id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
1402.3722
word2vec Explained: deriving Mikolov et al.'s negative-sampling word-embedding method
cs.CL cs.LG stat.ML
The word2vec software of Tomas Mikolov and colleagues (https://code.google.com/p/word2vec/ ) has gained a lot of traction lately, and provides state-of-the-art word embeddings. The learning models behind the software are described in two research papers. We found the description of the models in these papers to be somewhat cryptic and hard to follow. While the motivations and presentation may be obvious to the neural-networks language-modeling crowd, we had to struggle quite a bit to figure out the rationale behind the equations. This note is an attempt to explain equation (4) (negative sampling) in "Distributed Representations of Words and Phrases and their Compositionality" by Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg Corrado and Jeffrey Dean.
1402.3727
Multi-user Linear Precoding for Multi-polarized Massive MIMO System under Imperfect CSIT
cs.IT math.IT
The space limitation and the channel acquisition prevent Massive MIMO from being easily deployed in a practical setup. Motivated by current deployments of LTE-Advanced, the use of multi-polarized antennas can be an efficient solution to address the space constraint. Furthermore, the dual-structured precoding, in which a preprocessing based on the spatial correlation and a subsequent linear precoding based on the short-term channel state information at the transmitter (CSIT) are concatenated, can reduce the feedback overhead efficiently. By grouping and preprocessing spatially correlated mobile stations (MSs), the dimension of the precoding signal space is reduced and the corresponding short-term CSIT dimension is reduced. In this paper, to reduce the feedback overhead further, we propose a dual-structured multi-user linear precoding, in which the subgrouping method based on co-polarization is additionally applied to the spatially grouped MSs in the preprocessing stage. Furthermore, under imperfect CSIT, the proposed scheme is asymptotically analyzed based on random matrix theory. By investigating the behavior of the asymptotic performance, we also propose a new dual-structured precoding in which the precoding mode is switched between two dual-structured precoding strategies with 1) the preprocessing based only on the spatial correlation and 2) the preprocessing based on both the spatial correlation and polarization. Finally, we extend it to 3D dual-structured precoding.
1402.3735
Decentralized Goal Assignment and Safe Trajectory Generation in Multi-Robot Networks via Multiple Lyapunov Functions
cs.MA cs.RO cs.SY
This paper considers the problem of decentralized goal assignment and trajectory generation for multi-robot networks when only local communication is available, and proposes an approach based on methods related to switched systems and set invariance. A family of Lyapunov-like functions is employed to encode the (local) decision making among candidate goal assignments, under which a group of connected agents chooses the assignment that results in the shortest total distance to the goals. An additional family of Lyapunov-like barrier functions is activated in the case when the optimal assignment may lead to colliding trajectories, maintaining thus system safety while preserving the convergence guarantees. The proposed switching strategies give rise to feedback control policies that are computationally efficient and scalable with the number of agents, and therefore suitable for applications including first-response deployment of robotic networks under limited information sharing. The efficacy of the proposed method is demonstrated via simulation results and experiments with six ground robots.
1402.3749
Particle Computation: Designing Worlds to Control Robot Swarms with only Global Signals
cs.RO
Micro- and nanorobots are often controlled by global input signals, such as an electromagnetic or gravitational field. These fields move each robot maximally until it hits a stationary obstacle or another stationary robot. This paper investigates 2D motion-planning complexity for large swarms of simple mobile robots (such as bacteria, sensors, or smart building material). In previous work we proved it is NP-hard to decide whether a given initial configuration can be transformed into a desired target configuration; in this paper we prove a stronger result: the problem of finding an optimal control sequence is PSPACE-complete. On the positive side, we show we can build useful systems by designing obstacles. We present a reconfigurable hardware platform and demonstrate how to form arbitrary permutations and build a compact absolute encoder. We then take the same platform and use dual-rail logic to build a universal logic gate that concurrently evaluates AND, NAND, NOR and OR operations. Using many of these gates and appropriate interconnects we can evaluate any logical expression.
