title
stringlengths
15
153
paper_url
stringlengths
42
42
authors
listlengths
1
71
type
stringclasses
3 values
primary_area
stringclasses
0 values
abstract
large_stringlengths
422
2.2k
keywords
listlengths
0
0
TL;DR
large_stringclasses
0 values
submission_number
int64
2
10.2k
arxiv_id
stringlengths
10
10
arxiv_id_source
stringclasses
2 values
embedding
listlengths
768
768
On Discrete Prompt Optimization for Diffusion Models
https://openreview.net/forum?id=Fw4fBE2rqW
[ "Ruochen Wang", "Ting Liu", "Cho-Jui Hsieh", "Boqing Gong" ]
Poster
null
This paper introduces the first gradient-based framework for prompt optimization in text-to-image diffusion models. We formulate prompt engineering as a discrete optimization problem over the language space. Two major challenges arise in efficiently finding a solution to this problem: (1) Enormous Domain Space: Setting...
[]
null
10,207
2407.01606
title_snapshot
[ -0.029005568474531174, -0.02207097038626671, -0.010862091556191444, 0.07818451523780823, 0.023237142711877823, 0.033772312104701996, 0.017656298354268074, 0.0023722227197140455, 0.003975486382842064, -0.006495343055576086, -0.033706918358802795, 0.002007190603762865, -0.029573198407888412, ...
Multi-View Clustering by Inter-cluster Connectivity Guided Reward
https://openreview.net/forum?id=uEx2bSAJu8
[ "Hao Dai", "Yang Liu", "Peng Su", "Hecheng Cai", "Shudong Huang", "Jiancheng Lv" ]
Poster
null
Multi-view clustering has been widely explored for its effectiveness in harmonizing heterogeneity along with consistency in different views of data. Despite the significant progress made by recent works, the performance of most existing methods is heavily reliant on strong priori information regarding the true cluster ...
[]
null
10,190
null
null
[ -0.01635785959661007, -0.010016855783760548, 0.008748359978199005, 0.05831523239612579, 0.026984505355358124, 0.036001041531562805, 0.009738943539559841, -0.0026982638519257307, -0.025420324876904488, -0.046809010207653046, -0.02547571249306202, 0.013746246695518494, -0.06790728121995926, ...
Overestimation, Overfitting, and Plasticity in Actor-Critic: the Bitter Lesson of Reinforcement Learning
https://openreview.net/forum?id=5vZzmCeTYu
[ "Michal Nauman", "Michał Bortkiewicz", "Piotr Miłoś", "Tomasz Trzcinski", "Mateusz Ostaszewski", "Marek Cygan" ]
Poster
null
Recent advancements in off-policy Reinforcement Learning (RL) have significantly improved sample efficiency, primarily due to the incorporation of various forms of regularization that enable more gradient update steps than traditional agents. However, many of these techniques have been tested in limited settings, often...
[]
null
10,168
2403.00514
title_snapshot
[ -0.017980314791202545, -0.056917279958724976, 0.004743721801787615, 0.040997158735990524, 0.051658276468515396, -0.002001955872401595, 0.016939155757427216, -0.0000735143621568568, -0.06038643792271614, -0.027276471257209778, -0.0055153206922113895, 0.037408336997032166, -0.07924608141183853...
Copyright Traps for Large Language Models
https://openreview.net/forum?id=LDq1JPdc55
[ "Matthieu Meeus", "Igor Shilov", "Manuel Faysse", "Yves-Alexandre de Montjoye" ]
Poster
null
Questions of fair use of copyright-protected content to train Large Language Models (LLMs) are being actively debated. Document-level inference has been proposed as a new task: inferring from black-box access to the trained model whether a piece of content has been seen during training. SOTA methods however rely on nat...
[]
null
10,167
2402.09363
title_snapshot
[ -0.008653650060296059, -0.008818905800580978, -0.04982582852244377, 0.05854573845863342, 0.04955607280135155, -0.009293911047279835, 0.03844692185521126, 0.022408712655305862, -0.030352449044585228, 0.003585540223866701, -0.0234312005341053, 0.03754080459475517, -0.06733045727014542, -0.00...
Implicit meta-learning may lead language models to trust more reliable sources
https://openreview.net/forum?id=Fzp1DRzCIN
[ "Dmitrii Krasheninnikov", "Egor Krasheninnikov", "Bruno Kacper Mlodozeniec", "Tegan Maharaj", "David Krueger" ]
Poster
null
We demonstrate that large language models (LLMs) may learn indicators of document usefulness and modulate their updates accordingly. We introduce random strings ("tags") as indicators of usefulness in a synthetic fine-tuning dataset. Fine-tuning on this dataset leads to **implicit meta-learning (IML)**: in further fine...
[]
null
10,166
2310.15047
title_snapshot
[ -0.007131750229746103, 0.0034494162537157536, -0.008072172291576862, 0.045703113079071045, 0.03524107486009598, -0.012608935125172138, 0.032612890005111694, 0.04390069097280502, -0.03591356426477432, -0.0016159320948645473, -0.04002096876502037, 0.05568770319223404, -0.06496790796518326, -...
Dr. Strategy: Model-Based Generalist Agents with Strategic Dreaming
https://openreview.net/forum?id=HsseRq2FAx
[ "Hany Hamed", "Subin Kim", "Dongyeong Kim", "Jaesik Yoon", "Sungjin Ahn" ]
Poster
null
Model-based reinforcement learning (MBRL) has been a primary approach to ameliorating the sample efficiency issue as well as to make a generalist agent. However, there has not been much effort toward enhancing the strategy of dreaming itself. Therefore, it is a question *whether and how an agent can ``*dream better*''*...
[]
null
10,161
2402.18866
title_snapshot
[ -0.032165851444005966, 0.0026568486355245113, 0.009415887296199799, 0.009269989095628262, 0.045797064900398254, 0.0033431891351938248, 0.03405240178108215, 0.003738800063729286, -0.050060201436281204, -0.04164959490299225, -0.05357632786035538, 0.025966379791498184, -0.05962399020791054, -...
When Will Gradient Regularization Be Harmful?
https://openreview.net/forum?id=60vC1FY0dZ
[ "Yang Zhao", "Hao Zhang", "Xiuyuan Hu" ]
Poster
null
Gradient regularization (GR), which aims to penalize the gradient norm atop the loss function, has shown promising results in training modern over-parameterized deep neural networks. However, can we trust this powerful technique? This paper reveals that GR can cause performance degeneration in adaptive optimization sce...
[]
null
10,159
2406.09723
title_snapshot
[ 0.010384432971477509, -0.033499736338853836, 0.016326874494552612, 0.019307497888803482, 0.038603525608778, 0.03403163328766823, 0.04178997501730919, 0.006283102557063103, -0.04080827534198761, -0.07890496402978897, -0.005790241993963718, 0.003993523772805929, -0.06891251355409622, -0.0039...
Efficient Algorithms for Empirical Group Distributionally Robust Optimization and Beyond
https://openreview.net/forum?id=pOJbk4Nzmi
[ "Dingzhi Yu", "Yunuo Cai", "Wei Jiang", "Lijun Zhang" ]
Poster
null
In this paper, we investigate the empirical counterpart of Group Distributionally Robust Optimization (GDRO), which aims to minimize the maximal empirical risk across $m$ distinct groups. We formulate empirical GDRO as a *two-level* finite-sum convex-concave minimax optimization problem and develop an algorithm called ...
[]
null
10,129
2403.03562
title_snapshot
[ -0.040993161499500275, 0.027499970048666, 0.032525476068258286, 0.03587815910577774, 0.03959919139742851, 0.057727959007024765, 0.016340360045433044, -0.02229715883731842, -0.01313838642090559, -0.047049492597579956, 0.021729519590735435, -0.04077303409576416, -0.053462523967027664, -0.019...
Evaluation of Trajectory Distribution Predictions with Energy Score
https://openreview.net/forum?id=FCmWhJQ14I
[ "Novin Shahroudi", "Mihkel Lepson", "Meelis Kull" ]
Poster
null
Predicting the future trajectory of surrounding objects is inherently uncertain and vital in the safe and reliable planning of autonomous systems such as in self-driving cars. Although trajectory prediction models have become increasingly sophisticated in dealing with the complexities of spatiotemporal data, the evalua...
[]
null
10,124
null
null
[ -0.016225822269916534, -0.009125330485403538, 0.011096997186541557, 0.014686101116240025, 0.04797637462615967, 0.021670585498213768, 0.01037600077688694, 0.01951606199145317, -0.03642716258764267, -0.04801744595170021, -0.009298953227698803, 0.0015478383284062147, -0.05946258082985878, -0....
Causal Discovery with Fewer Conditional Independence Tests
https://openreview.net/forum?id=HpT19AKddu
[ "Kirankumar Shiragur", "Jiaqi Zhang", "Caroline Uhler" ]
Poster
null
Many questions in science center around the fundamental problem of understanding causal relationships. However, most constraint-based causal discovery algorithms, including the well-celebrated PC algorithm, often incur an _exponential_ number of conditional independence (CI) tests, posing limitations in various applica...
[]
null
10,108
2406.01823
title_snapshot
[ -0.03422063589096069, -0.04159848764538765, -0.02547483518719673, 0.053814347833395004, 0.037994518876075745, 0.028844846412539482, 0.037130601704120636, 0.006806834600865841, -0.010020914487540722, -0.039727360010147095, 0.005665380973368883, 0.027439221739768982, -0.0453069731593132, 0.0...
Referee Can Play: An Alternative Approach to Conditional Generation via Model Inversion
https://openreview.net/forum?id=hZ0fWhgVch
[ "Xuantong LIU", "Tianyang Hu", "Wenjia Wang", "Kenji Kawaguchi", "Yuan Yao" ]
Poster
null
As a dominant force in text-to-image generation tasks, Diffusion Probabilistic Models (DPMs) face a critical challenge in controllability, struggling to adhere strictly to complex, multi-faceted instructions. In this work, we aim to address this alignment challenge for conditional generation tasks. First, we provide an...
[]
null
10,107
2402.16305
title_snapshot
[ -0.006583697162568569, -0.0029335005674511194, -0.02171701192855835, 0.06824018806219101, 0.05176224559545517, 0.02735331654548645, 0.024444852024316788, -0.018763434141874313, -0.01899167336523533, -0.04507284611463547, 0.005000112112611532, -0.0014507777523249388, -0.069633848965168, -0....
HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal
https://openreview.net/forum?id=f3TUipYU3U
[ "Mantas Mazeika", "Long Phan", "Xuwang Yin", "Andy Zou", "Zifan Wang", "Norman Mu", "Elham Sakhaee", "Nathaniel Li", "Steven Basart", "Bo Li", "David Forsyth", "Dan Hendrycks" ]
Poster
null
Automated red teaming holds substantial promise for uncovering and mitigating the risks associated with the malicious use of large language models (LLMs), yet the field lacks a standardized evaluation framework to rigorously assess new methods. To address this issue, we introduce HarmBench, a standardized evaluation fr...
[]
null
10,106
2402.04249
title_snapshot
[ 0.017368199303746223, -0.019870489835739136, -0.04779093712568283, 0.029769914224743843, 0.03912830725312233, -0.011694762855768204, 0.03959232568740845, 0.0035341738257557154, -0.018627453595399857, -0.02674632892012596, -0.023479148745536804, 0.01482401043176651, -0.06698243319988251, -0...
