymptotic behavior of minima and minimizers as the
discretization scale vanishes. In particular, if the forcing term has bounded
variation, we show that any sequence of minimizers converges in the sense of
varifolds to the graph of the forcing term, but with tangent component which is
a combination of the horizontal and vertical directions.
If the forcing term is more regular, we also prove that minimizers actually
develop a microstructure that looks like a piecewise constant function at a
suitable scale, which is intermediate between the macroscopic scale and the
scale of the discretization.
| arxiv topic:math.AP math.FA math.OC |
arxiv_dataset-186192306.08752 | Dynamics of infectious diseases in predator-prey populations: a
stochastic model, sustainability, and invariant measure
q-bio.PE math.DS
This paper introduces an innovative model for infectious diseases in
predator-prey populations. We not only prove the existence of global
non-negative solutions but also establish essential criteria for the system's
decline and sustainability. Furthermore, we demonstrate the presence of a Borel
invariant measure, adding a new dimension to our understanding of the system.
To illustrate the practical implications of our findings, we present numerical
results. With our model's comprehensive approach, we aim to provide valuable
insights into the dynamics of infectious diseases and their impact on
predator-prey populations.
| arxiv topic:q-bio.PE math.DS |
arxiv_dataset-186202306.08852 | BED: Bi-Encoder-Based Detectors for Out-of-Distribution Detection
cs.CL
This paper introduces a novel method leveraging bi-encoder-based detectors
along with a comprehensive study comparing different out-of-distribution (OOD)
detection methods in NLP using different feature extractors. The feature
extraction stage employs popular methods such as Universal Sentence Encoder
(USE), BERT, MPNET, and GLOVE to extract informative representations from
textual data. The evaluation is conducted on several datasets, including
CLINC150, ROSTD-Coarse, SNIPS, and YELLOW. Performance is assessed using
metrics such as F1-Score, MCC, FPR@90, FPR@95, AUPR, an AUROC. The experimental
results demonstrate that the proposed bi-encoder-based detectors outperform
other methods, both those that require OOD labels in training and those that do
not, across all datasets, showing great potential for OOD detection in NLP. The
simplicity of the training process and the superior detection performance make
them applicable to real-world scenarios. The presented methods and benchmarking
metrics serve as a valuable resource for future research in OOD detection,
enabling further advancements in this field. The code and implementation
details can be found on our GitHub repository:
https://github.com/yellowmessenger/ood-detection.
| arxiv topic:cs.CL |
arxiv_dataset-186212306.08952 | Towards Benchmarking and Improving the Temporal Reasoning Capability of
Large Language Models
cs.CL cs.AI
Reasoning about time is of fundamental importance. Many facts are
time-dependent. For example, athletes change teams from time to time, and
different government officials are elected periodically. Previous
time-dependent question answering (QA) datasets tend to be biased in either
their coverage of time spans or question types. In this paper, we introduce a
comprehensive probing dataset \tempreason to evaluate the temporal reasoning
capability of large language models. Our dataset includes questions of three
temporal reasoning levels. In addition, we also propose a novel learning
framework to improve the temporal reasoning capability of large language
models, based on temporal span extraction and time-sensitive reinforcement
learning. We conducted experiments in closed book QA, open book QA, and
reasoning QA settings and demonstrated the effectiveness of our approach. Our
code and data are released on https://github.com/DAMO-NLP-SG/TempReason.
| arxiv topic:cs.CL cs.AI |
arxiv_dataset-186222306.09052 | Scalar-Induced Gravitational Waves from Ghost Inflation and Parity
Violation
gr-qc astro-ph.CO hep-th
We calculate the scalar-induced gravitational wave energy density in the
theory of Ghost Inflation, assuming scale invariance and taking into account
both the power spectrum- and trispectrum-induced contributions. For the latter
we consider the leading cubic and quartic couplings of the comoving curvature
perturbation in addition to two parity-violating quartic operators. In the
parity-even case, we find the relative importance of the trispectrum-induced
signal to be suppressed by the requirement of perturbativity, strengthening a
no-go theorem recently put forth. The parity-odd signal, even though also bound
to be small, is non-degenerate with the Gaussian contribution and may in
principle be comparable to the parity-even non-Gaussian part, thus potentially
serving as a probe of the Ghost Inflation scenario and of parity violating
physics during inflation.
| arxiv topic:gr-qc astro-ph.CO hep-th |
arxiv_dataset-186232306.09152 | Schematic Unification
cs.LO
We present a generalization of first-order unification to a term algebra
where variable indexing is part of the object language. We exploit variable
indexing by associating some sequences of variables ($X_0,\ X_1,\ X_2,\dots$)
with a mapping $\sigma$ whose domain is the variable sequence and whose range
consist of terms that may contain variables from the sequence. From a given
term $t$, an infinite sequence of terms may be produced by iterative
application of $\sigma$. Given a unification problem $U$ and mapping $\sigma$,
the \textit{schematic unification problem} asks whether all unification
problems $U$, $\sigma(U)$, $\sigma(\sigma(U))$, $\dots$ are unifiable. We
provide a terminating and sound algorithm. Our algorithm is \textit{complete}
if we further restrict ourselves to so-called $\infty$-stable problems. We
conjecture that this additional requirement is unnecessary for completeness.
Schematic unification is related to methods of inductive proof transformation
by resolution and inductive reasoning.
| arxiv topic:cs.LO |
arxiv_dataset-186242306.09252 | Dark ages, a window on the dark sector. Hunting for ultra-light axions
astro-ph.CO
Measurements of 21cm intensity mapping (IM) during the dark ages can
potentially provide us with an unprecedented window on high redshifts and small
scales. One of the main advantages this can bring involves the possibility to
probe the nature of dark matter. Tests of dark matter models with the
large-scale structure of the Universe are limited by non-linearities and
astrophysical effects, which are not present for IM measurements during the
dark ages. In this paper we focus on constraining the model in which dark
matter is comprised, totally or in part, by ultra-light axion-like particles
around the $10^{-18}-10^{-22}$ eV mass scale. For this model, the angular power
spectrum of 21cm brightness temperature fluctuations will exhibit a small-scale
suppression. However, this effect is intertwined with the imprint of
baryon-dark matter relative velocity at recombination, causing at the same time
an enhancement at large-scales, which is affected by the mass and abundance of
axion dark matter. In this work we forecast how future radio arrays will be
able to constrain ultra-light axion mass through both these effects on the
angular power spectrum.
| arxiv topic:astro-ph.CO |
arxiv_dataset-186252306.09352 | The effect of heavy reflector on neutronic parameters of core
physics.ins-det
The reactor baffle is an important component of a nuclear reactor that fixes
the location of the fuel assembly in the reactor core. The main types of
baffles are called light or heavy. The light baffle has mostly the form of
steel plates with outer space filled by water, and the heavy baffle is mostly a
forged steel element. Both concepts have advantages as well as disadvantages.
In the case of the light baffle, one does not need to solve the issue of void
swelling, but the neutron economy is not ideal, while the heavy baffle has a
good neutron economy, but void swelling is an issue. This paper deals with the
effect of the heavy VVER-1000 baffle on criticality. Criticality was measured
using a well-defined core composed of 6 fuel assemblies moved to a simulator of
the VVER-1000 internals, which is located at the LR-0 reactor. The experiments
confirm the fact that the water filling the cooling channels in the baffle has
a strong neutron absorbing effect. The keff calculated using the ENDF/B-VIII.0
library significantly underpredicts the experiment, whereas calculations using
a new evaluation of 56Fe by the IAEA (INDEN collaboration) give a better
agreement. Generally, the presented results are suitable for validation of iron
cross sections.
| arxiv topic:physics.ins-det |
arxiv_dataset-186262306.09452 | Distillation Strategies for Discriminative Speech Recognition Rescoring
eess.AS
Second-pass rescoring is employed in most state-of-the-art speech recognition
systems. Recently, BERT based models have gained popularity for re-ranking the
n-best hypothesis by exploiting the knowledge from masked language model
pre-training. Further, fine-tuning with discriminative loss such as minimum
word error rate (MWER) has shown to perform better than likelihood-based loss.
Streaming applications with low latency requirements impose significant
constraints on the size of the models, thereby limiting the word error rate
(WER) performance gains. In this paper, we propose effective strategies for
distilling from large models discriminatively trained with the MWER objective.
We experiment on Librispeech and production scale internal dataset for
voice-assistant. Our results demonstrate relative improvements of upto 7% WER
over student models trained with MWER. We also show that the proposed
distillation can reduce the WER gap between the student and the teacher by 62%
upto 100%.
| arxiv topic:eess.AS |
arxiv_dataset-186272306.09552 | Retrospective: EIE: Efficient Inference Engine on Sparse and Compressed
Neural Network
cs.AR
EIE proposed to accelerate pruned and compressed neural networks, exploiting
weight sparsity, activation sparsity, and 4-bit weight-sharing in neural
network accelerators. Since published in ISCA'16, it opened a new design space
to accelerate pruned and sparse neural networks and spawned many
algorithm-hardware co-designs for model compression and acceleration, both in
academia and commercial AI chips. In retrospect, we review the background of
this project, summarize the pros and cons, and discuss new opportunities where
pruning, sparsity, and low precision can accelerate emerging deep learning
workloads.
| arxiv topic:cs.AR |
arxiv_dataset-186282306.09652 | A New Low-Rank Learning Robust Quaternion Tensor Completion Method for
Color Video Inpainting Problem and Fast Algorithms
cs.CV cs.NA math.NA
The color video inpainting problem is one of the most challenging problem in
the modern imaging science. It aims to recover a color video from a small part
of pixels that may contain noise. However, there are less of robust models that
can simultaneously preserve the coupling of color channels and the evolution of
color video frames. In this paper, we present a new robust quaternion tensor
completion (RQTC) model to solve this challenging problem and derive the exact
recovery theory. The main idea is to build a quaternion tensor optimization
model to recover a low-rank quaternion tensor that represents the targeted
color video and a sparse quaternion tensor that represents noise. This new
model is very efficient to recover high dimensional data that satisfies the
prior low-rank assumption. To solve the case without low-rank property, we
introduce a new low-rank learning RQTC model, which rearranges similar patches
classified by a quaternion learning method into smaller tensors satisfying the
prior low-rank assumption. We also propose fast algorithms with global
convergence guarantees. In numerical experiments, the proposed methods
successfully recover color videos with eliminating color contamination and
keeping the continuity of video scenery, and their solutions are of higher
quality in terms of PSNR and SSIM values than the state-of-the-art algorithms.
| arxiv topic:cs.CV cs.NA math.NA |
arxiv_dataset-186292306.09752 | Politeness Stereotypes and Attack Vectors: Gender Stereotypes in
Japanese and Korean Language Models
cs.CL cs.AI cs.CY cs.LG
In efforts to keep up with the rapid progress and use of large language
models, gender bias research is becoming more prevalent in NLP. Non-English
bias research, however, is still in its infancy with most work focusing on
English. In our work, we study how grammatical gender bias relating to
politeness levels manifests in Japanese and Korean language models. Linguistic
studies in these languages have identified a connection between gender bias and
politeness levels, however it is not yet known if language models reproduce
these biases. We analyze relative prediction probabilities of the male and
female grammatical genders using templates and find that informal polite speech
is most indicative of the female grammatical gender, while rude and formal
speech is most indicative of the male grammatical gender. Further, we find
politeness levels to be an attack vector for allocational gender bias in
cyberbullying detection models. Cyberbullies can evade detection through simple
techniques abusing politeness levels. We introduce an attack dataset to (i)
identify representational gender bias across politeness levels, (ii)
demonstrate how gender biases can be abused to bypass cyberbullying detection
models and (iii) show that allocational biases can be mitigated via training on
our proposed dataset. Through our findings we highlight the importance of bias
research moving beyond its current English-centrism.