1402.3783
Map-Aware Models for Indoor Wireless Localization Systems: An Experimental Study
cs.IT math.IT stat.AP
The accuracy of indoor wireless localization systems can be substantially enhanced by map-awareness, i.e., by the knowledge of the map of the environment in which localization signals are acquired. In fact, this knowledge can be exploited to cancel out, at least to some extent, the signal degradation due to propagation through physical obstructions, i.e., to the so called non-line-of-sight bias. This result can be achieved by developing novel localization techniques that rely on proper map-aware statistical modelling of the measurements they process. In this manuscript a unified statistical model for the measurements acquired in map-aware localization systems based on time-of-arrival and received signal strength techniques is developed and its experimental validation is illustrated. Finally, the accuracy of the proposed map-aware model is assessed and compared with that offered by its map-unaware counterparts. Our numerical results show that, when the quality of acquired measurements is poor, map-aware modelling can enhance localization accuracy by up to 110% in certain scenarios.
1402.3797
Scalable Positional Analysis for Studying Evolution of Nodes in Networks
cs.SI physics.soc-ph
In social network analysis, the fundamental idea behind the notion of position is to discover actors who have similar structural signatures. Positional analysis of social networks involves partitioning the actors into disjoint sets using a notion of equivalence which captures the structure of relationships among actors. Classical approaches to Positional Analysis, such as Regular equivalence and Equitable Partitions, are too strict in grouping actors and often lead to trivial partitioning of actors in real world networks. An Epsilon Equitable Partition (EEP) of a graph, which is similar in spirit to Stochastic Blockmodels, is a useful relaxation to the notion of structural equivalence which results in meaningful partitioning of actors. In this paper we propose and implement a new scalable distributed algorithm based on MapReduce methodology to find EEP of a graph. Empirical studies on random power-law graphs show that our algorithm is highly scalable for sparse graphs, thereby giving us the ability to study positional analysis on very large scale networks. We also present the results of our algorithm on time evolving snapshots of the facebook and flickr social graphs. Results show the importance of positional analysis on large dynamic networks.
1402.3801
On Heterogeneous Regenerating Codes and Capacity of Distributed Storage Systems
cs.IT math.IT
Heterogeneous Distributed Storage Systems (DSS) are close to real world applications for data storage. Internet caching system and peer-to-peer storage clouds are the examples of such DSS. In this work, we calculate the capacity formula for such systems where each node store different number of packets and each having a different repair bandwidth (node can be repaired by contacting a specific set of nodes). The tradeoff curve between storage and repair bandwidth is studied for such heterogeneous DSS. By analyzing the capacity formula new minimum bandwidth regenerating (MBR) and minimum storage regenerating (MBR) points are obtained on the curve. It is shown that in some cases these are better than the homogeneous DSS.
1402.3811
Dropout Rademacher Complexity of Deep Neural Networks
cs.NE stat.ML
Great successes of deep neural networks have been witnessed in various real applications. Many algorithmic and implementation techniques have been developed, however, theoretical understanding of many aspects of deep neural networks is far from clear. A particular interesting issue is the usefulness of dropout, which was motivated from the intuition of preventing complex co-adaptation of feature detectors. In this paper, we study the Rademacher complexity of different types of dropout, and our theoretical results disclose that for shallow neural networks (with one or none hidden layer) dropout is able to reduce the Rademacher complexity in polynomial, whereas for deep neural networks it can amazingly lead to an exponential reduction of the Rademacher complexity.
1402.3847
Towards the reproducibility in soil erosion modeling: a new Pan-European soil erosion map
cs.SY cs.CE physics.geo-ph
Soil erosion by water is a widespread phenomenon throughout Europe and has the potentiality, with his on-site and off-site effects, to affect water quality, food security and floods. Despite the implementation of numerous and different models for estimating soil erosion by water in Europe, there is still a lack of harmonization of assessment methodologies. Often, different approaches result in soil erosion rates significantly different. Even when the same model is applied to the same region the results may differ. This can be due to the way the model is implemented (i.e. with the selection of different algorithms when available) and/or to the use of datasets having different resolution or accuracy. Scientific computation is emerging as one of the central topic of the scientific method, for overcoming these problems there is thus the necessity to develop reproducible computational method where codes and data are available. The present study illustrates this approach. Using only public available datasets, we applied the Revised Universal Soil loss Equation (RUSLE) to locate the most sensitive areas to soil erosion by water in Europe. A significant effort was made for selecting the better simplified equations to be used when a strict application of the RUSLE model is not possible. In particular for the computation of the Rainfall Erosivity factor (R) the reproducible research paradigm was applied. The calculation of the R factor was implemented using public datasets and the GNU R language. An easily reproducible validation procedure based on measured precipitation time series was applied using MATLAB language. Designing the computational modelling architecture with the aim to ease as much as possible the future reuse of the model in analysing climate change scenarios is also a challenging goal of the research.