Adversarial Attacks on Combinatorial Multi-Armed Bandits
https://openreview.net/forum?id=0tPBk24xNj
[ "Rishab Balasubramanian", "Jiawei Li", "Prasad Tadepalli", "Huazheng Wang", "Qingyun Wu", "Haoyu Zhao" ]
Poster
null
We study reward poisoning attacks on Combinatorial Multi-armed Bandits (CMAB). We first provide a sufficient and necessary condition for the attackability of CMAB, a notion to capture the vulnerability and robustness of CMAB. The attackability condition depends on the intrinsic properties of the corresponding CMAB inst...
[]
null
10,092
2310.05308
title_snapshot
[ -0.01477771531790495, -0.025406518951058388, -0.016221318393945694, 0.05859646573662758, 0.0425599031150341, -0.004841598682105541, 0.026307182386517525, -0.009477141313254833, -0.02286042831838131, -0.03413312882184982, -0.03746802359819412, 0.009223097003996372, -0.06210273876786232, -0....
Exact Conversion of In-Context Learning to Model Weights in Linearized-Attention Transformers
https://openreview.net/forum?id=LVF4P1NNwO
[ "Brian K Chen", "Tianyang Hu", "Hui Jin", "Hwee Kuan Lee", "Kenji Kawaguchi" ]
Poster
null
In-Context Learning (ICL) has been a powerful emergent property of large language models that has attracted increasing attention in recent years. In contrast to regular gradient-based learning, ICL is highly interpretable and does not require parameter updates. In this paper, we show that, for linearized transformer ne...
[]
null
10,091
2406.02847
title_snapshot
[ -0.016283785924315453, 0.012183541432023048, -0.011952987872064114, 0.024777688086032867, 0.009856403805315495, 0.020418940111994743, 0.018665261566638947, 0.04501945152878761, -0.03149150684475899, 0.017129942774772644, -0.0307305958122015, 0.022371985018253326, -0.05147488787770271, -0.0...
Reshape and Adapt for Output Quantization (RAOQ): Quantization-aware Training for In-memory Computing Systems
https://openreview.net/forum?id=fM9xTkpAdu
[ "Bonan Zhang", "Chia-Yu Chen", "Naveen Verma" ]
Poster
null
In-memory computing (IMC) has emerged as a promising solution to address both computation and data-movement challenges, by performing computation on data in-place directly in the memory array. IMC typically relies on analog operation, which makes analog-to-digital converters (ADCs) necessary, for converting results bac...
[]
null
10,086
null
null
[ -0.03203420713543892, -0.03894379362463951, -0.027550382539629936, 0.0017211755039170384, 0.07072620838880539, 0.0676705464720726, 0.0020104344002902508, -0.01154628861695528, -0.04775640740990639, -0.03162934258580208, 0.02691168338060379, 0.0006702038226649165, -0.06778758019208908, -0.0...
Learning in Feature Spaces via Coupled Covariances: Asymmetric Kernel SVD and Nyström method
https://openreview.net/forum?id=Gp0xZDmrA2
[ "Qinghua Tao", "Francesco Tonin", "Alex Lambert", "Yingyi Chen", "Panagiotis Patrinos", "Johan Suykens" ]
Poster
null
In contrast with Mercer kernel-based approaches as used e.g. in Kernel Principal Component Analysis (KPCA), it was previously shown that Singular Value Decomposition (SVD) inherently relates to asymmetric kernels and Asymmetric Kernel Singular Value Decomposition (KSVD) has been proposed. However, the existing formulat...
[]
null
10,082
2406.08748
title_snapshot
[ -0.03817343711853027, -0.004111702088266611, 0.027407001703977585, 0.014708522707223892, 0.02781824953854084, 0.05947284772992134, 0.034831758588552475, -0.021667489781975746, -0.006636662874370813, -0.04605502635240555, -0.014628845266997814, 0.030746884644031525, -0.05348747596144676, 0....
Probabilistic Inference in Language Models via Twisted Sequential Monte Carlo
https://openreview.net/forum?id=frA0NNBS1n
[ "Stephen Zhao", "Rob Brekelmans", "Alireza Makhzani", "Roger Baker Grosse" ]
Oral
null
Numerous capability and safety techniques of Large Language Models (LLMs), including RLHF, automated red-teaming, prompt engineering, and infilling, can be cast as sampling from an unnormalized target distribution defined by a given reward or potential function over the full sequence. In this work, we leverage the rich...
[]
null
10,076
2404.17546
title_snapshot
[ -0.019519096240401268, -0.012829498387873173, -0.02037636563181877, 0.05688134953379631, 0.041175611317157745, 0.022300679236650467, 0.032682280987501144, 0.0021839269902557135, -0.019269827753305435, 0.008121699094772339, -0.014967676252126694, 0.03580092638731003, -0.07092513889074326, -...
Q-Star Meets Scalable Posterior Sampling: Bridging Theory and Practice via HyperAgent
https://openreview.net/forum?id=OF7e0w1uon
[ "Yingru Li", "Jiawei Xu", "Lei Han", "Zhi-Quan Luo" ]
Poster
null
We propose HyperAgent, a reinforcement learning (RL) algorithm based on the hypermodel framework for exploration in RL. HyperAgent allows for the efficient incremental approximation of posteriors associated with an optimal action-value function ($Q^\star$) without the need for conjugacy and follows the greedy policies ...
[]
null
10,071
2402.10228
title_snapshot
[ -0.02863902971148491, -0.01780306175351143, 0.006599251180887222, 0.06486168503761292, 0.032721616327762604, 0.003132913261651993, 0.022000988945364952, -0.03198328614234924, -0.01862945966422558, -0.04002338647842407, -0.020714398473501205, 0.02714828960597515, -0.08395279943943024, -0.00...
Simplicity Bias via Global Convergence of Sharpness Minimization
https://openreview.net/forum?id=VUTyzH63Xa
[ "Khashayar Gatmiry", "Zhiyuan Li", "Sashank J. Reddi", "Stefanie Jegelka" ]
Poster
null
The remarkable generalization ability of neural networks is usually attributed to the implicit bias of SGD, which often yields models with lower complexity using simpler (e.g. linear) and low-rank features. Recent works have provided empirical and theoretical evidence for the bias of particular variants of SGD (such as...
[]
null
10,069
2410.16401
title_snapshot
[ -0.062380146235227585, 0.002589419251307845, 0.007344633340835571, 0.06867723912000656, -0.008409091271460056, 0.03479345142841339, 0.036289770156145096, 0.0221363864839077, -0.040586356073617935, -0.0414770245552063, 0.008022251538932323, -0.007097999565303326, -0.07868686318397522, 0.026...
STEER: Assessing the Economic Rationality of Large Language Models
https://openreview.net/forum?id=nU1mtFDtMX
[ "Narun Krishnamurthi Raman", "Taylor Lundy", "Samuel Joseph Amouyal", "Yoav Levine", "Kevin Leyton-Brown", "Moshe Tennenholtz" ]
Poster
null
There is increasing interest in using LLMs as decision-making "agents". Doing so includes many degrees of freedom: which model should be used; how should it be prompted; should it be asked to introspect, conduct chain-of-thought reasoning, etc? Settling these questions---and more broadly, determining whether an LLM age...
[]
null
10,056
2402.09552
title_snapshot
[ -0.042591776698827744, -0.010495811700820923, -0.012651089578866959, 0.04558373615145683, 0.06883613765239716, 0.012824281118810177, 0.006660154089331627, 0.050096433609724045, -0.025917960330843925, -0.0029169500339776278, -0.02152855508029461, 0.06749099493026733, -0.06450192630290985, -...
Beyond Chinchilla-Optimal: Accounting for Inference in Language Model Scaling Laws
https://openreview.net/forum?id=0bmXrtTDUu
[ "Nikhil Sardana", "Jacob Portes", "Sasha Doubov", "Jonathan Frankle" ]
Poster
null
Large language model (LLM) scaling laws are empirical formulas that estimate changes in model quality as a result of increasing parameter count and training data. However, these formulas, including the popular Deepmind Chinchilla scaling laws, neglect to include the cost of inference. We modify the Chinchilla scaling l...
[]
null
10,055
2401.00448
title_snapshot
[ -0.03193217143416405, -0.018737394362688065, -0.014423975721001625, 0.008589398115873337, 0.05357302352786064, 0.024253614246845245, 0.04677556827664375, 0.025215202942490578, -0.0350518561899662, -0.0035045924596488476, 0.002205109689384699, 0.020901652052998543, -0.06489413231611252, 0.0...
One Prompt is not Enough: Automated Construction of a Mixture-of-Expert Prompts
https://openreview.net/forum?id=edHLN40DWu
[ "Ruochen Wang", "Sohyun An", "Minhao Cheng", "Tianyi Zhou", "Sung Ju Hwang", "Cho-Jui Hsieh" ]
Poster
null
Large Language Models (LLMs) exhibit strong generalization capabilities to novel tasks when prompted with language instructions and in-context demos. Since this ability sensitively depends on the quality of prompts, various methods have been explored to automate the instruction design. While these methods demonstrated ...
[]
null
10,053
2407.00256
title_snapshot
[ -0.013711515814065933, -0.03630524501204491, -0.03465261682868004, 0.06070379540324211, 0.0372529998421669, 0.00930940080434084, 0.019007619470357895, 0.005379662849009037, -0.013977267779409885, 0.001281630015000701, -0.06249111145734787, 0.06329495459794998, -0.04716292396187782, -0.0340...
Double-Step Alternating Extragradient with Increasing Timescale Separation for Finding Local Minimax Points: Provable Improvements
https://openreview.net/forum?id=nUVForc3VP
[ "Kyuwon Kim", "Donghwan Kim" ]
Poster
null
In nonconvex-nonconcave minimax optimization, two-timescale gradient methods have shown their potential to find local minimax (optimal) points, provided that the timescale separation between the min and the max player is sufficiently large. However, existing two-timescale variants of gradient descent ascent and extragr...
[]
null
10,043
null
null
[ -0.04691266641020775, -0.026313520967960358, 0.03719548508524895, 0.028520731255412102, 0.02974729798734188, 0.03913210332393646, 0.023744387552142143, -0.03187986835837364, -0.029760226607322693, -0.0463554672896862, 0.007259836420416832, -0.019185304641723633, -0.05109553784132004, 0.010...
Unveiling and Harnessing Hidden Attention Sinks: Enhancing Large Language Models without Training through Attention Calibration
https://openreview.net/forum?id=DLTjFFiuUJ
[ "Zhongzhi Yu", "Zheng Wang", "Yonggan Fu", "Huihong Shi", "Khalid Shaikh", "Yingyan Celine Lin" ]
Poster
null
Attention is a fundamental component behind the remarkable achievements of large language models (LLMs). However, our current understanding of the attention mechanism, especially regarding how attention distributions are established, remains limited. Inspired by recent studies that explore the presence of attention sin...
[]
null
10,042
2406.15765
title_snapshot
[ 0.007713523227721453, -0.020012367516756058, 0.010455931536853313, 0.022905632853507996, 0.03791651502251625, 0.024335850030183792, 0.025615962222218513, 0.034663472324609756, -0.026683485135436058, 0.008369328454136848, -0.029005691409111023, 0.014414186589419842, -0.04245411232113838, 0....
TENG: Time-Evolving Natural Gradient for Solving PDEs With Deep Neural Nets Toward Machine Precision
https://openreview.net/forum?id=v1I4zRAjMb
[ "Zhuo Chen", "Jacob McCarran", "Esteban Vizcaino", "Marin Soljacic", "Di Luo" ]
Poster
null
Partial differential equations (PDEs) are instrumental for modeling dynamical systems in science and engineering. The advent of neural networks has initiated a significant shift in tackling these complexities though challenges in accuracy persist, especially for initial value problems. In this paper, we introduce the *...