| arxiv topic:cs.CL cs.AI cs.CY cs.LG |
arxiv_dataset-186302306.09852 | Actor-Critic Model Predictive Control: Differentiable Optimization meets
Reinforcement Learning
cs.RO
An open research question in robotics is how to combine the benefits of
model-free reinforcement learning (RL) - known for its strong task performance
and flexibility in optimizing general reward formulations - with the robustness
and online replanning capabilities of model predictive control (MPC). This
paper provides an answer by introducing a new framework called Actor-Critic
Model Predictive Control. The key idea is to embed a differentiable MPC within
an actor-critic RL framework. This integration allows for short-term predictive
optimization of control actions through MPC, while leveraging RL for end-to-end
learning and exploration over longer horizons. Through various ablation
studies, we expose the benefits of the proposed approach: it achieves better
out-of-distribution behaviour, better robustness to changes in the dynamics and
improved sample efficiency. Additionally, we conduct an empirical analysis that
reveals a relationship between the critic's learned value function and the cost
function of the differentiable MPC, providing a deeper understanding of the
interplay between the critic's value and the MPC cost functions. Finally, we
validate our method in a drone racing task on different tracks, in both
simulation and the real world. Our method achieves the same superhuman
performance as state-of-the-art model-free RL, showcasing speeds of up to 21
m/s. We show that the proposed architecture can achieve real-time control
performance, learn complex behaviors via trial and error, and retain the
predictive properties of the MPC to better handle out-of-distribution behavior.
| arxiv topic:cs.RO |
arxiv_dataset-186312306.09952 | Convective heat transfer in the Burgers-Rayleigh-B\'enard system
physics.flu-dyn nlin.CD
The dynamics of heat transfer in a model system of Rayleigh-B\'enard (RB)
convection reduced to its essential, here dubbed Burgers-Rayleigh-B\'enard
(BRB), is studied. The system is spatially one-dimensional, the flow field is
compressible and its evolution is described by the Burgers equation forced by
an active temperature field. The BRB dynamics shares some remarkable
similarities with realistic RB thermal convection in higher spatial dimensions:
i) it has a supercritical pitchfork instability for the onset of convection
which solely depends on the Rayleigh number $(Ra)$ and not on Prandlt $(Pr)$,
occurring at the critical value $Ra_c = (2\pi)^4$ ii) the convective regime is
spatially organized in distinct boundary-layers and bulk regions, iii) the
asymptotic high $Ra$ limit displays the Nusselt and Reynolds numbers scaling
regime $Nu = \sqrt{RaPr}/4$ for $Pr\ll 1$, $Nu=\sqrt{Ra}/(4\sqrt{\pi})$ for
$Pr\gg1$ and $Re = \sqrt{Ra/Pr}/\sqrt{12}$, thus making BRB the simplest
wall-bounded convective system exhibiting the so called ultimate regime of
convection. These scaling laws, derived analytically through a matched
asymptotic analysis are fully supported by the results of the accompanying
numerical simulations. A major difference with realistic natural convection is
the absence of turbulence. The BRB dynamics is stationary at any $Ra$ number
above the onset of convection. This feature results from a nonlinear saturation
mechanism whose existence is grasped by means of a two-mode truncated equation
system and via a stability analysis of the convective regime.
| arxiv topic:physics.flu-dyn nlin.CD |
arxiv_dataset-186322306.10052 | Assigning AI: Seven Approaches for Students, with Prompts
cs.CY cs.AI
This paper examines the transformative role of Large Language Models (LLMs)
in education and their potential as learning tools, despite their inherent
risks and limitations. The authors propose seven approaches for utilizing AI in
classrooms: AI-tutor, AI-coach, AI-mentor, AI-teammate, AI-tool, AI-simulator,
and AI-student, each with distinct pedagogical benefits and risks. The aim is
to help students learn with and about AI, with practical strategies designed to
mitigate risks such as complacency about the AI's output, errors, and biases.
These strategies promote active oversight, critical assessment of AI outputs,
and complementarity of AI's capabilities with the students' unique insights. By
challenging students to remain the "human in the loop," the authors aim to
enhance learning outcomes while ensuring that AI serves as a supportive tool
rather than a replacement. The proposed framework offers a guide for educators
navigating the integration of AI-assisted learning in classrooms
| arxiv topic:cs.CY cs.AI |
arxiv_dataset-186332306.10152 | Low-Resource Text-to-Speech Using Specific Data and Noise Augmentation
eess.AS cs.SD
Many neural text-to-speech architectures can synthesize nearly natural speech
from text inputs. These architectures must be trained with tens of hours of
annotated and high-quality speech data. Compiling such large databases for
every new voice requires a lot of time and effort. In this paper, we describe a
method to extend the popular Tacotron-2 architecture and its training with data
augmentation to enable single-speaker synthesis using a limited amount of
specific training data. In contrast to elaborate augmentation methods proposed
in the literature, we use simple stationary noises for data augmentation. Our
extension is easy to implement and adds almost no computational overhead during
training and inference. Using only two hours of training data, our approach was
rated by human listeners to be on par with the baseline Tacotron-2 trained with
23.5 hours of LJSpeech data. In addition, we tested our model with a
semantically unpredictable sentences test, which showed that both models
exhibit similar intelligibility levels.
| arxiv topic:eess.AS cs.SD |
arxiv_dataset-186342306.10252 | Thermodynamics of $f(R)$ Gravity: The Double Well Potential Case
gr-qc hep-th
In this work we further extend the analysis of $f(R)$ theories of gravity in
the metric formalism under the approach of a Thermodynamics analogy, proposed
in arXiv:1911.04830v3. Here we assume a double-well inflationary potential in
the Einstein frame and obtain a parametric form of $f(R)$ in the corresponding
Jordan frame.
The whole Thermodynamics picture then follows: an equation of state, binodal
and spinodal curves, phase transition, critical quantities (pressure, volume
and temperature), entropy jumps, specific-heat divergence (and the
corresponding critical exponent) and a butterfly catastrophe.
| arxiv topic:gr-qc hep-th |
arxiv_dataset-186352306.10352 | Deep Reinforcement Learning for Flipper Control of Tracked Robots
cs.RO
The autonomous control of flippers plays an important role in enhancing the
intelligent operation of tracked robots within complex environments. While
existing methods mainly rely on hand-crafted control models, in this paper, we
introduce a novel approach that leverages deep reinforcement learning (DRL)
techniques for autonomous flipper control in complex terrains. Specifically, we
propose a new DRL network named AT-D3QN, which ensures safe and smooth flipper
control for tracked robots. It comprises two modules, a feature extraction and
fusion module for extracting and integrating robot and environment state
features, and a deep Q-Learning control generation module for incorporating
expert knowledge to obtain a smooth and efficient control strategy. To train
the network, a novel reward function is proposed, considering both learning
efficiency and passing smoothness. A simulation environment is constructed
using the Pymunk physics engine for training. We then directly apply the
trained model to a more realistic Gazebo simulation for quantitative analysis.
The consistently high performance of the proposed approach validates its
superiority over manual teleoperation.
| arxiv topic:cs.RO |
arxiv_dataset-186362306.10452 | MISMATCH: Fine-grained Evaluation of Machine-generated Text with
Mismatch Error Types
cs.CL
With the growing interest in large language models, the need for evaluating
the quality of machine text compared to reference (typically human-generated)
text has become focal attention. Most recent works focus either on
task-specific evaluation metrics or study the properties of machine-generated
text captured by the existing metrics. In this work, we propose a new
evaluation scheme to model human judgments in 7 NLP tasks, based on the
fine-grained mismatches between a pair of texts. Inspired by the recent efforts
in several NLP tasks for fine-grained evaluation, we introduce a set of 13
mismatch error types such as spatial/geographic errors, entity errors, etc, to
guide the model for better prediction of human judgments. We propose a neural
framework for evaluating machine texts that uses these mismatch error types as
auxiliary tasks and re-purposes the existing single-number evaluation metrics
as additional scalar features, in addition to textual features extracted from
the machine and reference texts. Our experiments reveal key insights about the
existing metrics via the mismatch errors. We show that the mismatch errors
between the sentence pairs on the held-out datasets from 7 NLP tasks align well
with the human evaluation.
| arxiv topic:cs.CL |
arxiv_dataset-186372306.10552 | Weighted Subsequential ergodic theorems on Orlicz spaces
math.OA
For a semifinite von Neumann algebra M, individual convergence of
subsequential, \mathcal{Z}(M) (center of M) valued weighted ergodic averages
are studied in noncommutative Orlicz spaces. In the process, we also derive a
maximal ergodic inequality corresponding to such averages in noncommutative
L^p~ (1 \leq p < \infty) spaces using the weak (1,1) inequality obtained by
Yeadon.
| arxiv topic:math.OA |
arxiv_dataset-186382306.10652 | Unrestricted component count in multiphase lattice Boltzmann: a
fugacity-based approach
physics.flu-dyn
Studies of multiphase fluids utilizing the lattice Boltzmann method (LBM) are
typically severely restricted by the number of components or chemical species
being modeled. This restriction is particularly pronounced for multiphase
systems exhibiting partial miscibility and significant interfacial mass
exchange, which is a common occurrence in realistic multiphase systems.
Modeling such systems becomes increasingly complex as the number of chemical
species increases due to the increased role of molecular interactions and the
types of thermodynamic behavior that become possible. The recently introduced
fugacity-based LBM [M. Soomro, L. F. Ayala, C. Peng, and O. M. Ayala, Phys.