1402.3849
Scalable Kernel Clustering: Approximate Kernel k-means
cs.CV cs.DS cs.LG
Kernel-based clustering algorithms have the ability to capture the non-linear structure in real world data. Among various kernel-based clustering algorithms, kernel k-means has gained popularity due to its simple iterative nature and ease of implementation. However, its run-time complexity and memory footprint increase quadratically in terms of the size of the data set, and hence, large data sets cannot be clustered efficiently. In this paper, we propose an approximation scheme based on randomization, called the Approximate Kernel k-means. We approximate the cluster centers using the kernel similarity between a few sampled points and all the points in the data set. We show that the proposed method achieves better clustering performance than the traditional low rank kernel approximation based clustering schemes. We also demonstrate that its running time and memory requirements are significantly lower than those of kernel k-means, with only a small reduction in the clustering quality on several public domain large data sets. We then employ ensemble clustering techniques to further enhance the performance of our algorithm.
1402.3869
FTVd is beyond Fast Total Variation regularized Deconvolution
cs.CV
In this paper, we revisit the "FTVd" algorithm for Fast Total Variation Regularized Deconvolution, which has been widely used in the past few years. Both its original version implemented in the MATLAB software FTVd 3.0 and its related variant implemented in the latter version FTVd 4.0 are considered \cite{Wang08FTVdsoftware}. We propose that the intermediate results during the iterations are the solutions of a series of combined Tikhonov and total variation regularized image deconvolution models and therefore some of them often have even better image quality than the final solution, which is corresponding to the pure total variation regularized model.
1402.3891
Performance Evaluation of Machine Learning Classifiers in Sentiment Mining
cs.LG cs.CL cs.IR
In recent years, the use of machine learning classifiers is of great value in solving a variety of problems in text classification. Sentiment mining is a kind of text classification in which, messages are classified according to sentiment orientation such as positive or negative. This paper extends the idea of evaluating the performance of various classifiers to show their effectiveness in sentiment mining of online product reviews. The product reviews are collected from Amazon reviews. To evaluate the performance of classifiers various evaluation methods like random sampling, linear sampling and bootstrap sampling are used. Our results shows that support vector machine with bootstrap sampling method outperforms others classifiers and sampling methods in terms of misclassification rate.
1402.3892
Simulating Congestion Dynamics of Train Rapid Transit using Smart Card Data
cs.MA physics.soc-ph
Investigating congestion in train rapid transit systems (RTS) in today's urban cities is a challenge compounded by limited data availability and difficulties in model validation. Here, we integrate information from travel smart card data, a mathematical model of route choice, and a full-scale agent-based model of the Singapore RTS to provide a more comprehensive understanding of the congestion dynamics than can be obtained through analytical modelling alone. Our model is empirically validated, and allows for close inspection of the dynamics including station crowdedness, average travel duration, and frequency of missed trains---all highly pertinent factors in service quality. Using current data, the crowdedness in all 121 stations appears to be distributed log-normally. In our preliminary scenarios, we investigate the effect of population growth on service quality. We find that the current population (2 million) lies below a critical point; and increasing it beyond a factor of $\sim10\%$ leads to an exponential deterioration in service quality. We also predict that incentivizing commuters to avoid the most congested hours can bring modest improvements to the service quality provided the population remains under the critical point. Finally, our model can be used to generate simulated data for analytical modelling when such data are not empirically available, as is often the case.