[]
null
10,038
2404.10771
title_snapshot
[ -0.03554004803299904, -0.005881422199308872, 0.018062584102153778, 0.04289912059903145, 0.028361797332763672, 0.06584601104259491, 0.026057830080389977, 0.027505667880177498, -0.04962391406297684, -0.05811860412359238, 0.008162430487573147, -0.011993437074124813, -0.05506594479084015, 0.03...
Learning Adaptive and View-Invariant Vision Transformer for Real-Time UAV Tracking
https://openreview.net/forum?id=eaNLvrP8n1
[ "Yongxin Li", "Mengyuan Liu", "You Wu", "Xucheng Wang", "Xiangyang Yang", "Shuiwang Li" ]
Poster
null
Harnessing transformer-based models, visual tracking has made substantial strides. However, the sluggish performance of current trackers limits their practicality on devices with constrained computational capabilities, especially for real-time unmanned aerial vehicle (UAV) tracking. Addressing this challenge, we introd...
[]
null
10,036
2412.20002
title_judge
[ 0.026548197492957115, -0.006340066436678171, 0.01925099641084671, 0.02072303369641304, 0.023611394688487053, 0.03164118900895119, 0.043837834149599075, 0.016983283683657646, -0.03335479646921158, -0.05083828791975975, -0.049647632986307144, 0.0001485620014136657, -0.07284252345561981, -0.0...
Conditional Common Entropy for Instrumental Variable Testing and Partial Identification
https://openreview.net/forum?id=Wnni3cu39x
[ "Ziwei Jiang", "Murat Kocaoglu" ]
Poster
null
Instrumental variables (IVs) are widely used for estimating causal effects. There are two main challenges when using instrumental variables. First of all, using IV without additional assumptions such as linearity, the causal effect may still not be identifiable. Second, when selecting an IV, the validity of the selecte...
[]
null
10,033
null
null
[ -0.011853410862386227, -0.007178619038313627, -0.045488763600587845, 0.030593199655413628, 0.024948541074991226, 0.056916363537311554, 0.05286942049860954, 0.00306136067956686, -0.038805361837148666, -0.03879941254854202, -0.014418790116906166, 0.010666734538972378, -0.06348808109760284, -...
Towards Robust Model-Based Reinforcement Learning Against Adversarial Corruption
https://openreview.net/forum?id=Z0S6fUdW68
[ "Chenlu Ye", "Jiafan He", "Quanquan Gu", "Tong Zhang" ]
Poster
null
This study tackles the challenges of adversarial corruption in model-based reinforcement learning (RL), where the transition dynamics can be corrupted by an adversary. Existing studies on corruption-robust RL mostly focus on the setting of model-free RL, where robust least-square regression is often employed for value ...
[]
null
10,022
2402.08991
title_snapshot
[ -0.03441418334841728, -0.01718471758067608, -0.015591415576636791, 0.053920648992061615, 0.03333878144621849, 0.006103238556534052, 0.01490907184779644, -0.012812884524464607, -0.042607810348272324, -0.028030410408973694, -0.006416202988475561, 0.016245901584625244, -0.07725732028484344, -...
Prompting is a Double-Edged Sword: Improving Worst-Group Robustness of Foundation Models
https://openreview.net/forum?id=fdroxYsgzQ
[ "Amrith Setlur", "Saurabh Garg", "Virginia Smith", "Sergey Levine" ]
Poster
null
Machine learning models fail catastrophically under distribution shift, but a surprisingly effective way to empirically improve robustness to some types of shift (*e.g.*, Imagenet-A/C) is to use stronger open-vocabulary classifiers derived from foundation models. In this work, we first note that for shifts governed by ...
[]
null
10,017
null
null
[ -0.004071144852787256, -0.048305317759513855, 0.007271439768373966, 0.04710351303219795, 0.03188061714172363, 0.010485121980309486, 0.04644785448908806, 0.006108514498919249, -0.02449641190469265, -0.03426024317741394, -0.007733191829174757, 0.03032948076725006, -0.06701118499040604, 0.005...
Precise Accuracy / Robustness Tradeoffs in Regression: Case of General Norms
https://openreview.net/forum?id=btYeH65fI3
[ "Elvis Dohmatob", "Meyer Scetbon" ]
Poster
null
In this paper, we investigate the impact of test-time adversarial attacks on linear regression models and determine the optimal level of robustness that any model can reach while maintaining a given level of standard predictive performance (accuracy). Through quantitative estimates, we uncover fundamental tradeoffs bet...
[]
null
10,007
2308.00556
title_judge
[ -0.00603918032720685, -0.015508118085563183, -0.011741730384528637, 0.03262322023510933, 0.04915144294500351, 0.01473439484834671, 0.0638059750199318, -0.029031114652752876, -0.017358284443616867, -0.02779431827366352, -0.015518390573561192, 0.013249870389699936, -0.08559378981590271, 0.00...
Hard Tasks First: Multi-Task Reinforcement Learning Through Task Scheduling
https://openreview.net/forum?id=haUOhXo70o
[ "Myungsik Cho", "Jongeui Park", "Suyoung Lee", "Youngchul Sung" ]
Poster
null
Multi-task reinforcement learning (RL) faces the significant challenge of varying task difficulties, often leading to negative transfer when simpler tasks overshadow the learning of more complex ones. To overcome this challenge, we propose a novel algorithm, Scheduled Multi-Task Training (SMT), that strategically prior...
[]
null
9,999
null
null
[ -0.032566603273153305, -0.03698175773024559, -0.013058277778327465, 0.04441862925887108, 0.03707704693078995, 0.013855540193617344, 0.0021044283639639616, -0.005525586660951376, -0.037383340299129486, -0.03189968317747116, -0.039909642189741135, 0.03817589208483696, -0.07482622563838959, -...
Fault Tolerant ML: Efficient Meta-Aggregation and Synchronous Training
https://openreview.net/forum?id=Ht20wtgaty
[ "Tehila Dahan", "Kfir Yehuda Levy" ]
Poster
null
In this paper, we investigate the challenging framework of Byzantine-robust training in distributed machine learning (ML) systems, focusing on enhancing both efficiency and practicality. As distributed ML systems become integral for complex ML tasks, ensuring resilience against Byzantine failures—where workers may cont...
[]
null
9,997
2405.14759
title_snapshot
[ -0.005187807604670525, -0.029613759368658066, -0.013120125979185104, 0.0430404432117939, 0.0457611158490181, 0.00820828601717949, 0.04758426547050476, -0.0035909705329686403, -0.021186962723731995, -0.026001138612627983, -0.004142756573855877, -0.0042732791043818, -0.07359082251787186, 0.0...
Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations
https://openreview.net/forum?id=xye7iNsgXn
[ "Jiaqi Zhai", "Lucy Liao", "Xing Liu", "Yueming Wang", "Rui Li", "Xuan Cao", "Leon Gao", "Zhaojie Gong", "Fangda Gu", "Jiayuan He", "Yinghai Lu", "Yu Shi" ]
Poster
null
Large-scale recommendation systems are characterized by their reliance on high cardinality, heterogeneous features and the need to handle tens of billions of user actions on a daily basis. Despite being trained on huge volume of data with thousands of features, most Deep Learning Recommendation Models (DLRMs) in indust...
[]
null
9,994
2402.17152
title_snapshot
[ 0.028072509914636612, -0.05344441533088684, -0.013647624291479588, 0.02959250472486019, 0.01792205311357975, 0.03210878372192383, 0.018154185265302658, 0.02887292578816414, 0.0066887494176626205, -0.03364007547497749, -0.013183634728193283, -0.002417474752292037, -0.043985623866319656, 0.0...
Trustworthy Alignment of Retrieval-Augmented Large Language Models via Reinforcement Learning
https://openreview.net/forum?id=XwnABAdH5y
[ "Zongmeng Zhang", "Yufeng Shi", "Jinhua Zhu", "Wengang Zhou", "Xiang Qi", "peng zhang", "Houqiang Li" ]
Poster
null
Trustworthiness is an essential prerequisite for the real-world application of large language models. In this paper, we focus on the trustworthiness of language models with respect to retrieval augmentation. Despite being supported with external evidence, retrieval-augmented generation still suffers from hallucinations...
[]
null
9,991
2410.16843
title_snapshot
[ -0.023222846910357475, 0.0029863326344639063, -0.025211455300450325, 0.06912022829055786, 0.023442355915904045, 0.019994331523776054, 0.0431990772485733, 0.03162797540426254, -0.019979946315288544, -0.02044740691781044, -0.04798164591193199, 0.0569852739572525, -0.07705529779195786, -0.023...
Instruction Tuning for Secure Code Generation
https://openreview.net/forum?id=MgTzMaYHvG
[ "Jingxuan He", "Mark Vero", "Gabriela Krasnopolska", "Martin Vechev" ]
Poster
null
Modern language models (LMs) have gained widespread acceptance in everyday and professional contexts, particularly in programming. An essential procedure enabling this adoption is instruction tuning, which substantially enhances LMs' practical utility by training them to follow user instructions and human preferences. ...
[]
null
9,980
2402.09497
title_snapshot
[ -0.020007839426398277, -0.007003476843237877, -0.026297274976968765, 0.052264176309108734, 0.07108110189437866, 0.02293262630701065, 0.05713900923728943, -0.013107157312333584, -0.0060287462547421455, -0.03142141178250313, -0.033224571496248245, 0.01983814686536789, -0.05877820402383804, -...
Don't be so Negative! Score-based Generative Modeling with Oracle-assisted Guidance
https://openreview.net/forum?id=H8pMSJwRD5
[ "Saeid Naderiparizi", "Xiaoxuan Liang", "Setareh Cohan", "Berend Zwartsenberg", "Frank Wood" ]
Poster
null
Score-based diffusion models are a powerful class of generative models, widely utilized across diverse domains. Despite significant advancements in large-scale tasks such as text-to-image generation, their application to constrained domains has received considerably less attention. This work addresses model learning in...
[]
null
9,973
2307.16463
title_snapshot
[ -0.008836681954562664, -0.005117095075547695, 0.0057445187121629715, 0.042048901319503784, 0.05336330831050873, 0.026117263361811638, -0.005031020846217871, -0.04142334312200546, -0.0019957078620791435, -0.04996807500720024, -0.0073022497817873955, 0.017184164375066757, -0.05858208239078522,...
LangCell: Language-Cell Pre-training for Cell Identity Understanding
https://openreview.net/forum?id=GcZjpKA37R
[ "Suyuan Zhao", "Jiahuan Zhang", "Yushuai Wu", "YIZHEN LUO", "Zaiqing Nie" ]
Poster
null
Cell identity encompasses various semantic aspects of a cell, including cell type, pathway information, disease information, and more, which are essential for biologists to gain insights into its biological characteristics. Understanding cell identity from the transcriptomic data, such as annotating cell types, has bec...
[]
null
9,961
2405.06708
title_snapshot
[ -0.0116935009136796, -0.010997581295669079, -0.025364646688103676, 0.02115515246987343, 0.054514478892087936, 0.026075340807437897, 0.007304630242288113, 0.0038502190727740526, -0.04735792055726051, 0.0025939131155610085, -0.012948649935424328, 0.003859015181660652, -0.05918040871620178, 0...