Rev. E 107, 015304 (2023)] has provided a thermodynamically-consistent modeling
platform for multicomponent, partially-miscible LBM simulations. However, until
now, this fugacity-based LB model had lacked a comprehensive demonstration of
its ability to accurately reproduce thermodynamic behavior beyond binary
mixtures and to remove any restrictions in a number of components for
multiphase LBM. In this paper, we closely explore these fugacity-based LBM
capabilities by showcasing comprehensive, thermodynamically-consistent
simulations of multiphase mixtures of up to ten chemical components. The paper
begins by validating the model against the Young-Laplace equation for a droplet
composed of three components. The model is then applied to study mixtures with
a range of component numbers from one to six, showing agreement with rigorous
thermodynamic predictions and demonstrating linear scaling of computational
time with the number of components. We further investigate--which has been
previously absent in LB literature--ternary systems in detail, by exploring a
wide range of temperature, pressure, and overall composition conditions to
produce various characteristic ternary diagrams. [cont. in pdf]
| arxiv topic:physics.flu-dyn |
arxiv_dataset-186392306.10752 | Are Shortfall Systemic Risk Measures One Dimensional?
q-fin.MF
Shortfall systemic (multivariate) risk measures $\rho$ defined through an
$N$-dimensional multivariate utility function $U$ and random allocations can be
represented as classical (one dimensional) shortfall risk measures associated
to an explicitly determined $1$-dimensional function constructed from $U$. This
finding allows for simplifying the study of several properties of $\rho$, such
as dual representations, law invariance and stability.
| arxiv topic:q-fin.MF |
arxiv_dataset-186402306.10852 | Formation of liquid shells in active droplet systems
cond-mat.soft physics.bio-ph
We study a chemically active binary mixture undergoing phase separation and
show that under non-equilibrium conditions, stable liquid spherical shells can
form via a spinodal instability in the droplet center. A single liquid shell
tends to grow until it undergoes a shape instability beyond a critical size. In
an active emulsion, many stable and stationary liquid shells can coexist. We
discuss conditions under which liquid shells are stable and dominant as
compared to regimes where droplets undergo shape instabilities and divide.
| arxiv topic:cond-mat.soft physics.bio-ph |
arxiv_dataset-186412306.10952 | Evaluation of an automated choroid segmentation algorithm in a
longitudinal kidney donor and recipient cohort
q-bio.QM eess.IV physics.med-ph
Purpose: To evaluate the performance of an automated choroid segmentation
algorithm in optical coherence tomography (OCT) data using a longitudinal
kidney donor and recipient cohort. Methods: We assessed 22 donors and 23
patients requiring renal transplantation over up to 1 year post-transplant. We
measured choroidal thickness (CT) and area and compared our automated CT
measurements to manual ones at the same locations. We estimated associations
between choroidal measurements and markers of renal function (estimated
glomerular filtration rate (eGFR), serum creatinine and urea) using correlation
and linear mixed-effects (LME) modelling. Results: There was good agreement
between manual and automated CT. Automated measures were more precise because
of smaller measurement error over time. External adjudication of major
discrepancies were in favour of automated measures. Significant differences
were observed in the choroid pre- and post-transplant in both cohorts, and LME
modelling revealed significant linear associations observed between choroidal
measures and renal function in recipients. Significant associations were mostly
stronger with automated CT (eGFR P<0.001, creatinine P=0.004, urea P=0.04)
compared to manual CT (eGFR P=0.002, creatinine P=0.01, urea P=0.03).
Conclusions: Our automated approach has greater precision than human-performed
manual measurements, which may explain stronger associations with renal
function compared to manual measurements. To improve detection of meaningful
associations with clinical endpoints in longitudinal studies of OCT, reducing
measurement error should be a priority, and automated measurements help achieve
this. Translational relevance: We introduce a novel choroid segmentation
algorithm which can replace manual grading for studying the choroid in renal
disease, and other clinical conditions.
| arxiv topic:q-bio.QM eess.IV physics.med-ph |
arxiv_dataset-186422306.11052 | A spatio-temporal network for video semantic segmentation in surgical
videos
cs.CV
Semantic segmentation in surgical videos has applications in intra-operative
guidance, post-operative analytics and surgical education. Segmentation models
need to provide accurate and consistent predictions since temporally
inconsistent identification of anatomical structures can impair usability and
hinder patient safety. Video information can alleviate these challenges leading
to reliable models suitable for clinical use. We propose a novel architecture
for modelling temporal relationships in videos. The proposed model includes a
spatio-temporal decoder to enable video semantic segmentation by improving
temporal consistency across frames. The encoder processes individual frames
whilst the decoder processes a temporal batch of adjacent frames. The proposed
decoder can be used on top of any segmentation encoder to improve temporal
consistency. Model performance was evaluated on the CholecSeg8k dataset and a
private dataset of robotic Partial Nephrectomy procedures. Segmentation
performance was improved when the temporal decoder was applied across both
datasets. The proposed model also displayed improvements in temporal
consistency.
| arxiv topic:cs.CV |
arxiv_dataset-186432306.11152 | Few-shot Learning for Inference in Medical Imaging with Subspace Feature
Representations
math.NA cs.NA
Unlike the field of visual scene recognition where tremendous advances have
taken place due to the availability of very large datasets to train deep neural
networks, inference from medical images is often hampered by the fact that only
small amounts of data may be available. When working with very small dataset
problems, of the order of a few hundred items of data, the power of deep
learning may still be exploited by using a model pre-trained on natural images
as a feature extractor and carrying out classic pattern recognition techniques
in this feature space, the so-called few-shot learning problem. In regimes
where the dimension of this feature space is comparable to or even larger than
the number of items of data, dimensionality reduction is a necessity and is
often achieved by principal component analysis, i.e., singular value
decomposition (SVD). In this paper, noting the inappropriateness of using SVD
for this setting, we usher in and explore two alternatives based on
discriminant analysis and non-negative matrix factorization (NMF). Using 14
different datasets spanning $11$ distinct disease types, we demonstrate that
discriminant subspaces at low dimensions achieve significant improvements over
SVD-based subspaces and the original feature space. We also show that NMF at
modest dimensions is a competitive alternative to SVD in this setting.
| arxiv topic:math.NA cs.NA |
arxiv_dataset-186442306.11252 | HK-LegiCoST: Leveraging Non-Verbatim Transcripts for Speech Translation
cs.CL cs.LG
We introduce HK-LegiCoST, a new three-way parallel corpus of
Cantonese-English translations, containing 600+ hours of Cantonese audio, its
standard traditional Chinese transcript, and English translation, segmented and
aligned at the sentence level. We describe the notable challenges in corpus
preparation: segmentation, alignment of long audio recordings, and
sentence-level alignment with non-verbatim transcripts. Such transcripts make
the corpus suitable for speech translation research when there are significant
differences between the spoken and written forms of the source language. Due to
its large size, we are able to demonstrate competitive speech translation
baselines on HK-LegiCoST and extend them to promising cross-corpus results on
the FLEURS Cantonese subset. These results deliver insights into speech
recognition and translation research in languages for which non-verbatim or
``noisy'' transcription is common due to various factors, including vernacular
and dialectal speech.
| arxiv topic:cs.CL cs.LG |
arxiv_dataset-186452306.11352 | Chemical Mapping of Excitons in Halide Double Perovskites
cond-mat.mtrl-sci cond-mat.mes-hall physics.chem-ph
Halide double perovskites are an emerging class of semiconductors with
tremendous chemical and electronic diversity. While their bandstructure
features can be understood from frontier-orbital models, chemical intuition for
optical excitations remains incomplete. Here, we use \textit{ab initio}
many-body perturbation theory within the $GW$ and the Bethe-Salpeter Equation
approach to calculate excited-state properties of a representative range of
Cs$_2$BB$'$Cl$_6$ double perovskites. Our calculations reveal that double
perovskites with different combinations of B and B$'$ cations display a broad
variety of electronic bandstructures and dielectric properties, and form
excitons with binding energies ranging over several orders of magnitude. We
correlate these properties with the orbital-induced anisotropy of
charge-carrier effective masses and the long-range behavior of the dielectric
function, by comparing with the canonical conditions of the Wannier-Mott model.
Furthermore, we derive chemically intuitive rules for predicting the nature of
excitons in halide double perovskites using electronic structure information
obtained from computationally inexpensive DFT calculations.
| arxiv topic:cond-mat.mtrl-sci cond-mat.mes-hall physics.chem-ph |
arxiv_dataset-186462306.11452 | Stripes polymorphism and water-like anomaly in hard core-soft corona
dumbbells
cond-mat.soft
In this paper we investigate the phase diagram of a dumbbell model composed
of two hard-core soft-corona beads through $NpT$ simulations. This particular
system was chosen due to its ability to exhibit a diverse range of stripe
patterns. Analyzing the thermodynamic and structural changes along compression
isotherms, we explore the transition between these distinct patterns. In
addition to the stripe and Low-Density-Triangular solid phases obtained, we
observed a Nematic Anisotropic phase characterized by a polymer-like pattern at
high temperatures and intermediate pressures. Furthermore, we demonstrate the
significant role played by the new characteristic length scale, which arises
from the anisotropic geometry of the dumbbell structure, in the transition
between the stripes patterns. Notably, not only do the structural properties
exhibit intriguing behavior, but the diffusion and density in the nematic fluid
phase also displays a water-like anomalous increase under compression. Those
findings can be valuable in guiding the design of materials based on
nanoparticles, with the aim of achieving specific mesopatterns.
| arxiv topic:cond-mat.soft |
arxiv_dataset-186472306.11552 | Inter-Cell Network Slicing With Transfer Learning Empowered Multi-Agent
Deep Reinforcement Learning
cs.NI cs.LG cs.MA
Network slicing enables operators to efficiently support diverse applications
on a common physical infrastructure. The ever-increasing densification of
network deployment leads to complex and non-trivial inter-cell interference,
which requires more than inaccurate analytic models to dynamically optimize
resource management for network slices. In this paper, we develop a DIRP
algorithm with multiple deep reinforcement learning (DRL) agents to
cooperatively optimize resource partition in individual cells to fulfill the
requirements of each slice, based on two alternative reward functions.
Nevertheless, existing DRL approaches usually tie the pretrained model
parameters to specific network environments with poor transferability, which
raises practical deployment concerns in large-scale mobile networks. Hence, we
design a novel transfer learning-aided DIRP (TL-DIRP) algorithm to ease the
transfer of DIRP agents across different network environments in terms of
sample efficiency, model reproducibility, and algorithm scalability. The
TL-DIRP algorithm first centrally trains a generalized model and then transfers
the "generalist" to each local agent as "specialist" with distributed
finetuning and execution. TL-DIRP consists of two steps: 1) centralized
training of a generalized distributed model, 2) transferring the "generalist"
to each "specialist" with distributed finetuning and execution. The numerical
results show that not only DIRP outperforms existing baseline approaches in
terms of faster convergence and higher reward, but more importantly, TL-DIRP
significantly improves the service performance, with reduced exploration cost,
accelerated convergence rate, and enhanced model reproducibility. As compared
to a traffic-aware baseline, TL-DIRP provides about 15% less violation ratio of
the quality of service (QoS) for the worst slice service and 8.8% less
violation on the average service QoS.
| arxiv topic:cs.NI cs.LG cs.MA |
arxiv_dataset-186482306.11652 | Sparse Bayesian Estimation of Parameters in Linear-Gaussian State-Space
Models
stat.CO stat.ME
State-space models (SSMs) are a powerful statistical tool for modelling
time-varying systems via a latent state. In these models, the latent state is
never directly observed. Instead, a sequence of data points related to the
state are obtained. The linear-Gaussian state-space model is widely used, since
it allows for exact inference when all model parameters are known, however this
is rarely the case. The estimation of these parameters is a very challenging
but essential task to perform inference and prediction. In the linear-Gaussian
model, the state dynamics are described via a state transition matrix. This
model parameter is known to behard to estimate, since it encodes the
relationships between the state elements, which are never observed. In many
applications, this transition matrix is sparse since not all state components
directly affect all other state components. However, most parameter estimation
methods do not exploit this feature. In this work we propose SpaRJ, a fully
probabilistic Bayesian approach that obtains sparse samples from the posterior
distribution of the transition matrix. Our method explores sparsity by
traversing a set of models that exhibit differing sparsity patterns in the
transition matrix. Moreover, we also design new effective rules to explore
transition matrices within the same level of sparsity. This novel methodology
has strong theoretical guarantees, and unveils the latent structure of the data
generating process, thereby enhancing interpretability. The performance of
SpaRJ is showcased in example with dimension 144 in the parameter space, and in
a numerical example with real data.