1402.3895
Bounding Multiple Unicasts through Index Coding and Locally Repairable Codes
cs.IT math.IT
We establish a duality result between linear index coding and Locally Repairable Codes (LRCs). Specifically, we show that a natural extension of LRCs we call Generalized Locally Repairable Codes (GLCRs) are exactly dual to linear index codes. In a GLRC, every node is decodable from a specific set of other nodes and these sets induce a recoverability directed graph. We show that the dual linear subspace of a GLRC is a solution to an index coding instance where the side information graph is this GLRC recoverability graph. We show that the GLRC rate is equivalent to the complementary index coding rate, i.e. the number of transmissions saved by coding. Our second result uses this duality to establish a new upper bound for the multiple unicast network coding problem. In multiple unicast network coding, we are given a directed acyclic graph and r sources that want to send independent messages to r corresponding destinations. Our new upper bound is efficiently computable and relies on a strong approximation result for complementary index coding. We believe that our bound could lead to a logarithmic approximation factor for multiple unicast network coding if a plausible connection we state is verified.
1402.3898
Graph Theory versus Minimum Rank for Index Coding
cs.IT math.IT
We obtain novel index coding schemes and show that they provably outperform all previously known graph theoretic bounds proposed so far. Further, we establish a rather strong negative result: all known graph theoretic bounds are within a logarithmic factor from the chromatic number. This is in striking contrast to minrank since prior work has shown that it can outperform the chromatic number by a polynomial factor in some cases. The conclusion is that all known graph theoretic bounds are not much stronger than the chromatic number.
1402.3902
Sparse Polynomial Learning and Graph Sketching
cs.LG
Let $f:\{-1,1\}^n$ be a polynomial with at most $s$ non-zero real coefficients. We give an algorithm for exactly reconstructing f given random examples from the uniform distribution on $\{-1,1\}^n$ that runs in time polynomial in $n$ and $2s$ and succeeds if the function satisfies the unique sign property: there is one output value which corresponds to a unique set of values of the participating parities. This sufficient condition is satisfied when every coefficient of f is perturbed by a small random noise, or satisfied with high probability when s parity functions are chosen randomly or when all the coefficients are positive. Learning sparse polynomials over the Boolean domain in time polynomial in $n$ and $2s$ is considered notoriously hard in the worst-case. Our result shows that the problem is tractable for almost all sparse polynomials. Then, we show an application of this result to hypergraph sketching which is the problem of learning a sparse (both in the number of hyperedges and the size of the hyperedges) hypergraph from uniformly drawn random cuts. We also provide experimental results on a real world dataset.
1402.3926
Sparse Coding Approach for Multi-Frame Image Super Resolution
cs.CV
An image super-resolution method from multiple observation of low-resolution images is proposed. The method is based on sub-pixel accuracy block matching for estimating relative displacements of observed images, and sparse signal representation for estimating the corresponding high-resolution image. Relative displacements of small patches of observed low-resolution images are accurately estimated by a computationally efficient block matching method. Since the estimated displacements are also regarded as a warping component of image degradation process, the matching results are directly utilized to generate low-resolution dictionary for sparse image representation. The matching scores of the block matching are used to select a subset of low-resolution patches for reconstructing a high-resolution patch, that is, an adaptive selection of informative low-resolution images is realized. When there is only one low-resolution image, the proposed method works as a single-frame super-resolution method. The proposed method is shown to perform comparable or superior to conventional single- and multi-frame super-resolution methods through experiments using various real-world datasets.
1402.3928
Parametrization of completeness in symbolic abstraction of bounded input linear systems
cs.SY
A good state-time quantized symbolic abstraction of an already input quantized control system would satisfy three conditions: proximity, soundness and completeness. Extant approaches for symbolic abstraction of unstable systems limit to satisfying proximity and soundness but not completeness. Instability of systems is an impediment to constructing fully complete state-time quantized symbolic models for bounded and quantized input unstable systems, even using supervisory feedback. Therefore, in this paper we come up with a way of parametrization of completeness of the symbolic model through the quintessential notion of Trimmed-Input Approximate Bisimulation which is introduced in the paper. The amount of completeness is specified by a parameter called trimming of the set of input trajectories. We subsequently discuss a procedure of constructing state-time quantized symbolic models which are near-complete in addition to being sound and proximate with respect to the time quantized models.