Causal Representation Learning from Multiple Distributions: A General Setting
https://openreview.net/forum?id=Pte6iiXvpf
[ "Kun Zhang", "Shaoan Xie", "Ignavier Ng", "Yujia Zheng" ]
Poster
null
In many problems, the measured variables (e.g., image pixels) are just mathematical functions of the latent causal variables (e.g., the underlying concepts or objects). For the purpose of making predictions in changing environments or making proper changes to the system, it is helpful to recover the latent causal varia...
[]
null
9,954
2402.05052
title_snapshot
[ -0.014989711344242096, -0.02127048932015896, -0.01239000353962183, 0.04179493710398674, 0.03622274845838547, 0.05956229940056801, 0.02247254177927971, 0.014131080359220505, -0.013423633761703968, -0.03890008479356766, -0.004671166185289621, 0.004017102066427469, -0.06585310399532318, 0.009...
Diffusion Models Demand Contrastive Guidance for Adversarial Purification to Advance
https://openreview.net/forum?id=2NUGeV64y2
[ "Mingyuan Bai", "Wei Huang", "Tenghui Li", "Andong Wang", "Junbin Gao", "Cesar F Caiafa", "Qibin Zhao" ]
Poster
null
In adversarial defense, adversarial purification can be viewed as a special generation task with the purpose to remove adversarial attacks and diffusion models excel in adversarial purification for their strong generative power. With different predetermined generation requirements, various types of guidance have been p...
[]
null
9,947
null
null
[ 0.0032346651423722506, -0.008138296194374561, 0.005478430539369583, 0.038737840950489044, 0.04505223408341408, 0.006870422977954149, 0.018528694286942482, -0.037876881659030914, -0.01469921413809061, -0.02290264703333378, -0.020999496802687645, -0.01088635716587305, -0.05415656790137291, 0...
PairNet: Training with Observed Pairs to Estimate Individual Treatment Effect
https://openreview.net/forum?id=o5SVr80Rgg
[ "Lokesh Nagalapatti", "Pranava Singhal", "Avishek Ghosh", "Sunita Sarawagi" ]
Poster
null
Given a dataset of individuals each described by a covariate vector, a treatment, and an observed outcome on the treatment, the goal of the individual treatment effect (ITE) estimation task is to predict outcome changes resulting from a change in treatment. A fundamental challenge is that in the observational data, a c...
[]
null
9,945
2406.03864
title_snapshot
[ 0.028581256046891212, -0.027126891538500786, -0.028492532670497894, 0.030230935662984848, 0.0026574593503028154, 0.031313829123973846, 0.05849411338567734, -0.001491404720582068, 0.003560505574569106, -0.045806579291820526, 0.004072277806699276, 0.009333081543445587, -0.048218436539173126, ...
Position: Open-Endedness is Essential for Artificial Superhuman Intelligence
https://openreview.net/forum?id=Bc4vZ2CX7E
[ "Edward Hughes", "Michael D Dennis", "Jack Parker-Holder", "Feryal Behbahani", "Aditi Mavalankar", "Yuge Shi", "Tom Schaul", "Tim Rocktäschel" ]
Oral
null
In recent years there has been a tremendous surge in the general capabilities of AI systems, mainly fuelled by training foundation models on internet-scale data. Nevertheless, the creation of open-ended, ever self-improving AI remains elusive. **In this position paper, we argue that the ingredients are now in place to ...
[]
null
9,943
2406.04268
title_judge
[ -0.029170695692300797, -0.014926622621715069, -0.03471316769719124, 0.025196991860866547, 0.045899223536252975, 0.008863951079547405, 0.03466225042939186, 0.011410282924771309, -0.0263837780803442, -0.04218512400984764, -0.033298309892416, 0.006625360809266567, -0.06598366051912308, -0.029...
Simple linear attention language models balance the recall-throughput tradeoff
https://openreview.net/forum?id=e93ffDcpH3
[ "Simran Arora", "Sabri Eyuboglu", "Michael Zhang", "Aman Timalsina", "Silas Alberti", "James Zou", "Atri Rudra", "Christopher Re" ]
Spotlight
null
Recent work has shown that attention-based language models excel at "recall", the ability to ground generations in tokens previously seen in context. However, the efficiency of attention-based models is bottle-necked during inference by the KV-cache's aggressive memory consumption. In this work, we explore whether we c...
[]
null
9,942
2402.18668
title_snapshot
[ -0.025929033756256104, 0.0006206304533407092, 0.008410836569964886, 0.028004450723528862, 0.002182665513828397, 0.028652934357523918, 0.04165520891547203, 0.03596339374780655, -0.03276636451482773, -0.02343447506427765, 0.0017745193326845765, 0.017468253150582314, -0.07123729586601257, 0.0...
Meta-Learners for Partially-Identified Treatment Effects Across Multiple Environments
https://openreview.net/forum?id=s5PLISyNyP
[ "Jonas Schweisthal", "Dennis Frauen", "Mihaela van der Schaar", "Stefan Feuerriegel" ]
Poster
null
Estimating the conditional average treatment effect (CATE) from observational data is relevant for many applications such as personalized medicine. Here, we focus on the widespread setting where the observational data come from multiple environments, such as different hospitals, physicians, or countries. Furthermore, w...
[]
null
9,939
2406.02464
title_snapshot
[ 0.0010668353643268347, -0.009745979681611061, -0.01524930540472269, 0.018325520679354668, 0.04519805312156677, 0.018396452069282532, 0.06863696873188019, -0.015825923532247543, -0.0200221985578537, -0.01869141310453415, -0.0035352990962564945, 0.013584780506789684, -0.054631803184747696, -...
Principled Gradient-Based MCMC for Conditional Sampling of Text
https://openreview.net/forum?id=AwLLSlJAeJ
[ "Li Du", "Afra Amini", "Lucas Torroba Hennigen", "Xinyan Velocity Yu", "Holden Lee", "Jason Eisner", "Ryan Cotterell" ]
Poster
null
We consider the problem of sampling text from an energy-based model. This arises, for example, when sampling text from a neural language model subject to soft constraints. Although the target distribution is discrete, the internal computations of the energy function (given by the language model) are differentiable, so ...
[]
null
9,938
null
null
[ -0.0284504983574152, -0.014825551770627499, -0.0007447762181982398, 0.07048855721950531, 0.05038721114397049, 0.04044965282082558, 0.012601104564964771, 0.0012357301311567426, -0.031061027199029922, -0.013456365093588829, -0.009475958533585072, 0.03476627543568611, -0.0718822330236435, -0....
Maestro: Uncovering Low-Rank Structures via Trainable Decomposition
https://openreview.net/forum?id=7bjyambg4x
[ "Samuel Horváth", "Stefanos Laskaridis", "Shashank Rajput", "Hongyi Wang" ]
Poster
null
Deep Neural Networks (DNNs) have been a large driver for AI breakthroughs in recent years, ranging from self-driving cars to intelligent assistants. However, these models have been getting increasingly large as they become more accurate and safe. This means that their training becomes increasingly costly and time-consu...
[]
null
9,930
2308.14929
title_snapshot
[ -0.019652677699923515, -0.05518834665417671, -0.0023144043516367674, 0.04634597525000572, 0.035037893801927567, 0.021301116794347763, 0.008848635479807854, -0.016215726733207703, -0.0331694521009922, -0.046471789479255676, -0.013073176145553589, 0.010321428999304771, -0.04027324542403221, ...
Infinite-Horizon Distributionally Robust Regret-Optimal Control
https://openreview.net/forum?id=h3SGdpI4Ta
[ "Taylan Kargin", "Joudi Hajar", "Vikrant Malik", "Babak Hassibi" ]
Poster
null
We study the infinite-horizon distributionally robust (DR) control of linear systems with quadratic costs, where disturbances have unknown, possibly time-correlated distribution within a Wasserstein-2 ambiguity set. We aim to minimize the worst-case expected regret—the excess cost of a causal policy compared to a non-c...
[]
null
9,927
2406.07248
title_snapshot
[ -0.0537421852350235, 0.017216386273503304, -0.0029613745864480734, 0.03798194229602814, 0.0623527392745018, 0.014821063727140427, -0.0013653065543621778, 0.0096795205026865, 0.0017347343964502215, -0.06140845641493797, -0.03133905306458473, -0.008325954899191856, -0.05346599593758583, 0.01...
Conformal Validity Guarantees Exist for Any Data Distribution (and How to Find Them)
https://openreview.net/forum?id=F3936hVwQa
[ "Drew Prinster", "Samuel Don Stanton", "Anqi Liu", "Suchi Saria" ]
Poster
null
As artificial intelligence (AI) / machine learning (ML) gain widespread adoption, practitioners are increasingly seeking means to quantify and control the risk these systems incur. This challenge is especially salient when such systems have autonomy to collect their own data, such as in black-box optimization and activ...
[]
null
9,926
2405.06627
title_snapshot
[ -0.02280748076736927, -0.021811706945300102, -0.008442307822406292, 0.03917564079165459, 0.061646074056625366, 0.0044410377740859985, 0.018465379253029823, -0.013558774255216122, -0.00951225496828556, -0.042031001299619675, -0.013765501789748669, 0.004098162520676851, -0.09112510830163956, ...
StrWAEs to Invariant Representations
https://openreview.net/forum?id=kLZZWvqlEm
[ "Hyunjong Lee", "Yedarm Seong", "Sungdong Lee", "Joong-Ho Won" ]
Poster
null
Autoencoders have become an indispensable tool for generative modeling and representation learning in high dimensions. Imposing structural constraints such as conditional independence in order to capture invariance of latent variables to nuisance information has been attempted through adding *ad hoc* penalties to the l...
[]
null
9,920
null
null
[ -0.004574158228933811, 0.00394876953214407, -0.012110273353755474, 0.030540114268660545, 0.018858376890420914, 0.05687488988041878, 0.06869138032197952, -0.010225470177829266, -0.020817313343286514, -0.03537262603640556, -0.029199961572885513, -0.009540251456201077, -0.059922125190496445, ...
Debiased Offline Representation Learning for Fast Online Adaptation in Non-stationary Dynamics
https://openreview.net/forum?id=BrZPj9rEpN
[ "Xinyu Zhang", "Wenjie Qiu", "Yi-Chen Li", "Lei Yuan", "Chengxing Jia", "Zongzhang Zhang", "Yang Yu" ]
Poster
null
Developing policies that can adapt to non-stationary environments is essential for real-world reinforcement learning applications. Nevertheless, learning such adaptable policies in offline settings, with only a limited set of pre-collected trajectories, presents significant challenges. A key difficulty arises because t...
[]
null
9,919
2402.11317
title_snapshot
[ -0.03859991207718849, -0.028301626443862915, -0.007159136235713959, 0.04060259833931923, 0.02811097539961338, 0.02252095192670822, 0.016787804663181305, -0.0013483415823429823, -0.041421424597501755, -0.0270848385989666, -0.0247973520308733, -0.0018108125077560544, -0.08292555063962936, -0...
Linguistic Calibration of Long-Form Generations
https://openreview.net/forum?id=rJVjQSQ8ye
[ "Neil Band", "Xuechen Li", "Tengyu Ma", "Tatsunori Hashimoto" ]
Poster
null
Language models (LMs) may lead their users to make suboptimal downstream decisions when they confidently hallucinate. This issue can be mitigated by having the LM verbally convey the probability that its claims are correct, but existing models cannot produce long-form text with calibrated confidence statements. Through...
[]
null
9,911
2404.00474
title_snapshot
[ -0.011373719200491905, 0.004577461630105972, -0.01991705782711506, -0.016926515847444534, 0.059059228748083115, 0.056496214121580124, 0.03242390230298042, 0.04264034703373909, -0.03571762144565582, -0.02621539495885372, -0.02286546118557453, 0.04283388704061508, -0.05447780340909958, -0.00...