| arxiv topic:stat.CO stat.ME |
arxiv_dataset-186492306.11752 | Geometric phases for a thermal two-dimensional mixed spin 1/2 system
quant-ph
Quantum mechanical methods for getting geometric phases for mixed states are
analyzed. Parallel transport equations for pure states are generalized to mixed
states by which dynamical phases are eliminated. The geometric phases of mixed
states are obtained as Pancharatnam phases which are valid also for open
cycles. The geometric phases are derived here by SU(2) transformations of mixed
thermal states which are different from those used in NMR and neutron
interference experiments. While the zeroth order Hamiltonian is given by the
interaction of a magnetic moment and constant magnetic field in the z
direction, the high order perturbations assumed in the present article are
composed of two oscillating magnetic fields in the same z direction. These
assumptions lead to a special form of the SU(2) unitary transformation of the
mixed thermal states by which results for geometric phase and for interference
intensity are derived.
| arxiv topic:quant-ph |
arxiv_dataset-186502306.11852 | Einstein beams and the diffractive aspect of gravitationally-lensed
light
astro-ph.IM physics.optics
The study of light lensed by cosmic matter has yielded much information about
astrophysical questions. Observations are explained using geometrical optics
following a ray-based description of light. After deflection the lensed light
interferes, but observing this diffractive aspect of gravitational lensing has
not been possible due to coherency challenges caused by the finite size of the
sources or lack of near-perfect alignment. In this article, we report on the
observation of these wave effects of gravitational lensing by recreating the
lensing conditions in the laboratory via electro-optic deflection of coherent
laser light. The lensed light produces a beam containing regularities,
caustics, and chromatic modulations of intensity that depend on the symmetry
and structure of the lensing object. We were also able to observe previous and
new geometric-optical lensing situations that can be compared to astrophysical
observations. This platform could be a useful tool for testing
numerical/analytical simulations, and for performing analog simulations of
lensing situations when they are difficult to obtain otherwise. We found that
laboratory lensed beams constitute a new class of beams, with long-range, low
expansion, and self-healing properties, opening new possibilities for
non-astrophysical applications.
| arxiv topic:astro-ph.IM physics.optics |
arxiv_dataset-186512306.11952 | First-principles prediction of structural, magnetic properties of
Cr-substituted strontium hexaferrite, and its site preference
cond-mat.mtrl-sci
To investigate the structural and magnetic properties of Cr-doped M-type
strontium hexaferrite (SrFe$_{12}$O$_{19}$) with x = (0.0, 0.5, 1.0), we
perform first-principles total-energy calculations relied on density functional
theory. Based on the calculation of the substitution energy of Cr in strontium
hexaferrite and formation probability analysis, we conclude that the doped Cr
atoms prefer to occupy the 2a, 12k, and 4f$_{2}$ sites which is in good
agreement with the experimental findings. Due to Cr$^{3+}$ ion moment, 3
{$\mu_B$}, smaller than that of Fe$^{3+}$ ion, 5 {$\mu_B$}, saturation
magnetization (M$_{s}$) reduce rapidly as the concentration of Cr increases in
strontium hexaferrite. The magnetic anisotropic field $\left(H_{a}\right)$
rises with an increasing fraction of Cr despite a significant reduction of
magnetization and a slight increase of magnetocrystalline anisotropy
$\left(K_{1}\right)$.The cause for the rise in magnetic anisotropy field
$\left(H_{a}\right)$ with an increasing fraction of Cr is further emphasized by
our formation probability study. Cr$^{3+}$ ions prefer to occupy the 2a sites
at lower temperatures, but as the temperature rises, it is more likely that
they will occupy the 12k site. Cr$^{3+}$ ions are more likely to occupy the 12k
site than the 2a site at a specific annealing temperature (>700{\deg}C).
| arxiv topic:cond-mat.mtrl-sci |
arxiv_dataset-186522306.12052 | Some invariants of $U(1,1;\mathbb{H})$ and diagonalization
math.RA math.CV math.GR
Denote by $\mathbb{H}$ the set of all quaternions. We are interested in the
group $U(1,1;\mathbb{H})$, which is a subgroup of $2\times 2$ quaternionic
matrix group and is sometimes called $Sp(1,1)$. As well known,
$U(1,1;\mathbb{H})$ corresponds to the quaternionic M\"{o}bius transformations
on the unit ball in $\mathbb{H}$. In this article, some similar invariants on
$U(1,1;\mathbb{H})$ are discussed. Our main result shows that each matrix $T\in
U(1,1;\mathbb{H})$, which corresponds to an elliptic quaternionic M\"{o}bius
transformation $g_T(z)$, could be $U(1,1;\mathbb{H})$-similar to a diagonal
matrix.
| arxiv topic:math.RA math.CV math.GR |
arxiv_dataset-186532306.12152 | Exploiting Multimodal Synthetic Data for Egocentric Human-Object
Interaction Detection in an Industrial Scenario
cs.CV
In this paper, we tackle the problem of Egocentric Human-Object Interaction
(EHOI) detection in an industrial setting. To overcome the lack of public
datasets in this context, we propose a pipeline and a tool for generating
synthetic images of EHOIs paired with several annotations and data signals
(e.g., depth maps or segmentation masks). Using the proposed pipeline, we
present EgoISM-HOI a new multimodal dataset composed of synthetic EHOI images
in an industrial environment with rich annotations of hands and objects. To
demonstrate the utility and effectiveness of synthetic EHOI data produced by
the proposed tool, we designed a new method that predicts and combines
different multimodal signals to detect EHOIs in RGB images. Our study shows
that exploiting synthetic data to pre-train the proposed method significantly
improves performance when tested on real-world data. Moreover, to fully
understand the usefulness of our method, we conducted an in-depth analysis in
which we compared and highlighted the superiority of the proposed approach over
different state-of-the-art class-agnostic methods. To support research in this
field, we publicly release the datasets, source code, and pre-trained models at
https://iplab.dmi.unict.it/egoism-hoi.
| arxiv topic:cs.CV |
arxiv_dataset-186542306.12252 | Noble gas functional defect with unusual relaxation pattern in solids
cond-mat.mtrl-sci
The conventional understanding has always been that noble gases are
chemically inert and do not affect materials properties. This belief has led to
their use as a standard reference in various experimental applications through
noble gas implantation. However, in our research, using first-principles
calculations, we delve into the effects of noble gas defects on the properties
of several functional oxides, thereby questioning this long-held assumption. We
provide evidence that noble gases can indeed serve as functional defects. They
have the potential to decentralize the localized defect states and prompt a
shift of electrons from a localized state to the main conduction band. Our
investigation unveils that noble gas defects can indeed significantly alter
material properties. Thus, we underscore the importance of factoring in such
defects when assessing material properties.
| arxiv topic:cond-mat.mtrl-sci |
arxiv_dataset-186552306.12352 | Evidence of Radius Inflation in Radiative GCM Models of WASP-76b due to
the Advection of Potential Temperature
astro-ph.EP
Understanding the discrepancy between the radii of observed hot Jupiters and
standard 'radiative-convective' models remains a hotly debated topic in the
exoplanet community. One mechanism which has been proposed to bridge this gap,
and which has recently come under scrutiny, is the vertical advection of
potential temperature from the irradiated outer atmosphere deep into the
interior, heating the deep, unirradiated, atmosphere, warming the internal
adiabat, and resulting in radius inflation. Specifically, a recent study which
explored the atmosphere of WASP-76b using a 3D, non-grey, GCM suggested that
their models lacked radius inflation, and hence any vertical enthalpy
advection. Here we perform additional analysis of these, and related, models,
focusing on an explicit analysis of vertical enthalpy transport and the
resulting heating of the deep atmosphere compared with 1D models. Our results
indicate that, after any evolution linked with initialisation, all the WASP-76b
models considered here exhibit significant vertical enthalpy transport, heating
the deep atmosphere significantly when compared with standard 1D models.
Furthermore, comparison of a long time-scale (and hence near steady-state)
model with a Jupiter-like internal-structure model suggests not only strong
radius-inflation, but also that the model radius, $1.98 \mathrm{R_{J}}$, may be
comparable with observations ($1.83\pm0.06 \mathrm{R_{J}}$). We thus conclude
that the vertical advection of potential temperature alone is enough to explain
the radius inflation of WASP-76b, and potentially other irradiated gas giants,
albeit with the proviso that the exact strength of the vertical advection
remains sensitive to model parameters, such as the inclusion of deep
atmospheric drag.
| arxiv topic:astro-ph.EP |
arxiv_dataset-186562306.12452 | Impacts of nature deprivations during the COVID-19 pandemic: A pre-post
comparison
physics.soc-ph
Nature provides a myriad of intangible and non-material services to people.
However, urbanites are increasingly disconnected from the natural world. The
consequences of this progressive disconnection from nature remain difficult to
measure as this process is slow and long-term monitoring or large-scale
manipulation on nature experiences are scarce. Measures to contain the spread
of the recent covid-19 pandemic (i.e., lockdowns) have potentially reduced or
even suppressed nature experiences in cities. This situation provided an
opportunity for conducting a longitudinal study that can serve as a sort of
natural experiment to quantify the effects of nature deprivation on
individuals' health, well-being and relationship to nature. We collected data
on these variables from the same individuals inhabiting a large metropolis (Tel
Aviv, Israel) twice, in 2018 (before) and during the lockdown in 2020. Our
results confirmed that frequency, duration and quality of nature interactions
dropped during the lockdown, while environmental attitudes and affinity towards
nature remained similar. This was particularly true for people living in the
least green neighborhoods, where a significant decrease in personal and social
well-being was also found. Finally, affinity towards nature influenced
well-being through nature experiences in 2018. The mediation effect was not
significant in 2020, probably due to the decrease in nature experiences during
the lockdown, but the direct relationship between affinity towards nature and
well-being remained strong. These results provide insights into the means
required to align the public health and conservation agendas to safeguard
urbanites' health and well-being during a pandemic and mitigate the
biodiversity crisis.
| arxiv topic:physics.soc-ph |
arxiv_dataset-186572306.12552 | SituatedGen: Incorporating Geographical and Temporal Contexts into
Generative Commonsense Reasoning
cs.CL
Recently, commonsense reasoning in text generation has attracted much
attention. Generative commonsense reasoning is the task that requires machines,
given a group of keywords, to compose a single coherent sentence with
commonsense plausibility. While existing datasets targeting generative
commonsense reasoning focus on everyday scenarios, it is unclear how well
machines reason under specific geographical and temporal contexts. We formalize
this challenging task as SituatedGen, where machines with commonsense should
generate a pair of contrastive sentences given a group of keywords including
geographical or temporal entities. We introduce a corresponding English dataset
consisting of 8,268 contrastive sentence pairs, which are built upon several
existing commonsense reasoning benchmarks with minimal manual labor.