1402.3939
IMRank: Influence Maximization via Finding Self-Consistent Ranking
cs.SI cs.DS
Influence maximization, fundamental for word-of-mouth marketing and viral marketing, aims to find a set of seed nodes maximizing influence spread on social network. Early methods mainly fall into two paradigms with certain benefits and drawbacks: (1)Greedy algorithms, selecting seed nodes one by one, give a guaranteed accuracy relying on the accurate approximation of influence spread with high computational cost; (2)Heuristic algorithms, estimating influence spread using efficient heuristics, have low computational cost but unstable accuracy. We first point out that greedy algorithms are essentially finding a self-consistent ranking, where nodes' ranks are consistent with their ranking-based marginal influence spread. This insight motivates us to develop an iterative ranking framework, i.e., IMRank, to efficiently solve influence maximization problem under independent cascade model. Starting from an initial ranking, e.g., one obtained from efficient heuristic algorithm, IMRank finds a self-consistent ranking by reordering nodes iteratively in terms of their ranking-based marginal influence spread computed according to current ranking. We also prove that IMRank definitely converges to a self-consistent ranking starting from any initial ranking. Furthermore, within this framework, a last-to-first allocating strategy and a generalization of this strategy are proposed to improve the efficiency of estimating ranking-based marginal influence spread for a given ranking. In this way, IMRank achieves both remarkable efficiency and high accuracy by leveraging simultaneously the benefits of greedy algorithms and heuristic algorithms. As demonstrated by extensive experiments on large scale real-world social networks, IMRank always achieves high accuracy comparable to greedy algorithms, with computational cost reduced dramatically, even about $10-100$ times faster than other scalable heuristics.
1402.3941
The Saddlepoint Approximation: Unified Random Coding Asymptotics for Fixed and Varying Rates
cs.IT math.IT
This paper presents a saddlepoint approximation of the random-coding union bound of Polyanskiy et al. for i.i.d. random coding over discrete memoryless channels. The approximation is single-letter, and can thus be computed efficiently. Moreover, it is shown to be asymptotically tight for both fixed and varying rates, unifying existing achievability results in the regimes of error exponents, second-order coding rates, and moderate deviations. For fixed rates, novel exact-asymptotics expressions are specified to within a multiplicative 1+o(1) term. A numerical example is provided for which the approximation is remarkably accurate even at short block lengths.
1402.3973
Dimensionality reduction with subgaussian matrices: a unified theory
cs.IT cs.DS math.IT stat.ML
We present a theory for Euclidean dimensionality reduction with subgaussian matrices which unifies several restricted isometry property and Johnson-Lindenstrauss type results obtained earlier for specific data sets. In particular, we recover and, in several cases, improve results for sets of sparse and structured sparse vectors, low-rank matrices and tensors, and smooth manifolds. In addition, we establish a new Johnson-Lindenstrauss embedding for data sets taking the form of an infinite union of subspaces of a Hilbert space.
1402.3986
New Mechanism for Multiagent Extensible Negotiations
cs.MA
Multiagent negotiation mechanisms advise original solutions to several problems for which usual problem solving methods are inappropriate. Mainly negotiation models are based on agents' interactions through messages. Agents interact in order to reach an agreement for solving a specific problem. In this work, we study a new variant of negotiations, which has not yet been addressed in existing works. This negotiation form is denoted extensible negotiation. In contrast with current negotiation models, this form of negotiation allows the agents to dynamically extend the set of items under negotiation. This facility gives more acceptable solutions for the agents in their negotiation. The advantage of enlarging the negotiation space is to certainly offer more facilities for the agents for reaching new agreements which would not have been obtained using usual negotiation methods. This paper presents the protocol and the strategies used by the agents to deal with such negotiations.
1402.4004
Design of a Hybrid Robot Control System using Memristor-Model and Ant-Inspired Based Information Transfer Protocols
cs.RO cs.ET cs.SY
It is not always possible for a robot to process all the information from its sensors in a timely manner and thus quick and yet valid approximations of the robot's situation are needed. Here we design hybrid control for a robot within this limit using algorithms inspired by ant worker placement behaviour and based on memristor-based non-linearity.