R2E: Turning any Github Repository into a Programming Agent Environment
https://openreview.net/forum?id=kXHgEYFyf3
[ "Naman Jain", "Manish Shetty", "Tianjun Zhang", "King Han", "Koushik Sen", "Ion Stoica" ]
Poster
null
While Large Language Models’ (LLMs) coding capabilities have advanced rapidly, corresponding evaluation benchmarks on real-world programming setups are yet to catch up. Building a scalable and interactive testbed for evaluating general-purpose AI coding agents for real-world code has been challenging, particularly due ...
[]
null
9,908
null
null
[ 0.011233076453208923, -0.03641493618488312, -0.03961693495512009, 0.03252451866865158, 0.051375843584537506, 0.03791718930006027, 0.03833327442407608, -0.00015373087080661207, -0.02963436208665371, -0.02841702662408352, -0.03362980857491493, 0.019489971920847893, -0.08332617580890656, -0.0...
Learning to Explore for Stochastic Gradient MCMC
https://openreview.net/forum?id=aECamk9izk
[ "SeungHyun Kim", "Seohyeon Jung", "SeongHyeon Kim", "Juho Lee" ]
Poster
null
Bayesian Neural Networks(BNNs) with high-dimensional parameters pose a challenge for posterior inference due to the multi-modality of the posterior distributions. Stochastic Gradient Markov Chain Monte Carlo(SGMCMC) with cyclical learning rate scheduling is a promising solution, but it requires a large number of sampli...
[]
null
9,897
2408.09140
title_snapshot
[ -0.015533422119915485, -0.004147551953792572, -0.014889592304825783, 0.060311317443847656, 0.03415513038635254, 0.02896459773182869, 0.03202466666698456, 0.000047990317398216575, -0.03711917996406555, -0.032039448618888855, 0.006682619918137789, 0.016428858041763306, -0.040757447481155396, ...
Efficient Adaptation in Mixed-Motive Environments via Hierarchical Opponent Modeling and Planning
https://openreview.net/forum?id=disVlUOH4b
[ "Yizhe Huang", "Anji Liu", "Fanqi Kong", "Yaodong Yang", "Song-Chun Zhu", "Xue Feng" ]
Poster
null
Despite the recent successes of multi-agent reinforcement learning (MARL) algorithms, efficiently adapting to co-players in mixed-motive environments remains a significant challenge. One feasible approach is to hierarchically model co-players' behavior based on inferring their characteristics. However, these methods of...
[]
null
9,889
2406.08002
title_snapshot
[ -0.05167948827147484, -0.010604343377053738, -0.0033640884794294834, 0.01066631730645895, 0.04620135575532913, 0.01570264622569084, 0.02922576293349266, -0.020862864330410957, -0.03891979157924652, -0.0399324893951416, 0.008478894829750061, 0.025181496515870094, -0.04543672502040863, -0.02...
How to Explore with Belief: State Entropy Maximization in POMDPs
https://openreview.net/forum?id=LbcNAIgNnB
[ "Riccardo Zamboni", "Duilio Cirino", "Marcello Restelli", "Mirco Mutti" ]
Poster
null
Recent works have studied *state entropy maximization* in reinforcement learning, in which the agent's objective is to learn a policy inducing high entropy over states visitation (Hazan et al., 2019). They typically assume full observability of the state of the system, so that the entropy of the observations is maximiz...
[]
null
9,887
2406.02295
title_snapshot
[ -0.04012569785118103, -0.007521188817918301, -0.009022234007716179, 0.05503242835402489, 0.05291104316711426, 0.03388034552335739, 0.029169712215662003, -0.014867586083710194, -0.03231124207377434, -0.04831227287650108, -0.028397422283887863, 0.0017589041963219643, -0.0732639953494072, -0....
Submodular framework for structured-sparse optimal transport
https://openreview.net/forum?id=bfQCO9Vqhk
[ "Piyushi Manupriya", "Pratik Jawanpuria", "Karthik S. Gurumoorthy", "SakethaNath Jagarlapudi", "Bamdev Mishra" ]
Poster
null
Unbalanced optimal transport (UOT) has recently gained much attention due to its flexible framework for handling un-normalized measures and its robustness properties. In this work, we explore learning (structured) sparse transport plans in the UOT setting, i.e., transport plans have an upper bound on the number of non-...
[]
null
9,879
2406.04914
title_snapshot
[ -0.009339362382888794, -0.05768953636288643, 0.01801074668765068, 0.03598421812057495, 0.05063281208276749, 0.01861436851322651, -0.006584042217582464, -0.01422109268605709, -0.016031529754400253, -0.07694216817617416, 0.03452669829130173, -0.016333438456058502, -0.07104005664587021, -0.00...
Graph-Triggered Rising Bandits
https://openreview.net/forum?id=bPsohGR6gD
[ "Gianmarco Genalti", "Marco Mussi", "Nicola Gatti", "Marcello Restelli", "Matteo Castiglioni", "Alberto Maria Metelli" ]
Poster
null
In this paper, we propose a novel generalization of rested and restless bandits where the evolution of the arms' expected rewards is governed by a graph defined over the arms. An edge connecting a pair of arms $(i,j)$ represents the fact that a pull of arm $i$ *triggers* the evolution of arm $j$, and vice versa. Intere...
[]
null
9,878
null
null
[ -0.03725507855415344, -0.027960680425167084, 0.01205617468804121, 0.03919088467955589, 0.03770681098103523, 0.011324744671583176, 0.02841801382601261, 0.01764524169266224, -0.025258904322981834, -0.0457184873521328, -0.011542859487235546, 0.008806473575532436, -0.0690421611070633, -0.02777...
Breadth-First Exploration on Adaptive Grid for Reinforcement Learning
https://openreview.net/forum?id=59MYoLghyk
[ "Youngsik Yoon", "Gangbok Lee", "Sungsoo Ahn", "Jungseul Ok" ]
Poster
null
Graph-based planners have gained significant attention for goal-conditioned reinforcement learning (RL), where they construct a graph consisting of confident transitions between *subgoals* as edges and run shortest path algorithms to exploit the confident edges. Meanwhile, identifying and avoiding unattainable transiti...
[]
null
9,877
null
null
[ -0.01759490743279457, -0.03602644056081772, 0.012810011394321918, 0.017459694296121597, 0.054408129304647446, 0.0029259431175887585, 0.015885522589087486, -0.007033071480691433, -0.03279584273695946, -0.05078411102294922, -0.00435372581705451, 0.006698639132082462, -0.06356094032526016, -0...
SILVER: Single-loop variance reduction and application to federated learning
https://openreview.net/forum?id=pOgMluzEIH
[ "Kazusato Oko", "Shunta Akiyama", "Denny Wu", "Tomoya Murata", "Taiji Suzuki" ]
Poster
null
Most variance reduction methods require multiple times of full gradient computation, which is time-consuming and hence a bottleneck in application to distributed optimization. We present a single-loop variance-reduced gradient estimator named SILVER (SIngle-Loop VariancE-Reduction) for the finite-sum non-convex optimiz...
[]
null
9,875
null
null
[ 0.00502160657197237, -0.021062128245830536, 0.01628504879772663, 0.03277195990085602, 0.02256838232278824, 0.035631049424409866, 0.043879877775907516, -0.016779376193881035, -0.021761232987046242, -0.04550690948963165, 0.014897345565259457, -0.03239871934056282, -0.06048671528697014, 0.027...
Generative Conditional Distributions by Neural (Entropic) Optimal Transport
https://openreview.net/forum?id=FoRqdsN4IA
[ "Bao Nguyen", "Binh Nguyen", "Hieu Trung Nguyen", "Viet Anh Nguyen" ]
Poster
null
Learning conditional distributions is challenging because the desired outcome is not a single distribution but multiple distributions that correspond to multiple instances of the covariates. We introduce a novel neural entropic optimal transport method designed to effectively learn generative models of conditional dist...
[]
null
9,874
2406.02317
title_snapshot
[ -0.043827082961797714, -0.026134956628084183, -0.007612141780555248, 0.0670539140701294, 0.05334656313061714, 0.051668401807546616, 0.0016678510000929236, -0.028931325301527977, -0.01286950334906578, -0.0678972452878952, -0.013367432169616222, 0.011122867465019226, -0.05781015753746033, 0....
Cell2Sentence: Teaching Large Language Models the Language of Biology
https://openreview.net/forum?id=EWt5wsEdvc
[ "Daniel Levine", "Syed A Rizvi", "Sacha Lévy", "Nazreen Pallikkavaliyaveetil", "David Zhang", "Xingyu Chen", "Sina Ghadermarzi", "Ruiming Wu", "Zihe Zheng", "Ivan Vrkic", "Anna Zhong", "Daphne Raskin", "Insu Han", "Antonio Henrique de Oliveira Fonseca", "Josue Ortega Caro", "Amin Karba...
Poster
null
We introduce Cell2Sentence (C2S), a novel method to directly adapt large language models to a biological context, specifically single-cell transcriptomics. By transforming gene expression data into "cell sentences," C2S bridges the gap between natural language processing and biology. We demonstrate cell sentences enabl...
[]
null
9,873
null
null
[ -0.027380529791116714, -0.04673641175031662, -0.0342317633330822, 0.03837263956665993, 0.040699686855077744, 0.023280242457985878, 0.02368944138288498, 0.0230818260461092, -0.023330416530370712, -0.013022049330174923, -0.007548066321760416, 0.03654500097036362, -0.05967181921005249, 0.0297...
Watermark Stealing in Large Language Models
https://openreview.net/forum?id=Wp054bnPq9
[ "Nikola Jovanović", "Robin Staab", "Martin Vechev" ]
Poster
null
LLM watermarking has attracted attention as a promising way to detect AI-generated content, with some works suggesting that current schemes may already be fit for deployment. In this work we dispute this claim, identifying *watermark stealing* (WS) as a fundamental vulnerability of these schemes. We show that querying ...
[]
null
9,867
2402.19361
title_snapshot
[ 0.010389112867414951, -0.032922256737947464, -0.00976791512221098, 0.05020572245121002, 0.061397820711135864, -0.006392114330083132, 0.031136706471443176, 0.007024089340120554, 0.00048331476864404976, -0.014866969548165798, -0.03363896161317825, -0.0028102390933781862, -0.05594432353973389, ...
A Simple Early Exiting Framework for Accelerated Sampling in Diffusion Models
https://openreview.net/forum?id=OnOaj3g9fi
[ "Taehong Moon", "Moonseok Choi", "EungGu Yun", "Jongmin Yoon", "Gayoung Lee", "Jaewoong Cho", "Juho Lee" ]
Poster
null
Diffusion models have shown remarkable performance in generation problems over various domains including images, videos, text, and audio. A practical bottleneck of diffusion models is their sampling speed, due to the repeated evaluation of score estimation networks during the inference. In this work, we propose a novel...
[]
null
9,866
2408.05927
title_snapshot
[ 0.0034427421633154154, -0.026243366301059723, 0.014195173978805542, 0.05424787476658821, 0.051732975989580154, 0.025306180119514465, 0.007953311316668987, -0.009100493974983692, -0.01666543446481228, -0.06952382624149323, 0.03541279584169388, -0.03131154552102089, -0.036895301192998886, 0....