Experiments show that state-of-the-art generative language models struggle to
generate sentences with commonsense plausibility and still lag far behind human
performance. Our dataset is publicly available at
https://github.com/yunx-z/situated_gen.
| arxiv topic:cs.CL |
arxiv_dataset-186582306.12652 | UltraGlove: Hand Pose Estimation with Mems-Ultrasonic Sensors
cs.CV cs.GR cs.HC cs.RO
Hand tracking is an important aspect of human-computer interaction and has a
wide range of applications in extended reality devices. However, current hand
motion capture methods suffer from various limitations. For instance,
visual-based hand pose estimation is susceptible to self-occlusion and changes
in lighting conditions, while IMU-based tracking gloves experience significant
drift and are not resistant to external magnetic field interference. To address
these issues, we propose a novel and low-cost hand-tracking glove that utilizes
several MEMS-ultrasonic sensors attached to the fingers, to measure the
distance matrix among the sensors. Our lightweight deep network then
reconstructs the hand pose from the distance matrix. Our experimental results
demonstrate that this approach is both accurate, size-agnostic, and robust to
external interference. We also show the design logic for the sensor selection,
sensor configurations, circuit diagram, as well as model architecture.
| arxiv topic:cs.CV cs.GR cs.HC cs.RO |
arxiv_dataset-186592306.12752 | When is the average number of saddle points typical?
cond-mat.stat-mech cond-mat.dis-nn
A common measure of a function's complexity is the count of its stationary
points. For complicated functions, this count grows exponentially with the
volume and dimension of their domain. In practice, the count is averaged over a
class of functions (the annealed average), but the large numbers involved can
produce averages biased by extremely rare samples. Typical counts are reliably
found by taking the average of the logarithm (the quenched average), which is
more difficult and not often done in practice. When most stationary points are
uncorrelated with each other, quenched and anneals averages are equal.
Equilibrium heuristics can guarantee when most of the lowest minima will be
uncorrelated. We show that these equilibrium heuristics cannot be used to draw
conclusions about other minima and saddles by producing examples among
Gaussian-correlated functions on the hypersphere where the count of certain
saddles and minima has different quenched and annealed averages, despite being
guaranteed `safe' in the equilibrium setting. We determine conditions for the
emergence of nontrivial correlations between saddles, and discuss the
implications for the geometry of those functions and what out-of-equilibrium
settings might be affected.
| arxiv topic:cond-mat.stat-mech cond-mat.dis-nn |
arxiv_dataset-186602306.12852 | Zygmund graphs are thin for doubling measures
math.CA math.FA
The Zygmund functions form an intermediate class between Lipschitz and
H\"older functions; their second order divided differences are uniformly
bounded. It is well known that for $d \geq 1$ the graph of any Lipschitz
function $f:\R^d \rightarrow \R$ is thin for doubling measures, and we extend
this result to the Zygmund class.
| arxiv topic:math.CA math.FA |
arxiv_dataset-186612306.12952 | A balanced finite-element method for an axisymmetrically loaded thin
shell
math.NA cs.NA
We analyse a finite-element discretisation of a differential equation
describing an axisymmetrically loaded thin shell. The problem is singularly
perturbed when the thickness of the shell becomes small. We prove robust
convergence of the method in a balanced norm that captures the layers present
in the solution. Numerical results confirm our findings.
| arxiv topic:math.NA cs.NA |
arxiv_dataset-186622306.13052 | Nonexistence of isoperimetric sets in spaces of positive curvature
math.DG math.MG
For every $d\ge 3$, we construct a noncompact smooth $d$-dimensional
Riemannian manifold with strictly positive sectional curvature without
isoperimetric sets for any volume below $1$. We construct a similar example
also for the relative isoperimetric problem in (unbounded) convex sets in
$\mathbb R^d$. The examples we construct have nondegenerate asymptotic cone.
The dimensional constraint $d\ge 3$ is sharp. Our examples exhibit
nonexistence of isoperimetric sets only for small volumes; indeed in
nonnegatively curved spaces with nondegenerate asymptotic cones isoperimetric
sets with large volumes always exist.
This is the first instance of noncollapsed nonnegatively curved space without
isoperimetric sets.
| arxiv topic:math.DG math.MG |
arxiv_dataset-186632306.13152 | t-design curves and mobile sampling on the sphere
math.MG cs.NA math.NA
In analogy to classical spherical t-design points, we introduce the concept
of t-design curves on the sphere. This means that the line integral along a
t-design curve integrates polynomials of degree t exactly. For low degrees we
construct explicit examples. We also derive lower asymptotic bounds on the
lengths of t-design curves. Our main results prove the existence of
asymptotically optimal t-design curves in the Euclidean 2-sphere and the
existence of t-design curves in the d-sphere.
| arxiv topic:math.MG cs.NA math.NA |
arxiv_dataset-186642306.13252 | Self-similar solutions to the hypoviscous Burgers and SQG equations at
criticality
physics.flu-dyn math.AP
After reviewing the source-type solution of the Burgers equation with
standard dissipativity, we study the hypoviscous counterpart of the Burgers
equation. 1) We determine an equation that governs the near-identity
transformation underlying its self-similar solution. 2) We develop its
approximation scheme and construct the first-order approximation. 3) We obtain
the source-type solution numerically by the Newton-Raphson iteration scheme and
find it to agree well with the first-order approximation. Implications of the
source-type solution are given, regarding the possibility of linearisation of
the hypoviscous Burgers equation. Finally we address the problems of the
incompressible fluid equations in two dimensions, centering on the surface
quasi-geostrophic equation with standard and hypoviscous dissipativity.
| arxiv topic:physics.flu-dyn math.AP |
arxiv_dataset-186652306.13352 | Kapitza-resistance-like exciton dynamics in atomically flat
MoSe$_{2}$-WSe$_{2}$ lateral heterojunction
cond-mat.mes-hall physics.optics
Being able to control the neutral excitonic flux is a mandatory step for the
development of future room-temperature two-dimensional excitonic devices.
Semiconducting Monolayer Transition Metal Dichalcogenides (TMD-ML) with
extremely robust and mobile excitons are highly attractive in this regard.
However, generating an efficient and controlled exciton transport over long
distances is a very challenging task. Here we demonstrate that an atomically
sharp TMD-ML lateral heterostructure (MoSe$_{2}$-WSe$_{2}$) transforms the
isotropic exciton diffusion into a unidirectional excitonic flow through the
junction. Using tip-enhanced photoluminescence spectroscopy (TEPL) and a
modified exciton transfer model, we show a discontinuity of the exciton density
distribution on each side of the interface. We introduce the concept of exciton
Kapitza resistance, by analogy with the interfacial thermal resistance referred
to as Kapitza resistance. By comparing different heterostructures with or
without top hexagonal boron nitride (hBN) layer, we deduce that the transport
properties can be controlled, over distances far greater than the junction
width, by the exciton density through near-field engineering and/or laser power
density. This work provides a new approach for controlling the neutral exciton
flow, which is key toward the conception of excitonic devices.
| arxiv topic:cond-mat.mes-hall physics.optics |
arxiv_dataset-186662306.13452 | A Graph Neural Network Approach for Temporal Mesh Blending and
Correspondence
cs.CV cs.GR
We have proposed a self-supervised deep learning framework for solving the
mesh blending problem in scenarios where the meshes are not in correspondence.
To solve this problem, we have developed Red-Blue MPNN, a novel graph neural
network that processes an augmented graph to estimate the correspondence. We
have designed a novel conditional refinement scheme to find the exact
correspondence when certain conditions are satisfied. We further develop a
graph neural network that takes the aligned meshes and the time value as input
and fuses this information to process further and generate the desired result.
Using motion capture datasets and human mesh designing software, we create a
large-scale synthetic dataset consisting of temporal sequences of human meshes
in motion. Our results demonstrate that our approach generates realistic
deformation of body parts given complex inputs.
| arxiv topic:cs.CV cs.GR |
arxiv_dataset-186672306.13552 | Techno-economic analysis of renewable energy generation at the South
Pole
physics.soc-ph hep-ex
Transitioning from fossil-fuel power generation to renewable energy
generation and energy storage in remote locations has the potential to reduce
both carbon emissions and cost. This study presents a techno-economic analysis
for implementation of a hybrid renewable energy system at the South Pole in
Antarctica, which currently hosts several high-energy physics experiments with
nontrivial power needs. A tailored model of resource availability and economics
for solar photovoltaics, wind turbine generators, lithium-ion energy storage,
and long-duration energy storage at this site is explored in different
combinations with and without existing diesel energy generation. The Renewable
Energy Integration and Optimization (REopt) platform is used to determine the
optimal system component sizing and the associated system economics and
environmental benefit. We find that the least-cost system includes all three
energy generation sources and lithium-ion energy storage. For an example
steady-state load of 170 kW, this hybrid system includes 180 kW-DC of
photovoltaic panels, 570 kW of wind turbines, and a 3.4 MWh lithium-ion battery
energy storage system. This system reduces diesel consumption by 95% compared
to an all-diesel configuration, resulting in approximately 1200 metric tons of
carbon footprint avoided annually. Over the course of a 15-year analysis period
the reduced diesel usage leads to a net savings of 57 million United States
dollars, with a time to payback of approximately two years. All the scenarios
modeled show that the transition to renewables is highly cost effective under
the unique economics and constraints of this extremely remote site.
| arxiv topic:physics.soc-ph hep-ex |
arxiv_dataset-186682306.13652 | Theory for dissipative time crystals in coupled parametric oscillators
cond-mat.stat-mech cond-mat.quant-gas
Discrete time crystals are novel phases of matter that break the discrete
time translational symmetry of a periodically driven system. In this work, we
propose a classical system of weakly-nonlinear parametrically-driven coupled
oscillators as a testbed to understand these phases. Such a system of
parametric oscillators can be used to model period-doubling instabilities of
Josephson junction arrays as well as semiconductor lasers. To show that this
instability leads to a discrete time crystal we first show that a certain limit
of the system is close to Langevin dynamics in a symmetry breaking potential.
We numerically show that this phase exists even in the presence of Ising
symmetry breaking using a Glauber dynamics approximation. We then use a field
theoretic argument to show that these results are robust to other
approximations including the semiclassical limit when applied to dissipative
quantum systems.
| arxiv topic:cond-mat.stat-mech cond-mat.quant-gas |
arxiv_dataset-186692306.13752 | Randomized compiling in fault-tolerant quantum computation
quant-ph
Studies of quantum error correction (QEC) typically focus on stochastic Pauli
errors because the existence of a threshold error rate below which stochastic
Pauli errors can be corrected implies that there exists a threshold below which
generic errors can be corrected. However, rigorous estimates of the threshold
for generic errors are typically orders of magnitude worse than the threshold
for stochastic Pauli errors. Specifically, coherent errors have a particularly
harmful impact on the encoded space because they can map encoded states to
superpositions of logical and error states. Further, coherent errors can add up
and interfere over multiple rounds of error correction or between syndrome
measurements, which may result in significantly worse errors than expected
under a stochastic Pauli error model. In this paper, we present an algorithm
which decoheres noise at the logical level, projecting the state of the system
onto a logical state with a well-defined error. The algorithm does not
significantly increase the depth of the logical circuit (and usually does not
lead to any increase in depth), and applies generally to most fault-tolerant
gadgets and error correction steps.
| arxiv topic:quant-ph |
arxiv_dataset-186702306.13852 | Gd-Based Solvated Shells for Defect Passivation of CsPbBr$_3$
Nanoplatelets Enabling Efficient Color-Saturated Blue Electroluminescence
physics.app-ph cond-mat.mtrl-sci
Reduced-dimensional CsPbBr$_3$ nanoplatelets (NPLs) are promising candidates
for color-saturated blue emitters, yet their electroluminescence performance is
hampered by non-radiative recombination, which is associated with bromine
vacancies. Here, we show that a post-synthetic treatment of CsPbBr$_3$ NPLs
with GdBr$_3$-dimethylformamide (DMF) can effectively eliminate defects while
preserving the color. According to a combined experimental and theoretical
study, Gd$^{3+}$ ions are less reactive with NPLs as a result of compact
interaction between them and DMF, and this stable Gd$^{3+}$-DMF solvation
structure makes Brions more available and allows them to move more freely.