1402.4007
Does the D.C. Response of Memristors Allow Robotic Short-Term Memory and a Possible Route to Artificial Time Perception?
cs.RO cs.ET cs.NE
Time perception is essential for task switching, and in the mammalian brain appears alongside other processes. Memristors are electronic components used as synapses and as models for neurons. The d.c. response of memristors can be considered as a type of short-term memory. Interactions of the memristor d.c. response within networks of memristors leads to the emergence of oscillatory dynamics and intermittent spike trains, which are similar to neural dynamics. Based on this data, the structure of a memristor network control for a robot as it undergoes task switching is discussed and it is suggested that these emergent network dynamics could improve the performance of role switching and learning in an artificial intelligence and perhaps create artificial time perception.
1402.4029
Connecting Spiking Neurons to a Spiking Memristor Network Changes the Memristor Dynamics
cs.ET cs.NE physics.bio-ph
Memristors have been suggested as neuromorphic computing elements. Spike-time dependent plasticity and the Hodgkin-Huxley model of the neuron have both been modelled effectively by memristor theory. The d.c. response of the memristor is a current spike. Based on these three facts we suggest that memristors are well-placed to interface directly with neurons. In this paper we show that connecting a spiking memristor network to spiking neuronal cells causes a change in the memristor network dynamics by: removing the memristor spikes, which we show is due to the effects of connection to aqueous medium; causing a change in current decay rate consistent with a change in memristor state; presenting more-linear $I-t$ dynamics; and increasing the memristor spiking rate, as a consequence of interaction with the spiking neurons. This demonstrates that neurons are capable of communicating directly with memristors, without the need for computer translation.
1402.4031
Estimation with Strategic Sensors
cs.GT cs.SY math.OC
We introduce a model of estimation in the presence of strategic, self-interested sensors. We employ a game-theoretic setup to model the interaction between the sensors and the receiver. The cost function of the receiver is equal to the estimation error variance while the cost function of the sensor contains an extra term which is determined by its private information. We start by the single sensor case in which the receiver has access to a noisy but honest side information in addition to the message transmitted by a strategic sensor. We study both static and dynamic estimation problems. For both these problems, we characterize a family of equilibria in which the sensor and the receiver employ simple strategies. Interestingly, for the dynamic estimation problem, we find an equilibrium for which the strategic sensor uses a memory-less policy. We generalize the static estimation setup to multiple sensors with synchronous communication structure (i.e., all the sensors transmit their messages simultaneously). We prove the maybe surprising fact that, for the constructed equilibrium in affine strategies, the estimation quality degrades as the number of sensors increases. However, if the sensors are herding (i.e., copying each other policies), the quality of the receiver's estimation improves as the number of sensors increases. Finally, we consider the asynchronous communication structure (i.e., the sensors transmit their messages sequentially).
1402.4033
Friendship Prediction in Composite Social Networks
cs.SI physics.soc-ph
Friendship prediction is an important task in social network analysis (SNA). It can help users identify friends and improve their level of activity. Most previous approaches predict users' friendship based on their historical records, such as their existing friendship, social interactions, etc. However, in reality, most users have limited friends in a single network, and the data can be very sparse. The sparsity problem causes existing methods to overfit the rare observations and suffer from serious performance degradation. This is particularly true when a new social network just starts to form. We observe that many of today's social networks are composite in nature, where people are often engaged in multiple networks. In addition, users' friendships are always correlated, for example, they are both friends on Facebook and Google+. Thus, by considering those overlapping users as the bridge, the friendship knowledge in other networks can help predict their friendships in the current network. This can be achieved by exploiting the knowledge in different networks in a collective manner. However, as each individual network has its own properties that can be incompatible and inconsistent with other networks, the naive merging of all networks into a single one may not work well. The proposed solution is to extract the common behaviors between different networks via a hierarchical Bayesian model. It captures the common knowledge across networks, while avoiding negative impacts due to network differences. Empirical studies demonstrate that the proposed approach improves the mean average precision of friendship prediction over state-of-the-art baselines on nine real-world social networking datasets significantly.