Stop Regressing: Training Value Functions via Classification for Scalable Deep RL
https://openreview.net/forum?id=dVpFKfqF3R
[ "Jesse Farebrother", "Jordi Orbay", "Quan Vuong", "Adrien Ali Taiga", "Yevgen Chebotar", "Ted Xiao", "Alex Irpan", "Sergey Levine", "Pablo Samuel Castro", "Aleksandra Faust", "Aviral Kumar", "Rishabh Agarwal" ]
Oral
null
Value functions are an essential component in deep reinforcement learning (RL), that are typically trained via mean squared error regression to match bootstrapped target values. However, scaling value-based RL methods to large networks has proven challenging. This difficulty is in stark contrast to supervised learning:...
[]
null
9,864
2403.03950
title_snapshot
[ -0.02744234912097454, -0.04824923351407051, -0.009278697893023491, 0.02133447490632534, 0.05596371367573738, 0.04381976276636124, 0.013994724489748478, -0.01713409088551998, -0.03218047320842743, -0.03454859182238579, -0.012409468181431293, 0.016598684713244438, -0.07079683989286423, 0.002...
Weisfeiler-Leman at the margin: When more expressivity matters
https://openreview.net/forum?id=HTNgNt8CTJ
[ "Billy Joe Franks", "Christopher Morris", "Ameya Velingker", "Floris Geerts" ]
Poster
null
The Weisfeiler--Leman algorithm (1-WL) is a well-studied heuristic for the graph isomorphism problem. Recently, the algorithm has played a prominent role in understanding the expressive power of message-passing graph neural networks (MPNNs) and being effective as a graph kernel. Despite its success, the 1-WL faces chal...
[]
null
9,855
2402.07568
title_snapshot
[ -0.017329327762126923, -0.025570577010512352, 0.023415466770529747, 0.053264912217855453, 0.03326885774731636, 0.013347964733839035, 0.043300725519657135, 0.010750516317784786, 0.004164202604442835, -0.0283369030803442, 0.028121164068579674, -0.012429788708686829, -0.07136403024196625, -0....
Allocation Requires Prediction Only if Inequality Is Low
https://openreview.net/forum?id=WUicA0hOF9
[ "Ali Shirali", "Rediet Abebe", "Moritz Hardt" ]
Spotlight
null
Algorithmic predictions are emerging as a promising solution concept for efficiently allocating societal resources. Fueling their use is an underlying assumption that such systems are necessary to identify individuals for interventions. We propose a principled framework for assessing this assumption: Using a simple mat...
[]
null
9,854
2406.13882
title_snapshot
[ -0.003408373100683093, -0.030525779351592064, -0.02541714534163475, 0.011487896554172039, 0.0570610873401165, 0.041556816548109055, 0.027191001921892166, -0.003852223977446556, -0.017755083739757538, -0.03427307307720184, -0.005867269821465015, -0.016577785834670067, -0.08240967988967896, ...
Decomposing and Editing Predictions by Modeling Model Computation
https://openreview.net/forum?id=rTBR0eqE4G
[ "Harshay Shah", "Andrew Ilyas", "Aleksander Madry" ]
Poster
null
*How does the internal computation of a machine learning model transform inputs into predictions?* To tackle this question, we introduce a framework called *component modeling* for decomposing a model prediction in terms of its components---architectural "building blocks" such as convolution filters or attention heads....
[]
null
9,846
2404.11534
title_snapshot
[ 0.015430647879838943, -0.0024713268503546715, -0.025078605860471725, 0.036926593631505966, 0.048440832644701004, 0.024974709376692772, 0.028632745146751404, 0.007165467832237482, -0.03710078448057175, -0.03454093635082245, -0.013751300051808357, 0.027700157836079597, -0.07786887139081955, ...
Scalable Wasserstein Gradient Flow for Generative Modeling through Unbalanced Optimal Transport
https://openreview.net/forum?id=dMhF96PfQi
[ "Jaemoo Choi", "Jaewoong Choi", "Myungjoo Kang" ]
Poster
null
Wasserstein gradient flow (WGF) describes the gradient dynamics of probability density within the Wasserstein space. WGF provides a promising approach for conducting optimization over the probability distributions. Numerically approximating the continuous WGF requires the time discretization method. The most well-known...
[]
null
9,845
2402.05443
title_snapshot
[ -0.01804957538843155, -0.037630751729011536, 0.05406315624713898, 0.02859138511121273, 0.041975781321525574, 0.02493743970990181, 0.004998582880944014, -0.0027213881257921457, 0.0075403060764074326, -0.06043998524546623, 0.0021482580341398716, -0.03337019681930542, -0.03791167959570885, 0....
Guiding LLMs The Right Way: Fast, Non-Invasive Constrained Generation
https://openreview.net/forum?id=pXaEYzrFae
[ "Luca Beurer-Kellner", "Marc Fischer", "Martin Vechev" ]
Poster
null
To ensure that text generated by large language models (LLMs) is in an expected format, constrained decoding methods propose to enforce strict formal language constraints during generation. However, as we show in this work, not only do such methods often incur performance overhead during generation, but many of them al...
[]
null
9,839
2403.06988
title_snapshot
[ -0.027528023347258568, -0.02665889635682106, -0.02618868462741375, 0.009596982039511204, 0.06548561900854111, 0.027721572667360306, 0.023180972784757614, 0.0125448452308774, -0.03426869958639145, 0.010052182711660862, -0.02838996611535549, 0.03731713443994522, -0.06479692459106445, 0.00034...
Peeking with PEAK: Sequential, Nonparametric Composite Hypothesis Tests for Means of Multiple Data Streams
https://openreview.net/forum?id=hcASxFvmZ5
[ "Brian M Cho", "Kyra Gan", "Nathan Kallus" ]
Poster
null
We propose a novel nonparametric sequential test for composite hypotheses for means of multiple data streams. Our proposed method, peeking with expectation-based averaged capital (PEAK), builds upon the testing-by-betting framework and provides a non-asymptotic $\alpha$-level test across any stopping time. Our contribu...
[]
null
9,838
2402.06122
title_snapshot
[ -0.01943393610417843, -0.03421073034405708, -0.026278214529156685, 0.06421981751918793, 0.04602346941828728, 0.007437325548380613, 0.03598110377788544, -0.009475002065300941, -0.0007313218084163964, -0.05489843338727951, 0.009988720528781414, 0.0008068299503065646, -0.0652383416891098, -0....
DUPLEX: Dual GAT for Complex Embedding of Directed Graphs
https://openreview.net/forum?id=M3uv4qDKOL
[ "Zhaoru Ke", "Hang Yu", "Jianguo Li", "Haipeng Zhang" ]
Poster
null
Current directed graph embedding methods build upon undirected techniques but often inadequately capture directed edge information, leading to challenges such as: (1) Suboptimal representations for nodes with low in/out-degrees, due to the insufficient neighbor interactions; (2) Limited inductive ability for representi...
[]
null
9,834
2406.05391
title_snapshot
[ 0.01904337853193283, -0.026369733735919, -0.014993102289736271, 0.051590852439403534, 0.032888393849134445, 0.03397301211953163, 0.052859753370285034, -0.016384823247790337, 0.009052237495779991, -0.046934641897678375, 0.03129289299249649, -0.009007377550005913, -0.08566360920667648, 0.002...
Exploiting Human-AI Dependence for Learning to Defer
https://openreview.net/forum?id=aiz79FxjaI
[ "Zixi Wei", "Yuzhou Cao", "Lei Feng" ]
Poster
null
The learning to defer (L2D) framework allows models to defer their decisions to human experts. For L2D, the Bayes optimality is the basic requirement of theoretical guarantees for the design of consistent surrogate loss functions, which requires the minimizer (i.e., learned classifier) by the surrogate loss to be the B...
[]
null
9,827
null
null
[ -0.016671502962708473, -0.024142080917954445, 0.006819101981818676, 0.0033059034030884504, 0.04106464982032776, 0.03479628637433052, 0.02793041616678238, 0.0010403175838291645, -0.006124014966189861, -0.03242073208093643, -0.00003621701762313023, 0.040396805852651596, -0.03686050698161125, ...
Iterative Search Attribution for Deep Neural Networks
https://openreview.net/forum?id=5ToHnqYxjB
[ "Zhiyu Zhu", "Huaming Chen", "Xinyi Wang", "Jiayu Zhang", "Zhibo Jin", "Jason Xue", "Jun Shen" ]
Poster
null
Deep neural networks (DNNs) have achieved state-of-the-art performance across various applications. However, ensuring the reliability and trustworthiness of DNNs requires enhanced interpretability of model inputs and outputs. As an effective means of Explainable Artificial Intelligence (XAI) research, the interpretabil...
[]
null
9,818
null
null
[ -0.02491132728755474, -0.000466563185909763, -0.0420590378344059, 0.07219891995191574, 0.03217998147010803, 0.06090665981173515, 0.006350788287818432, 0.012832420878112316, 0.005408134777098894, -0.04690520837903023, -0.03324439749121666, -0.032002899795770645, -0.04122769832611084, 0.0007...
Latent Logic Tree Extraction for Event Sequence Explanation from LLMs
https://openreview.net/forum?id=pwfcwEqdUz
[ "Zitao Song", "Chao Yang", "Chaojie Wang", "Bo An", "Shuang Li" ]
Poster
null
Modern high-stakes systems, such as healthcare or robotics, often generate vast streaming event sequences. Our goal is to design an efficient, plug-and-play tool to elicit logic tree-based explanations from Large Language Models (LLMs) to provide customized insights into each observed event sequence. Built on the tempo...
[]
null
9,816
2406.01124
title_snapshot
[ -0.004297745414078236, -0.022320836782455444, -0.035996321588754654, 0.04744429141283035, 0.0516374371945858, 0.019767891615629196, 0.012640364468097687, 0.026716377586126328, -0.025511616840958595, 0.021720755845308304, -0.01767858676612377, 0.030766025185585022, -0.046341508626937866, -0...
Understanding Finetuning for Factual Knowledge Extraction
https://openreview.net/forum?id=cPsn9AcOYh
[ "Gaurav Rohit Ghosal", "Tatsunori Hashimoto", "Aditi Raghunathan" ]
Poster
null
In this work, we study the impact of QA fine-tuning data on downstream factuality. We show that fine-tuning on lesser-known facts that are poorly stored during pretraining yields significantly worse factuality than fine-tuning on well-known facts, even when all facts are seen during pretraining. We prove this phenomeno...
[]
null
9,813
2406.14785
title_snapshot
[ 0.007338693831115961, -0.04415179789066315, 0.03391444310545921, 0.07170026004314423, 0.06603909283876419, -0.010683136992156506, 0.03219116851687431, 0.003970721736550331, -0.01831057295203209, 0.024113254621624947, -0.04271792247891426, 0.06545901298522949, -0.04682828485965729, -0.01741...
Pruner-Zero: Evolving Symbolic Pruning Metric From Scratch for Large Language Models
https://openreview.net/forum?id=1tRLxQzdep
[ "Peijie Dong", "Lujun Li", "Zhenheng Tang", "Xiang Liu", "Xinglin Pan", "Qiang Wang", "Xiaowen Chu" ]
Poster
null
Despite the remarkable capabilities, Large Language Models (LLMs) face deployment challenges due to their extensive size. Pruning methods drop a subset of weights to accelerate, but many of them require retraining, which is prohibitively expensive and computationally demanding. Recently, post-training pruning approache...