Consequently, defects are rapidly passivated and photoluminescence quantum
yield increases dramatically (from 35 to ~100%), while the surface ligand
density and emission color remain unchanged. The result is a remarkable
electroluminescence efficiency of 2.4% (at 464 nm), one of the highest in pure
blue perovskite NPL light-emitting diodes. It is noteworthy that the conductive
NPL film shows a high photoluminescence quantum yield of 80%, demonstrating
NPLs' significant electroluminescence potential with further device structure
design.
| arxiv topic:physics.app-ph cond-mat.mtrl-sci |
arxiv_dataset-186712306.13952 | Artificial intelligence and biological misuse: Differentiating risks of
language models and biological design tools
cs.CY
As advancements in artificial intelligence (AI) propel progress in the life
sciences, they may also enable the weaponisation and misuse of biological
agents. This article differentiates two classes of AI tools that could pose
such biosecurity risks: large language models (LLMs) and biological design
tools (BDTs). LLMs, such as GPT-4 and its successors, might provide dual-use
information and thus remove some barriers encountered by historical biological
weapons efforts. As LLMs are turned into multi-modal lab assistants and
autonomous science tools, this will increase their ability to support
non-experts in performing laboratory work. Thus, LLMs may in particular lower
barriers to biological misuse. In contrast, BDTs will expand the capabilities
of sophisticated actors. Concretely, BDTs may enable the creation of pandemic
pathogens substantially worse than anything seen to date and could enable forms
of more predictable and targeted biological weapons. In combination, the
convergence of LLMs and BDTs could raise the ceiling of harm from biological
agents and could make them broadly accessible. A range of interventions would
help to manage risks. Independent pre-release evaluations could help understand
the capabilities of models and the effectiveness of safeguards. Options for
differentiated access to such tools should be carefully weighed with the
benefits of openly releasing systems. Lastly, essential for mitigating risks
will be universal and enhanced screening of gene synthesis products.
| arxiv topic:cs.CY |
arxiv_dataset-186722306.14052 | A Survey on Graph Neural Network Acceleration: Algorithms, Systems, and
Customized Hardware
cs.LG cs.AR cs.DC
Graph neural networks (GNNs) are emerging for machine learning research on
graph-structured data. GNNs achieve state-of-the-art performance on many tasks,
but they face scalability challenges when it comes to real-world applications
that have numerous data and strict latency requirements. Many studies have been
conducted on how to accelerate GNNs in an effort to address these challenges.
These acceleration techniques touch on various aspects of the GNN pipeline,
from smart training and inference algorithms to efficient systems and
customized hardware. As the amount of research on GNN acceleration has grown
rapidly, there lacks a systematic treatment to provide a unified view and
address the complexity of relevant works. In this survey, we provide a taxonomy
of GNN acceleration, review the existing approaches, and suggest future
research directions. Our taxonomic treatment of GNN acceleration connects the
existing works and sets the stage for further development in this area.
| arxiv topic:cs.LG cs.AR cs.DC |
arxiv_dataset-186732306.14152 | Low-Rank Prune-And-Factorize for Language Model Compression
cs.CL
The components underpinning PLMs -- large weight matrices -- were shown to
bear considerable redundancy. Matrix factorization, a well-established
technique from matrix theory, has been utilized to reduce the number of
parameters in PLM. However, it fails to retain satisfactory performance under
moderate to high compression rate. In this paper, we identify the
\textit{full-rankness} of fine-tuned PLM as the fundamental bottleneck for the
failure of matrix factorization and explore the use of network pruning to
extract low-rank sparsity pattern desirable to matrix factorization. We find
such low-rank sparsity pattern exclusively exists in models generated by
first-order pruning, which motivates us to unite the two approaches and achieve
more effective model compression. We further propose two techniques:
sparsity-aware SVD and mixed-rank fine-tuning, which improve the initialization
and training of the compression procedure, respectively. Experiments on GLUE
and question-answering tasks show that the proposed method has superior
compression-performance trade-off compared to existing approaches.
| arxiv topic:cs.CL |
arxiv_dataset-186742306.14252 | Standing waves with prescribed $L^2$-norm to nonlinear Schr\"odinger
equations with combined inhomogeneous nonlinearities
math.AP
In this paper, we are concerned with solutions to the following nonlinear
Schr\"odinger equation with combined inhomogeneous nonlinearities, $$ -\Delta u
+ \lambda u= \mu |x|^{-b}|u|^{q-2} u + |x|^{-b}|u|^{p-2} u \quad \mbox{in} \,\,
\R^N, $$ under the $L^2$-norm constraint $$ \int_{\R^N} |u|^2 \, dx=c>0, $$
where $N \geq 1$, $\mu =\pm 1$, $2<q<p<{2(N-b)}/{(N-2)^+}$, $0<b<\min\{2, N\}$
and the parameter $\lambda \in \R$ appearing as Lagrange multiplier is unknown.
In the mass subcritical case, we establish the compactness of any minimizing
sequence to the minimization problem given by the underlying energy functional
restricted on the constraint. As a consequence of the compactness of any
minimizing sequence, orbital stability of minimizers is derived. In the mass
critical and supercritical cases, we investigate the existence, radial symmetry
and orbital instability of solutions. Meanwhile, we consider the existence,
radial symmetry and algebraical decay of ground states to the corresponding
zero mass equation with defocusing perturbation. In addition, dynamical
behaviors of solutions to the Cauchy problem for the associated dispersive
equation are discussed.
| arxiv topic:math.AP |
arxiv_dataset-186752306.14352 | A simple and self-contained proof for the Lindemann-Weierstrass theorem
math.NT
The famous result of Lindemann and Weierstrass says that if
$a_{1},a_{2},\ldots,a_{n}$ are distinct algebraic numbers, then
$e^{a_{1}},e^{a_{2}},\ldots,e^{a_{n}}$ are linearly independent complex numbers
over the field $\overline{\mathbb{Q}}$ of all algebraic numbers.
Starting from some basic ideas of Hermite, Lindemann, Hilbert, Hurwitz and
Baker, in this note we provide an easy to understand and self-contained proof
for the Lindemann-Weierstrass Theorem. In an introductory section we have
gathered all the algebraic number theory tools that are necessary to prove the
main theorem. All these auxiliary results are fully proved in a simple and
elementary way, so that the paper can be read even by an undergraduate student.
| arxiv topic:math.NT |
arxiv_dataset-186762306.14452 | Phenomenon of multiple reentrant localization in a double-stranded helix
with transverse electric field
cond-mat.mes-hall cond-mat.dis-nn cond-mat.str-el physics.comp-ph quant-ph
The present work explores the potential for observing multiple reentrant
localization behavior in a double-stranded helical (DSH) system, extending
beyond the conventional nearest-neighbor hopping interaction. The DSH system is
considered to have hopping dimerization in each strand, while also being
subjected to a transverse electric field. The inclusion of an electric field
serves the dual purpose of inducing quasiperiodic disorder and strand-wise
staggered site energies. Two reentrant localization regions are identified: one
exhibiting true extended behavior in the thermodynamic limit, while the second
region shows quasi-extended characteristics with partial spreading within the
helix. The DSH system exhibits three distinct single-particle mobility edges
linked to localization transitions present in the system. The analysis in this
study involves examining various parameters such as the single-particle energy
spectrum, inverse participation ratio, local probability amplitude, and more.
Our proposal, combining achievable hopping dimerization and induced correlated
disorder, presents a unique opportunity to study phenomenon of reentrant
localization, generating significant research interest.
| arxiv topic:cond-mat.mes-hall cond-mat.dis-nn cond-mat.str-el physics.comp-ph quant-ph |
arxiv_dataset-186772306.14552 | Radiative Decays of the Spin-$\nicefrac{3}{2}$ to Spin-$\nicefrac{1}{2}$
Doubly Heavy Baryons in QCD
hep-ph
The spin-$\nicefrac{3}{2}$ to spin-$\nicefrac{1}{2}$ doubly heavy baryon
transition magnetic dipole $G_M$ and electric quadrupole $G_E$ formfactors are
calculated in the framework of light cone sum rules method. Moreover, the decay
widths of corresponding radiative transitions are estimated. Obtained results
of magnetic dipole moments $G_M$ and decay widths are compared with the results
present in the literature.
| arxiv topic:hep-ph |
arxiv_dataset-186782306.14652 | Strongly coupled fermions in odd dimensions and the running cut-off
$\Lambda_d$
hep-th
I study the fermionic $U(N)$ Gross-Neveu model at imaginary chemical
potential and finite temperature for odd $d$ dimensions, in the strong coupling
regime, by using the gap (saddle point) equation for the fermion condensate of
the model. This equation describes the phase transitions from weak to strong
coupling regime. I point out that the higher odd dimensional gap equations are
linear combinations of the lower dimensional equations in a way that as the
dimension of the model increases the lower dimensions are weaker coupled but
still in the strong coupling regime. Interestingly, at a specific value of the
chemical potential, exactly in the middle of the thermal windows that separate
the fermionic from the bosonic (condensed) state of the fermions, I find the
mass of the fermion condensate for $d=3,5,7,9$. An anomaly occurs at the $5$
dimensional theory where it is stronger coupled against other theories in
higher dimensions and lower energy. The main idea of this work is that the
cut-off $\Lambda$ regulator for the UV divergent parts of the fermion mass
saddle point equation, plays the role of a physical parameter. This idea is
based on the identity of the asymptotic freedom of the Gross-Neveu model as a
toy model for QCD.
| arxiv topic:hep-th |
arxiv_dataset-186792306.14752 | MedLSAM: Localize and Segment Anything Model for 3D CT Images
cs.CV
Recent advancements in foundation models have shown significant potential in
medical image analysis. However, there is still a gap in models specifically
designed for medical image localization. To address this, we introduce MedLAM,
a 3D medical foundation localization model that accurately identifies any
anatomical part within the body using only a few template scans. MedLAM employs
two self-supervision tasks: unified anatomical mapping (UAM) and multi-scale
similarity (MSS) across a comprehensive dataset of 14,012 CT scans.