1402.4036
Is Spiking Logic the Route to Memristor-Based Computers?
cs.ET cond-mat.mtrl-sci cs.AR cs.NE
Memristors have been suggested as a novel route to neuromorphic computing based on the similarity between neurons (synapses and ion pumps) and memristors. The D.C. action of the memristor is a current spike, which we think will be fruitful for building memristor computers. In this paper, we introduce 4 different logical assignations to implement sequential logic in the memristor and introduce the physical rules, summation, `bounce-back', directionality and `diminishing returns', elucidated from our investigations. We then demonstrate how memristor sequential logic works by instantiating a NOT gate, an AND gate and a Full Adder with a single memristor. The Full Adder makes use of the memristor's memory to add three binary values together and outputs the value, the carry digit and even the order they were input in.
1402.4050
Minority Becomes Majority in Social Networks
cs.GT cs.DS cs.MA cs.SI
It is often observed that agents tend to imitate the behavior of their neighbors in a social network. This imitating behavior might lead to the strategic decision of adopting a public behavior that differs from what the agent believes is the right one and this can subvert the behavior of the population as a whole. In this paper, we consider the case in which agents express preferences over two alternatives and model social pressure with the majority dynamics: at each step an agent is selected and its preference is replaced by the majority of the preferences of her neighbors. In case of a tie, the agent does not change her current preference. A profile of the agents' preferences is stable if the preference of each agent coincides with the preference of at least half of the neighbors (thus, the system is in equilibrium). We ask whether there are network topologies that are robust to social pressure. That is, we ask if there are graphs in which the majority of preferences in an initial profile always coincides with the majority of the preference in all stable profiles reachable from that profile. We completely characterize the graphs with this robustness property by showing that this is possible only if the graph has no edge or is a clique or very close to a clique. In other words, except for this handful of graphs, every graph admits at least one initial profile of preferences in which the majority dynamics can subvert the initial majority. We also show that deciding whether a graph admits a minority that becomes majority is NP-hard when the minority size is at most 1/4-th of the social network size.
1402.4053
The Algebraic Approach to Phase Retrieval and Explicit Inversion at the Identifiability Threshold
math.FA cs.CV cs.IT math.AG math.IT stat.ML
We study phase retrieval from magnitude measurements of an unknown signal as an algebraic estimation problem. Indeed, phase retrieval from rank-one and more general linear measurements can be treated in an algebraic way. It is verified that a certain number of generic rank-one or generic linear measurements are sufficient to enable signal reconstruction for generic signals, and slightly more generic measurements yield reconstructability for all signals. Our results solve a few open problems stated in the recent literature. Furthermore, we show how the algebraic estimation problem can be solved by a closed-form algebraic estimation technique, termed ideal regression, providing non-asymptotic success guarantees.
1402.4067
Statistical Noise Analysis in SENSE Parallel MRI
cs.CV
A complete first and second order statistical characterization of noise in SENSE reconstructed data is proposed. SENSE acquisitions have usually been modeled as Rician distributed, since the data reconstruction takes place into the spatial domain, where Gaussian noise is assumed. However, this model just holds for the first order statistics and obviates other effects induced by coils correlations and the reconstruction interpolation. Those effects are properly taken into account in this study, in order to fully justify a final SENSE noise model. As a result, some interesting features of the reconstructed image arise: (1) There is a strong correlation between adjacent lines. (2) The resulting distribution is non-stationary and therefore the variance of noise will vary from point to point across the image. Closed equations for the calculation of the variance of noise and the correlation coefficient between lines are proposed. The proposed model is totally compatible with g-factor formulations.
1402.4069
Application of the Ring Theory in the Segmentation of Digital Images
cs.CV
Ring theory is one of the branches of the abstract algebra that has been broadly used in images. However, ring theory has not been very related with image segmentation. In this paper, we propose a new index of similarity among images using Zn rings and the entropy function. This new index was applied as a new stopping criterion to the Mean Shift Iterative Algorithm with the goal to reach a better segmentation. An analysis on the performance of the algorithm with this new stopping criterion is carried out. The obtained results proved that the new index is a suitable tool to compare images.