[]
null
9,811
2406.02924
title_snapshot
[ -0.024976834654808044, -0.04044464975595474, -0.023941393941640854, 0.017322558909654617, 0.033488404005765915, 0.059030499309301376, 0.043872661888599396, 0.02607630379498005, -0.032776542007923126, -0.018036188557744026, -0.03875148668885231, 0.04460492730140686, -0.05709218233823776, -0...
tinyBenchmarks: evaluating LLMs with fewer examples
https://openreview.net/forum?id=qAml3FpfhG
[ "Felipe Maia Polo", "Lucas Weber", "Leshem Choshen", "Yuekai Sun", "Gongjun Xu", "Mikhail Yurochkin" ]
Poster
null
The versatility of large language models (LLMs) led to the creation of diverse benchmarks that thoroughly test a variety of language models’ abilities. These benchmarks consist of tens of thousands of examples making evaluation of LLMs very expensive. In this paper, we investigate strategies to reduce the number of eva...
[]
null
9,806
2402.14992
title_snapshot
[ -0.03272611275315285, -0.0190343726426363, -0.010869750753045082, 0.038541071116924286, 0.04185185581445694, 0.024393176659941673, 0.010737845674157143, 0.02065463922917843, -0.035727404057979584, 0.02181714028120041, -0.009565027430653572, 0.040949497371912, -0.052997637540102005, 0.00832...
The Surprising Effectiveness of Skip-Tuning in Diffusion Sampling
https://openreview.net/forum?id=2pYTCy4GUV
[ "Jiajun Ma", "Shuchen Xue", "Tianyang Hu", "Wenjia Wang", "Zhaoqiang Liu", "Zhenguo Li", "Zhi-Ming Ma", "Kenji Kawaguchi" ]
Poster
null
With the incorporation of the UNet architecture, diffusion probabilistic models have become a dominant force in image generation tasks. One key design in UNet is the skip connections between the encoder and decoder blocks. Although skip connections have been shown to improve training stability and model performance, we...
[]
null
9,801
2402.15170
title_snapshot
[ -0.007244459819048643, -0.036421842873096466, -0.017784301191568375, 0.07382640242576599, 0.04997960850596428, 0.025692934170365334, 0.012377796694636345, -0.0012344655115157366, -0.014991541393101215, -0.08526165783405304, 0.0009992505656555295, -0.016638154163956642, -0.04085960611701012, ...
Unveiling the Cycloid Trajectory of EM Iterations in Mixed Linear Regression
https://openreview.net/forum?id=Yn8xnK90mS
[ "Zhankun Luo", "Abolfazl Hashemi" ]
Poster
null
We study the trajectory of iterations and the convergence rates of the Expectation-Maximization (EM) algorithm for two-component Mixed Linear Regression (2MLR). The fundamental goal of MLR is to learn the regression models from unlabeled observations. The EM algorithm finds extensive applications in solving the mixture...
[]
null
9,800
2405.18237
title_snapshot
[ -0.008709137327969074, 0.005203109700232744, -0.0029185214079916477, 0.0034460746683180332, 0.048948753625154495, 0.01878802850842476, 0.027869446203112602, -0.0017997589893639088, -0.038758400827646255, -0.01820005662739277, -0.02012648992240429, -0.0037291962653398514, -0.03175258263945579...
RoboMP$^2$: A Robotic Multimodal Perception-Planning Framework with Multimodal Large Language Models
https://openreview.net/forum?id=eJFQROkaj0
[ "Qi Lv", "Hao Li", "Xiang Deng", "Rui Shao", "Michael Y Wang", "Liqiang Nie" ]
Poster
null
Multimodal Large Language Models (MLLMs) have shown impressive reasoning abilities and general intelligence in various domains. It inspires researchers to train end-to-end MLLMs or utilize large models to generate policies with human-selected prompts for embodied agents. However, these methods exhibit limited generaliz...
[]
null
9,799
2404.04929
title_snapshot
[ 0.003417672123759985, -0.028814416378736496, 0.016712691634893417, 0.027844460681080818, 0.024915585294365883, 0.0156613327562809, 0.01763538084924221, 0.02627529576420784, -0.05830427631735802, -0.015243166126310825, -0.02924599125981331, 0.039220329374074936, -0.08258354663848877, -0.023...
Consistent Diffusion Meets Tweedie: Training Exact Ambient Diffusion Models with Noisy Data
https://openreview.net/forum?id=PlVjIGaFdH
[ "Giannis Daras", "Alex Dimakis", "Constantinos Costis Daskalakis" ]
Poster
null
Ambient diffusion is a recently proposed framework for training diffusion models using corrupted data. Both Ambient Diffusion and alternative SURE-based approaches for learning diffusion models from corrupted data resort to approximations which deteriorate performance. We present the first framework for training diffus...
[]
null
9,798
2404.10177
title_snapshot
[ 0.02193835936486721, -0.011072687804698944, -0.0022571601439267397, 0.060120463371276855, 0.06830763071775436, 0.0208426546305418, 0.013563941232860088, -0.0038675598334521055, -0.025331607088446617, -0.07153865694999695, 0.03239588811993599, 0.007738622836768627, -0.04680398106575012, 0.0...
Token-level Direct Preference Optimization
https://openreview.net/forum?id=1RZKuvqYCR
[ "Yongcheng Zeng", "Guoqing Liu", "Weiyu Ma", "Ning Yang", "Haifeng Zhang", "Jun Wang" ]
Poster
null
Fine-tuning pre-trained Large Language Models (LLMs) is essential to align them with human values and intentions. This process often utilizes methods like pairwise comparisons and KL divergence against a reference LLM, focusing on the evaluation of full answers generated by the models. However, the generation of these ...
[]
null
9,794
2404.11999
title_snapshot
[ -0.012714432552456856, 0.002720301039516926, 0.014139450155198574, 0.038447458297014236, 0.02033412456512451, 0.04841186851263046, -0.007419354747980833, 0.035125359892845154, -0.021484533324837685, 0.01757381670176983, -0.022767072543501854, 0.030350103974342346, -0.05860491096973419, -0....
Mean Field Langevin Actor-Critic: Faster Convergence and Global Optimality beyond Lazy Learning
https://openreview.net/forum?id=FOJE1kRcHG
[ "Kakei Yamamoto", "Kazusato Oko", "Zhuoran Yang", "Taiji Suzuki" ]
Poster
null
This work explores the feature learning capabilities of deep reinforcement learning algorithms in the pursuit of optimal policy determination. We particularly examine an over-parameterized neural actor-critic framework within the mean-field regime, where both actor and critic components undergo updates via policy gradi...
[]
null
9,792
null
null
[ -0.007370728999376297, -0.004564209375530481, 0.015644608065485954, 0.03501760959625244, 0.049125682562589645, 0.03011779487133026, 0.0214321818202734, -0.003345726989209652, -0.03488267958164215, -0.04516707360744476, 0.02110428176820278, 0.03161454573273659, -0.07268950343132019, -0.0008...
Enabling Uncertainty Estimation in Iterative Neural Networks
https://openreview.net/forum?id=N6A6t6xlKm
[ "Nikita Durasov", "Doruk Oner", "Jonathan Donier", "Hieu Le", "Pascal Fua" ]
Poster
null
Turning pass-through network architectures into iterative ones, which use their own output as input, is a well-known approach for boosting performance. In this paper, we argue that such architectures offer an additional benefit: The convergence rate of their successive outputs is highly correlated with the accuracy of ...
[]
null
9,790
2403.16732
title_snapshot
[ 0.010183369740843773, -0.007104723714292049, -0.00928787887096405, 0.046917159110307693, 0.03186134621500969, 0.043905168771743774, 0.018308820202946663, 0.021875306963920593, -0.030469553545117378, -0.05035713315010071, -0.00563642755150795, -0.01781196892261505, -0.05694253742694855, -0....
Clifford-Steerable Convolutional Neural Networks
https://openreview.net/forum?id=XTglHJjzQI
[ "Maksim Zhdanov", "David Ruhe", "Maurice Weiler", "Ana Lucic", "Johannes Brandstetter", "Patrick Forré" ]
Poster
null
We present Clifford-Steerable Convolutional Neural Networks (CS-CNNs), a novel class of ${\operatorname{E}}(p, q)$-equivariant CNNs. CS-CNNs process multivector fields on pseudo-Euclidean spaces $\mathbb{R}^{p,q}$. They specialize, for instance, to ${\operatorname{E}}(3)$-equivariance on $\mathbb{R}^3$ and Poincaré-equ...
[]
null
9,787
2402.14730
title_snapshot
[ -0.00011249948147451505, -0.022020529955625534, 0.02773071452975273, 0.03207821398973465, 0.0182674378156662, 0.0010703849839046597, 0.0034942657221108675, 0.030363522469997406, -0.009019049815833569, -0.057108230888843536, -0.027492063120007515, -0.019594706594944, -0.025678444653749466, ...
Keep the Momentum: Conservation Laws beyond Euclidean Gradient Flows
https://openreview.net/forum?id=hG6gddAKnJ
[ "Sibylle Marcotte", "Rémi Gribonval", "Gabriel Peyré" ]
Poster
null
Conservation laws are well-established in the context of Euclidean gradient flow dynamics, notably for linear or ReLU neural network training. Yet, their existence and principles for non-Euclidean geometries and momentum-based dynamics remain largely unknown. In this paper, we characterize "all" conservation laws in th...
[]
null
9,780
2405.12888
title_snapshot
[ -0.03429108485579491, -0.03671339154243469, 0.026580993086099625, 0.04992656037211418, 0.007375926710665226, 0.032349325716495514, 0.003906463738530874, 0.007351265288889408, -0.038698919117450714, -0.04689330607652664, -0.018209634348750114, -0.032361868768930435, -0.042579181492328644, 0...
Improving Robustness to Multiple Spurious Correlations by Multi-Objective Optimization
https://openreview.net/forum?id=CbbTF6tDhW
[ "Nayeong Kim", "Juwon Kang", "Sungsoo Ahn", "Jungseul Ok", "Suha Kwak" ]
Poster
null
We study the problem of training an unbiased and accurate model given a dataset with multiple biases. This problem is challenging since the multiple biases cause multiple undesirable shortcuts during training, and even worse, mitigating one may exacerbate the other. We propose a novel training method to tackle this cha...
[]
null
9,772
2409.03303
title_snapshot
[ -0.018763510510325432, 0.00005643654367304407, -0.0110212666913867, 0.04569374397397041, 0.017272144556045532, 0.04149267077445984, 0.018487006425857544, 0.001953619997948408, -0.02664896845817566, -0.05815230682492256, 0.0018589829560369253, 0.01984780840575695, -0.09300380200147629, -0.0...
Self-Supervised Coarsening of Unstructured Grid with Automatic Differentiation
https://openreview.net/forum?id=kMBvZ40Iu9
[ "Sergei Shumilin", "Alexander Ryabov", "Nikolay Yavich", "Evgeny Burnaev", "Vladimir Vanovskiy" ]
Poster
null
Due to the high computational load of modern numerical simulation, there is a demand for approaches that would reduce the size of discrete problems while keeping the accuracy reasonable. In this work, we present an original algorithm to coarsen an unstructured grid based on the concepts of differentiable physics. We ac...
[]
null
9,769
2507.18297
title_snapshot
[ -0.04921279475092888, -0.027441538870334625, 0.03196511045098305, 0.030480889603495598, 0.05815248191356659, 0.01607505790889263, 0.010464734397828579, -0.02796512469649315, -0.040396638214588165, -0.04529516398906708, 0.015005378983914852, -0.0007118807407096028, -0.034926533699035645, 0....