Furthermore, we developed MedLSAM by integrating MedLAM with the Segment
Anything Model (SAM). This innovative framework requires extreme point
annotations across three directions on several templates to enable MedLAM to
locate the target anatomical structure in the image, with SAM performing the
segmentation. It significantly reduces the amount of manual annotation required
by SAM in 3D medical imaging scenarios. We conducted extensive experiments on
two 3D datasets covering 38 distinct organs. Our findings are twofold: 1)
MedLAM can directly localize anatomical structures using just a few template
scans, achieving performance comparable to fully supervised models; 2) MedLSAM
closely matches the performance of SAM and its specialized medical adaptations
with manual prompts, while minimizing the need for extensive point annotations
across the entire dataset. Moreover, MedLAM has the potential to be seamlessly
integrated with future 3D SAM models, paving the way for enhanced segmentation
performance. Our code is public at \href{https://github.com/openmedlab/MedLSAM}
| arxiv topic:cs.CV |
arxiv_dataset-186802306.14852 | CoarsenConf: Equivariant Coarsening with Aggregated Attention for
Molecular Conformer Generation
cs.LG physics.chem-ph q-bio.BM
Molecular conformer generation (MCG) is an important task in cheminformatics
and drug discovery. The ability to efficiently generate low-energy 3D
structures can avoid expensive quantum mechanical simulations, leading to
accelerated virtual screenings and enhanced structural exploration. Several
generative models have been developed for MCG, but many struggle to
consistently produce high-quality conformers. To address these issues, we
introduce CoarsenConf, which coarse-grains molecular graphs based on torsional
angles and integrates them into an SE(3)-equivariant hierarchical variational
autoencoder. Through equivariant coarse-graining, we aggregate the fine-grained
atomic coordinates of subgraphs connected via rotatable bonds, creating a
variable-length coarse-grained latent representation. Our model uses a novel
aggregated attention mechanism to restore fine-grained coordinates from the
coarse-grained latent representation, enabling efficient generation of accurate
conformers. Furthermore, we evaluate the chemical and biochemical quality of
our generated conformers on multiple downstream applications, including
property prediction and oracle-based protein docking. Overall, CoarsenConf
generates more accurate conformer ensembles compared to prior generative
models.
| arxiv topic:cs.LG physics.chem-ph q-bio.BM |
arxiv_dataset-186812306.14952 | Spontaneous CP violation and $\mu-\tau$ symmetry in two-Higgs-doublet
models with flavour conservation
hep-ph
In multi-Higgs-doublet models, requiring simultaneously that (i) CP violation
only arises spontaneously, (ii) tree level scalar flavour changing couplings
are absent and (iii) the fermion mixing matrix is CP violating, can only be
achieved in a very specific manner. A general approach with new clarifying
insights on the question is presented. Considering the quark sector, that
peculiar possibility is not viable on phenomenological grounds. We show that,
considering the lepton sector, it is highly interesting and leads to viable
models with $\mu-\tau$ symmetric PMNS matrices. Phenomenological implications
of the models, both for Dirac and Majorana (in a type I seesaw scenario)
neutrinos, are analysed.
| arxiv topic:hep-ph |
arxiv_dataset-186822306.15052 | Collinear matching for next-to-leading power transverse-momentum
distributions
hep-ph hep-th
Phenomenological studies of transverse momentum dependent (TMD) parton
distributions rely on the expansion in small values of the transverse
separation of fields, where TMD parton distributions match onto collinear
parton distribution functions. In this work, we derive this expansion at
tree-level for the genuine next-to-leading power quark-gluon-quark TMDs, taking
into account all the target mass corrections. We find that only a limited
number of TMD parton distributions exhibit matching to twist-three collinear
distributions, leading to a significant simplification in the analysis of the
structure functions of semi-inclusive deep inelastic scattering.
| arxiv topic:hep-ph hep-th |
arxiv_dataset-186832306.15152 | Molecular optomechanics in the anharmonic regime: from nonclassical
mechanical states to mechanical lasing
quant-ph physics.optics
Cavity optomechanics aims to establish optical control over vibrations of
mechanical systems, to heat, cool or to drive them toward coherent, or
nonclassical states. This field was recently extended to include molecular
optomechanics, which describes the dynamics of THz molecular vibrations coupled
to the optical fields of lossy cavities via Raman transitions, and was
developed to understand the anomalous amplification of optical phonons in
Surface-Enhanced Raman Scattering experiments. But the molecular platform
should prove suitable for demonstrating more sophisticated optomechanical
effects, including engineering of nonclassical mechanical states, or inducing
coherent molecular vibrations. In this work, we propose two pathways towards
implementing these effects, enabled or revealed by the strong intrinsic
anharmonicities of molecular vibrations. First, to prepare a nonclassical
mechanical state, we propose an incoherent analogue of the mechanical blockade,
in which the molecular aharmonicity and optical response of hybrid cavities
isolate the two lowest-energy vibrational states. Secondly, we show that for a
strongly driven optomechanical system, the anharmonicity can effectively
suppress the mechanical amplification, shifting and reshaping the onset of
coherent mechanical oscillations. Our estimates indicate that both effects
should be within reach of the existing implementations of the Surface Enhanced
Raman Scattering, opening the pathway towards the coherent and nonclassical
effects in molecular optomechanics.
| arxiv topic:quant-ph physics.optics |
arxiv_dataset-186842306.15252 | Platinum-absorbed Defective 2D Monolayer Boron Nitride: A Promising
Electrocatalyst for O2 Reduction Reaction
cond-mat.mtrl-sci
The large bandgap and strong covalent bonds of hexagonal boron nitride (hBN)
had long been thought to be chemically inert. Due to its inertness with
saturated robust covalent bonds, the pristine 2D monolayer hBN cannot be
functionalized for applications of energy conversion. Therefore, it is
necessary to make the 2D hBN chemically reactive for potential applications.
Here, we have computationally designed a single nitrogen (N) and boron (B)
di-vacancy of the 2D monolayer hBN, noted by VBN defective-BN (d-BN), to
activate the chemical reactivity, which is an effective strategy to use the
d-BN for potential applications. Single Pt atom absorbed on the defective area
of the VBN d-BN acts as a single-atom catalyst which exhibits distinctive
performances for O2 reduction reaction (ORR). First-principles based
dispersion-corrected periodic hybrid Density Functional Theory (DFT-D) method
has been employed to investigate the equilibrium structure and properties of
the Pt-absorbed 2D defective boron nitride (Pt-d-BN). The present study shows
the semiconducting character of Pt-d-BN with an electronic bandgap of 1.30 eV,
which is an essential aspect of the ORR. The ORR mechanism on the surface of 2D
monolayer Pt-d-BN follows a 4e-reduction route because of the low barriers to
OOH formation and dissociation, H2O2 instability and water production at the
Pt-d-BN surface. Here, both the dissociative and associative ORR mechanisms
have been investigated, and it is found that results for both mechanisms with
the ORR pathways are almost equally favorable. Therefore, it can be mentioned
here that the 2D monolayer Pt-d-BN exhibits a high selectivity for the
four-electron reduction pathway. According to the calculations of the relative
adsorption energy of each step in ORR, the Pt-d-BN is anticipated to exhibit
substantial catalytic activity.
| arxiv topic:cond-mat.mtrl-sci |
arxiv_dataset-186852306.15352 | Theory of active self-organization of dense nematic structures in the
actin cytoskeleton
physics.bio-ph cond-mat.soft
The actin cytoskeleton is remarkably adaptable and multifunctional. It often
organizes into nematic bundles such as contractile rings or stress fibers.
However, how a uniform and isotropic actin gel self-organizes into dense
nematic bundles is not fully understood. Here, using an active gel model
accounting for nematic order and density variations, we identify an active
patterning mechanism leading to localized dense nematic structures. Linear
stability analysis and nonlinear finite element simulations establish the
conditions for nematic bundle self-assembly and how active gel parameters
control the architecture, orientation, connectivity and dynamics of
self-organized patterns. Finally, we substantiate with discrete network
simulations the main requirements for nematic bundle formation according to our
theory, namely increased active tension perpendicular to the nematic direction
and generalized active forces conjugate to nematic order. Our work portrays
actin gels a reconfigurable active materials with a spontaneous tendency to
develop patterns of dense nematic bundles.
| arxiv topic:physics.bio-ph cond-mat.soft |
arxiv_dataset-186862306.15452 | On fractional quasilinear equations with elliptic degeneracy
math.AP
In this work, we present a systematic approach to investigate the existence,
multiplicity, and local gradient regularity of solutions for nonlocal
quasilinear equations with local gradient degeneracy. Our method involves an
interactive geometric argument that interplays with uniqueness property for the
corresponding homogeneous problem, leading with gradient H\"older regularity
estimates. This approach is intrinsically developed for nonlocal scenarios,
where uniqueness holds for the local homogeneous problem. We illustrate our
results by showing classes of exterior data that exhibit multiple solutions,
while also highlighting relevant cases where uniqueness is confirmed.
| arxiv topic:math.AP |
arxiv_dataset-186872306.15552 | A Survey on Deep Learning Hardware Accelerators for Heterogeneous HPC
Platforms
cs.AR cs.ET cs.LG
Recent trends in deep learning (DL) have made hardware accelerators essential
for various high-performance computing (HPC) applications, including image
classification, computer vision, and speech recognition. This survey summarizes
and classifies the most recent developments in DL accelerators, focusing on
their role in meeting the performance demands of HPC applications. We explore
cutting-edge approaches to DL acceleration, covering not only GPU- and
TPU-based platforms but also specialized hardware such as FPGA- and ASIC-based
accelerators, Neural Processing Units, open hardware RISC-V-based accelerators,
and co-processors. This survey also describes accelerators leveraging emerging
memory technologies and computing paradigms, including 3D-stacked
Processor-In-Memory, non-volatile memories like Resistive RAM and Phase Change
Memories used for in-memory computing, as well as Neuromorphic Processing
Units, and Multi-Chip Module-based accelerators. Furthermore, we provide
insights into emerging quantum-based accelerators and photonics. Finally, this
survey categorizes the most influential architectures and technologies from
recent years, offering readers a comprehensive perspective on the rapidly
evolving field of deep learning acceleration.
| arxiv topic:cs.AR cs.ET cs.LG |
arxiv_dataset-186882306.15652 | Complex fluid models of mixed quantum-classical dynamics
math-ph math.MP physics.chem-ph physics.flu-dyn quant-ph
Several methods in nonadiabatic molecular dynamics are based on Madelung's
hydrodynamic description of nuclear motion, while the electronic component is
treated as a finite-dimensional quantum system. In this context, the quantum
potential leads to severe computational challenges and one often seeks to
neglect its contribution, thereby approximating nuclear motion as classical.