FlashST: A Simple and Universal Prompt-Tuning Framework for Traffic Prediction
https://openreview.net/forum?id=vye4OgLaTy
[ "Zhonghang Li", "Lianghao Xia", "Yong Xu", "Chao Huang" ]
Poster
null
The objective of traffic prediction is to accurately forecast and analyze the dynamics of transportation patterns, considering both space and time. However, the presence of distribution shift poses a significant challenge in this field, as existing models struggle to generalize well when faced with test data that signi...
[]
null
9,765
2405.17898
title_snapshot
[ -0.003648573998361826, -0.06488737463951111, 0.04107191413640976, 0.05151662230491638, 0.02983861416578293, 0.016724953427910805, 0.0234318058937788, 0.02881849743425846, -0.011268310248851776, -0.04388410225510597, -0.022071557119488716, 0.007926012389361858, -0.06772671639919281, -0.0141...
PIPER: Primitive-Informed Preference-based Hierarchical Reinforcement Learning via Hindsight Relabeling
https://openreview.net/forum?id=l6Hef6FVd0
[ "Utsav Singh", "Wesley A Suttle", "Brian M. Sadler", "Vinay P. Namboodiri", "Amrit Bedi" ]
Poster
null
In this work, we introduce PIPER: Primitive-Informed Preference-based Hierarchical reinforcement learning via Hindsight Relabeling, a novel approach that leverages preference-based learning to learn a reward model, and subsequently uses this reward model to relabel higher-level replay buffers. Since this reward is unaf...
[]
null
9,762
2404.13423
title_snapshot
[ -0.03210549056529999, -0.02361481823027134, 0.00012471521040424705, 0.04974932223558426, 0.04447764903306961, 0.036138132214546204, 0.017995543777942657, -0.013827957212924957, -0.03587526082992554, -0.05447182059288025, 0.008980427868664265, 0.02738448604941368, -0.06299669295549393, -0.0...
Robust Yet Efficient Conformal Prediction Sets
https://openreview.net/forum?id=MrNq6rbcUi
[ "Soroush H. Zargarbashi", "Mohammad Sadegh Akhondzadeh", "Aleksandar Bojchevski" ]
Poster
null
Conformal prediction (CP) can convert any model's output into prediction sets guaranteed to include the true label with any user-specified probability. However, same as the model itself, CP is vulnerable to adversarial test examples (evasion) and perturbed calibration data (poisoning). We derive provably robust sets by...
[]
null
9,761
2407.09165
title_snapshot
[ 0.031417522579431534, -0.0006763171986676753, 0.011794217862188816, 0.03981552645564079, 0.05632902309298515, -0.021266385912895203, 0.01881355606019497, -0.028922630473971367, -0.01165066659450531, -0.06207018718123436, -0.0072417003102600574, -0.02345886081457138, -0.07507173717021942, 0...
Finite Smoothing Algorithm for High-Dimensional Support Vector Machines and Quantile Regression
https://openreview.net/forum?id=RvwMTDYTOb
[ "Qian Tang", "Yikai Zhang", "Boxiang Wang" ]
Poster
null
This paper introduces a finite smoothing algorithm (FSA), a novel approach to tackle computational challenges in applying support vector machines (SVM) and quantile regression to high-dimensional data. The critical issue with these methods is the non-smooth nature of their loss functions, which traditionally limits the...
[]
null
9,755
null
null
[ -0.029590705409646034, -0.045639410614967346, 0.0036924430169165134, 0.01688493974506855, 0.054974254220724106, 0.056899264454841614, 0.023831747472286224, 0.00020580599084496498, -0.012370402924716473, -0.039290737360715866, -0.0032568813767284155, -0.0004728304920718074, -0.056915555149316...
Scale-Free Image Keypoints Using Differentiable Persistent Homology
https://openreview.net/forum?id=fNJbcxhxRj
[ "Giovanni Barbarani", "Francesco Vaccarino", "Gabriele Trivigno", "Marco Guerra", "Gabriele Berton", "Carlo Masone" ]
Poster
null
In computer vision, keypoint detection is a fundamental task, with applications spanning from robotics to image retrieval; however, existing learning-based methods suffer from scale dependency, and lack flexibility. This paper introduces a novel approach that leverages Morse theory and persistent homology, powerful too...
[]
null
9,753
2406.01315
title_snapshot
[ -0.01590869575738907, -0.0003819916455540806, 0.02682306244969368, 0.00203712098300457, 0.01970057375729084, 0.010199942626059055, 0.042398493736982346, 0.014558115042746067, -0.01550485473126173, -0.046052105724811554, -0.012657561339437962, -0.018559230491518974, -0.08227647095918655, 0....
Smooth Tchebycheff Scalarization for Multi-Objective Optimization
https://openreview.net/forum?id=m4dO5L6eCp
[ "Xi Lin", "Xiaoyuan Zhang", "Zhiyuan Yang", "Fei Liu", "Zhenkun Wang", "Qingfu Zhang" ]
Poster
null
Multi-objective optimization problems can be found in many real-world applications, where the objectives often conflict each other and cannot be optimized by a single solution. In the past few decades, numerous methods have been proposed to find Pareto solutions that represent optimal trade-offs among the objectives fo...
[]
null
9,745
2402.19078
title_snapshot
[ -0.01153250690549612, -0.01173481997102499, 0.023263011127710342, 0.025165805593132973, 0.03545355796813965, 0.05100386589765549, -0.014299331232905388, 0.006027549039572477, -0.008772588334977627, -0.0635780319571495, 0.01956472173333168, 0.0007796736899763346, -0.06575483083724976, 0.002...
DNCs Require More Planning Steps
https://openreview.net/forum?id=tu5fCCuua2
[ "Yara Shamshoum", "Nitzan Hodos", "Yuval Sieradzki", "Assaf Schuster" ]
Poster
null
Many recent works use machine learning models to solve various complex algorithmic problems. However, these models attempt to reach a solution without considering the problem's required computational complexity, which can be detrimental to their ability to solve it correctly. In this work we investigate the effect of c...
[]
null
9,744
2406.02187
title_snapshot
[ -0.04824642464518547, -0.02302272990345955, -0.044003721326589584, 0.04622539505362511, 0.049451153725385666, 0.0243927463889122, 0.01654011942446232, 0.015987280756235123, -0.04696664959192276, -0.03477105870842934, 0.0010863830102607608, 0.0011750273406505585, -0.06224379315972328, 0.002...
Self-Supervised Interpretable End-to-End Learning via Latent Functional Modularity
https://openreview.net/forum?id=dFEeI51O5j
[ "Hyunki Seong", "Hyunchul Shim" ]
Poster
null
We introduce MoNet, a novel functionally modular network for self-supervised and interpretable end-to-end learning. By leveraging its functional modularity with a latent-guided contrastive loss function, MoNet efficiently learns task-specific decision-making processes in latent space without requiring task-level superv...
[]
null
9,741
2403.18947
title_snapshot
[ 0.025274788960814476, -0.00984408799558878, -0.0010735868709161878, 0.022774046286940575, 0.03892778605222702, 0.01931088976562023, 0.010390345007181168, -0.01799483597278595, -0.03101513162255287, -0.03596186265349388, -0.025432512164115906, 0.01381334662437439, -0.04023035988211632, -0.0...
Offline Multi-Objective Optimization
https://openreview.net/forum?id=3AuoStfUIH
[ "Ke Xue", "Rongxi Tan", "Xiaobin Huang", "Chao Qian" ]
Poster
null
Offline optimization aims to maximize a black-box objective function with a static dataset and has wide applications. In addition to the objective function being black-box and expensive to evaluate, numerous complex real-world problems entail optimizing multiple conflicting objectives, i.e., multi-objective optimizatio...
[]
null
9,738
2406.03722
title_snapshot
[ -0.02375805377960205, -0.00714452238753438, 0.01435781829059124, -0.0004602487024385482, 0.03338133916258812, 0.03968321532011032, 0.004731856286525726, 0.021845931187272072, -0.03210101276636124, -0.02437961846590042, -0.031190574169158936, 0.033540017902851105, -0.08567662537097931, -0.0...
Model-Free Robust $\phi$-Divergence Reinforcement Learning Using Both Offline and Online Data
https://openreview.net/forum?id=Yug1IEkvcb
[ "Kishan Panaganti", "Adam Wierman", "Eric Mazumdar" ]
Poster
null
The robust $\phi$-regularized Markov Decision Process (RRMDP) framework focuses on designing control policies that are robust against parameter uncertainties due to mismatches between the simulator (nominal) model and real-world settings. This work makes *two* important contributions. First, we propose a *model-free* a...
[]
null
9,725
2405.05468
title_snapshot
[ -0.03383038938045502, -0.007885805331170559, 0.02299080789089203, 0.02917606011033058, 0.060661084949970245, 0.0020378241315484047, 0.01638631336390972, -0.018272748216986656, -0.014643989503383636, -0.05310980603098869, -0.009767554700374603, 0.010579254478216171, -0.07824870944023132, -0...
Graph Distillation with Eigenbasis Matching
https://openreview.net/forum?id=DYN66IJCI9
[ "Yang Liu", "Deyu Bo", "Chuan Shi" ]
Poster
null
The increasing amount of graph data places requirements on the efficient training of graph neural networks (GNNs). The emerging graph distillation (GD) tackles this challenge by distilling a small synthetic graph to replace the real large graph, ensuring GNNs trained on real and synthetic graphs exhibit comparable perf...
[]
null
9,721
2310.09202
title_snapshot
[ -0.017777152359485626, -0.013088543899357319, 0.016930904239416122, 0.04955313354730606, 0.03559460863471031, 0.01912933960556984, 0.03297507390379906, 0.008851218968629837, -0.02314988151192665, -0.04259670153260231, -0.003262128448113799, -0.025243762880563736, -0.09590935707092285, 0.00...
Random matrix theory improved Fréchet mean of symmetric positive definite matrices
https://openreview.net/forum?id=uQiFsBil3p
[ "Florent Bouchard", "Ammar Mian", "Malik Tiomoko", "Guillaume Ginolhac", "Frederic Pascal" ]
Poster
null
In this study, we consider the realm of covariance matrices in machine learning, particularly focusing on computing Fréchet means on the manifold of symmetric positive definite matrices, commonly referred to as Karcher or geometric means. Such means are leveraged in numerous machine learning tasks. Relying on advanced ...
[]
null
9,715
2405.06558
title_snapshot
[ -0.027116861194372177, -0.0014518297975882888, 0.04169796034693718, 0.011865626089274883, -0.0034546775277704, 0.01771637797355652, 0.06653551012277603, 0.023101720958948135, -0.006177637726068497, -0.0564403161406517, -0.04236271604895592, -0.0034844239708036184, -0.04821702837944031, -0....
MS$^3$D: A RG Flow-Based Regularization for GAN Training with Limited Data
https://openreview.net/forum?id=TuALw8xVum
[ "Jian Wang", "Xin Lan", "Yuxin Tian", "Jiancheng Lv" ]
Poster
null
Generative adversarial networks (GANs) have made impressive advances in image generation, but they often require large-scale training data to avoid degradation caused by discriminator overfitting. To tackle this issue, we investigate the challenge of training GANs with limited data, and propose a novel regularization m...
[]
null
9,703
2408.11135
title_snapshot
[ -0.009820681065320969, -0.028594836592674255, 0.024864917621016502, 0.03287404030561447, 0.02077408693730831, 0.017050402238965034, 0.01412195060402155, -0.01014031283557415, -0.047977522015571594, -0.05547057092189789, 0.000059562375099631026, -0.0057747079990804195, -0.06795992702245712, ...