The resulting model couples classical hydrodynamics for the nuclei to the
quantum motion of the electronic component, leading to the structure of a
complex fluid system. This type of mixed quantum-classical fluid models have
also appeared in solvation dynamics to describe the coupling between liquid
solvents and the quantum solute molecule. While these approaches represent a
promising direction, their mathematical structure requires a certain care. In
some cases, challenging higher-order gradients make these equations hardly
tractable. In other cases, these models are based on phase-space formulations
that suffer from well-known consistency issues. Here, we present a new complex
fluid system that resolves these difficulties. Unlike common approaches, the
current system is obtained by applying the fluid closure at the level of the
action principle of the original phase-space model. As a result, the system
inherits a Hamiltonian structure and retains energy/momentum balance. After
discussing some of its structural properties and dynamical invariants, we
illustrate the model in the case of pure-dephasing dynamics. We conclude by
presenting some invariant planar models.
| arxiv topic:math-ph math.MP physics.chem-ph physics.flu-dyn quant-ph |
arxiv_dataset-186892306.15752 | On the almost-palindromic width of free groups
math.GR math.CO
We answer a question of Bardakov (Kourovka Notebook, Problem 19.8) which asks
for the existence of a pair of natural numbers $(c, m)$ with the property that
every element in the free group on the two-element set $\{a, b\}$ can be
represented as a concatenation of $c$, or fewer, $m$-almost-palindromes in
letters $a^{\pm 1}, b^{\pm 1}$. Here, an $m$-almost-palindrome is a word which
can be obtained from a palindrome by changing at most $m$ letters. We show that
no such pair $(c, m)$ exists. In fact, we show that the analogous result holds
for all non-abelian free groups.
| arxiv topic:math.GR math.CO |
arxiv_dataset-186902306.15852 | Action-conditioned Deep Visual Prediction with RoAM, a new Indoor Human
Motion Dataset for Autonomous Robots
cs.RO cs.CV
With the increasing adoption of robots across industries, it is crucial to
focus on developing advanced algorithms that enable robots to anticipate,
comprehend, and plan their actions effectively in collaboration with humans. We
introduce the Robot Autonomous Motion (RoAM) video dataset, which is collected
with a custom-made turtlebot3 Burger robot in a variety of indoor environments
recording various human motions from the robot's ego-vision. The dataset also
includes synchronized records of the LiDAR scan and all control actions taken
by the robot as it navigates around static and moving human agents. The unique
dataset provides an opportunity to develop and benchmark new visual prediction
frameworks that can predict future image frames based on the action taken by
the recording agent in partially observable scenarios or cases where the
imaging sensor is mounted on a moving platform. We have benchmarked the dataset
on our novel deep visual prediction framework called ACPNet where the
approximated future image frames are also conditioned on action taken by the
robot and demonstrated its potential for incorporating robot dynamics into the
video prediction paradigm for mobile robotics and autonomous navigation
research.
| arxiv topic:cs.RO cs.CV |
arxiv_dataset-186912306.15952 | A minimal completion theorem and almost everywhere equivalence for
Completely Positive maps
math.OA quant-ph
A problem of completing a linear map on C*-algebras to a completely positive
map is analyzed. It is shown that whenever such a completion is feasible there
exists a unique minimal completion. This theorem is used to show that under
some very general conditions a completely positive map almost everywhere
equivalent to a quasi-pure map is actually equal to that map.
| arxiv topic:math.OA quant-ph |
arxiv_dataset-186922306.16052 | SVNR: Spatially-variant Noise Removal with Denoising Diffusion
cs.CV
Denoising diffusion models have recently shown impressive results in
generative tasks. By learning powerful priors from huge collections of training
images, such models are able to gradually modify complete noise to a clean
natural image via a sequence of small denoising steps, seemingly making them
well-suited for single image denoising. However, effectively applying denoising
diffusion models to removal of realistic noise is more challenging than it may
seem, since their formulation is based on additive white Gaussian noise, unlike
noise in real-world images. In this work, we present SVNR, a novel formulation
of denoising diffusion that assumes a more realistic, spatially-variant noise
model. SVNR enables using the noisy input image as the starting point for the
denoising diffusion process, in addition to conditioning the process on it. To
this end, we adapt the diffusion process to allow each pixel to have its own
time embedding, and propose training and inference schemes that support
spatially-varying time maps. Our formulation also accounts for the correlation
that exists between the condition image and the samples along the modified
diffusion process. In our experiments we demonstrate the advantages of our
approach over a strong diffusion model baseline, as well as over a
state-of-the-art single image denoising method.
| arxiv topic:cs.CV |
arxiv_dataset-186932306.16152 | Exponential Adoption of Battery Electric Cars
physics.soc-ph
The adoption of battery electric vehicles (BEVs) may significantly reduce
greenhouse gas emissions caused by road transport. However, there is wide
disagreement as to how soon battery electric vehicles will play a major role in
overall transportation. Focusing on battery electric passenger cars, we here
analyze BEV adoption across 17 individual countries, Europe, and the World, and
consistently find exponential growth trends. Modeling-based estimates of future
adoption given past trends suggests system-wide adoption substantially faster
than typical economic analyses have proposed so far. For instance, we estimate
the majority of passenger cars in Europe to be electric by about 2031. Within
regions, the predicted times of mass adoption are largely insensitive to model
details. Despite significant differences in current electric fleet sizes across
regions, their growth rates consistently indicate fast doubling times of
approximately 15 months, hinting at radical economic and infrastructural
consequences in the near future.
| arxiv topic:physics.soc-ph |
arxiv_dataset-186942306.16252 | Land Cover Segmentation with Sparse Annotations from Sentinel-2 Imagery
cs.CV
Land cover (LC) segmentation plays a critical role in various applications,
including environmental analysis and natural disaster management. However,
generating accurate LC maps is a complex and time-consuming task that requires
the expertise of multiple annotators and regular updates to account for
environmental changes. In this work, we introduce SPADA, a framework for fuel
map delineation that addresses the challenges associated with LC segmentation
using sparse annotations and domain adaptation techniques for semantic
segmentation. Performance evaluations using reliable ground truths, such as
LUCAS and Urban Atlas, demonstrate the technique's effectiveness. SPADA
outperforms state-of-the-art semantic segmentation approaches as well as
third-party products, achieving a mean Intersection over Union (IoU) score of
42.86 and an F1 score of 67.93 on Urban Atlas and LUCAS, respectively.
| arxiv topic:cs.CV |
arxiv_dataset-186952306.16352 | Information-Computation Tradeoffs for Learning Margin Halfspaces with
Random Classification Noise
cs.LG cs.DS math.ST stat.ML stat.TH
We study the problem of PAC learning $\gamma$-margin halfspaces with Random
Classification Noise. We establish an information-computation tradeoff
suggesting an inherent gap between the sample complexity of the problem and the
sample complexity of computationally efficient algorithms. Concretely, the
sample complexity of the problem is $\widetilde{\Theta}(1/(\gamma^2
\epsilon))$. We start by giving a simple efficient algorithm with sample
complexity $\widetilde{O}(1/(\gamma^2 \epsilon^2))$. Our main result is a lower
bound for Statistical Query (SQ) algorithms and low-degree polynomial tests
suggesting that the quadratic dependence on $1/\epsilon$ in the sample
complexity is inherent for computationally efficient algorithms. Specifically,
our results imply a lower bound of $\widetilde{\Omega}(1/(\gamma^{1/2}
\epsilon^2))$ on the sample complexity of any efficient SQ learner or
low-degree test.
| arxiv topic:cs.LG cs.DS math.ST stat.ML stat.TH |
arxiv_dataset-186962306.16452 | Exact description of transport and non-reciprocity in monitored quantum
devices
quant-ph cond-mat.mes-hall cond-mat.stat-mech
We study non-interacting fermionic systems undergoing continuous monitoring
and driven by biased reservoirs. Averaging over the measurement outcomes, we
derive exact formulas for the particle and heat flows in the system. We show
that these currents feature competing elastic and inelastic components, which
depend non-trivially on the monitoring strength $\gamma$. We highlight that
monitor-induced inelastic processes lead to non-reciprocal currents, allowing
to extract work from measurements without active feedback control. We
illustrate our formalism with two distinct monitoring schemes providing
measurement-induced power or cooling.~Optimal performances are found for values
of the monitoring strength $\gamma$ which are hard to address with perturbative
approaches.
| arxiv topic:quant-ph cond-mat.mes-hall cond-mat.stat-mech |
arxiv_dataset-186972306.16552 | Learning Fair Classifiers via Min-Max F-divergence Regularization
cs.LG cs.AI cs.CY cs.IT math.IT
As machine learning (ML) based systems are adopted in domains such as law
enforcement, criminal justice, finance, hiring and admissions, ensuring the
fairness of ML aided decision-making is becoming increasingly important. In
this paper, we focus on the problem of fair classification, and introduce a
novel min-max F-divergence regularization framework for learning fair
classification models while preserving high accuracy. Our framework consists of
two trainable networks, namely, a classifier network and a bias/fairness
estimator network, where the fairness is measured using the statistical notion
of F-divergence. We show that F-divergence measures possess convexity and
differentiability properties, and their variational representation make them
widely applicable in practical gradient based training methods. The proposed
framework can be readily adapted to multiple sensitive attributes and for high
dimensional datasets. We study the F-divergence based training paradigm for two
types of group fairness constraints, namely, demographic parity and equalized
odds. We present a comprehensive set of experiments for several real-world data
sets arising in multiple domains (including COMPAS, Law Admissions, Adult
Income, and CelebA datasets). To quantify the fairness-accuracy tradeoff, we
introduce the notion of fairness-accuracy receiver operating characteristic
(FA-ROC) and a corresponding \textit{low-bias} FA-ROC, which we argue is an
appropriate measure to evaluate different classifiers. In comparison to several
existing approaches for learning fair classifiers (including pre-processing,
post-processing and other regularization methods), we show that the proposed
F-divergence based framework achieves state-of-the-art performance with respect
to the trade-off between accuracy and fairness.
| arxiv topic:cs.LG cs.AI cs.CY cs.IT math.IT |
arxiv_dataset-186982306.16652 | TimeClave: Oblivious In-enclave Time series Processing System
cs.CR
Cloud platforms are widely adopted by many systems, such as time series
processing systems, to store and process massive amounts of sensitive time
series data. Unfortunately, several incidents have shown that cloud platforms
are vulnerable to internal and external attacks that lead to critical data
breaches. Adopting cryptographic protocols such as homomorphic encryption and
secure multi-party computation adds high computational and network overhead to
query operations.
We present TimeClave, a fully oblivious in-enclave time series processing
system: TimeClave leverages Intel SGX to support aggregate statistics on time
series with minimal memory consumption inside the enclave. To hide the access
pattern inside the enclave, we introduce a non-blocking read-optimised ORAM
named RoORAM. TimeClave integrates RoORAM to obliviously and securely handle
client queries with high performance. With an aggregation time interval of
$10s$, $2^{14}$ summarised data blocks and 8 aggregate functions, TimeClave run
point query in $0.03ms$ and a range query of 50 intervals in $0.46ms$. Compared
to the ORAM baseline, TimeClave achieves lower query latency by up to
$2.5\times$ and up to $2\times$ throughput, with up to 22K queries per second.
| arxiv topic:cs.CR |
arxiv_dataset-186992306.16752 | Rapid FRD determination for multiplexed fibre systems -- I. The
quasi-near field model and its uncertainties
astro-ph.IM
Focal Ratio Degradation (FRD) in fibres is a crucial factor to control in
astronomical instruments in order to minimize light loss. As astronomical
instrumentation has advanced, the integration of large populations of fibres
has become common. However, determining FRD in multiplexed fibre systems has
become a challenging and time-consuming task. The Integral Field Unit for the
Fiber Arrayed Solar Optical Telescope (FASOT-IFU) represents the most densely
arranged fibre-based IFU in a single unit. Due to the close packing of fibres
in the V-groove of the slit end, measuring FRD is particularly challenging as
the output spots are prone to overlapping with adjacent fibres. In this paper,
a novel method based on the quasi-near field model is proposed to enable rapid
FRD measurement in highly multiplexed fibre systems like IFUs and multi-object
observation systems. The principle and uncertainties associated with the method
are investigated. The method's validity is demonstrated by applying it to
determine the FRD in FASOT-IFU, with the achieved FRD performance meeting the
acceptable requirements of FASOT-IFU, where the output focal ratio primarily
falls within the range of 5.0-7.0. The results indicate that the proposed
method offers several advantages, including the simultaneous and rapid
measurement of FRD in multiple fibres with high accuracy (error smaller than
0.35 in F-ratio). Furthermore, besides FRD, the method exhibits potential for
extensive measurements of throughput, scrambling, and spectral analysis.
| arxiv topic:astro-ph.IM |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.