1402.4073
Threshold and Symmetric Functions over Bitmaps
cs.DB cs.DS
Bitmap indexes are routinely used to speed up simple aggregate queries in databases. Set operations such as intersections, unions and complements can be represented as logical operations (AND, OR, NOT). However, less is known about the application of bitmap indexes to more advanced queries. We want to extend the applicability of bitmap indexes. As a starting point, we consider symmetric Boolean queries (e.g., threshold functions). For example, we might consider stores as sets of products, and ask for products that are on sale in 2 to 10 stores. Such symmetric Boolean queries generalize intersection, union, and T-occurrence queries. It may not be immediately obvious to an engineer how to use bitmap indexes for symmetric Boolean queries. Yet, maybe surprisingly, we find that the best of our bitmap-based algorithms are competitive with the state-of-the-art algorithms for important special cases (e.g., MergeOpt, MergeSkip, DivideSkip, ScanCount). Moreover, unlike the competing algorithms, the result of our computation is again a bitmap which can be further processed within a bitmap index. We review algorithmic design issues such as the aggregation of many compressed bitmaps. We conclude with a discussion on other advanced queries that bitmap indexes might be able to support efficiently.
1402.4084
Selective Sampling with Drift
cs.LG
Recently there has been much work on selective sampling, an online active learning setting, in which algorithms work in rounds. On each round an algorithm receives an input and makes a prediction. Then, it can decide whether to query a label, and if so to update its model, otherwise the input is discarded. Most of this work is focused on the stationary case, where it is assumed that there is a fixed target model, and the performance of the algorithm is compared to a fixed model. However, in many real-world applications, such as spam prediction, the best target function may drift over time, or have shifts from time to time. We develop a novel selective sampling algorithm for the drifting setting, analyze it under no assumptions on the mechanism generating the sequence of instances, and derive new mistake bounds that depend on the amount of drift in the problem. Simulations on synthetic and real-world datasets demonstrate the superiority of our algorithms as a selective sampling algorithm in the drifting setting.
1402.4100
Generalized Area Spectral Efficiency: An Effective Performance Metric for Green Wireless Communications
cs.NI cs.IT math.IT
Area spectral efficiency (ASE) was introduced as a metric to quantify the spectral utilization efficiency of cellular systems. Unlike other performance metrics, ASE takes into account the spatial property of cellular systems. In this paper, we generalize the concept of ASE to study arbitrary wireless transmissions. Specifically, we introduce the notion of affected area to characterize the spatial property of arbitrary wireless transmissions. Based on the definition of affected area, we define the performance metric, generalized area spectral efficiency (GASE), to quantify the spatial spectral utilization efficiency as well as the greenness of wireless transmissions. After illustrating its evaluation for point-to-point transmission, we analyze the GASE performance of several different transmission scenarios, including dual-hop relay transmission, three-node cooperative relay transmission and underlay cognitive radio transmission. We derive closed-form expressions for the GASE metric of each transmission scenario under Rayleigh fading environment whenever possible. Through mathematical analysis and numerical examples, we show that the GASE metric provides a new perspective on the design and optimization of wireless transmissions, especially on the transmitting power selection. We also show that introducing relay nodes can greatly improve the spatial utilization efficiency of wireless systems. We illustrate that the GASE metric can help optimize the deployment of underlay cognitive radio systems.
1402.4101
First steps to Virtual Mammography: Simulating external compressions of the breast with the Surface Evolver
cs.CE physics.med-ph
In this paper we introduce a computational modelling that reproduces the breast compression processes used to obtain the mammogram. The main result is a programme in which one can track the first steps of virtual mammography. On the one hand, our modelling enables addition of structures that represent different tissues, muscles and glands in the breast. On the other hand, we shall validate and implement it by means of laboratory tests with phantoms. To the best of our knowledge, these two characteristics do confer originality to our research. This is because their interrelation seems not to be properly established elsewhere yet. We conclude that our model reproduces the same shapes and measurements really taken from the volunteer's breasts.
1402.4102
Stochastic Gradient Hamiltonian Monte Carlo
stat.ME cs.LG stat.ML
Hamiltonian Monte Carlo (HMC) sampling methods provide a mechanism for defining distant proposals with high acceptance probabilities in a Metropolis-Hastings framework, enabling more efficient exploration of the state space than standard random-walk proposals. The popularity of such methods has grown significantly in recent years. However, a limitation of HMC methods is the required gradient computation for simulation of the Hamiltonian dynamical system-such computation is infeasible in problems involving a large sample size or streaming data. Instead, we must rely on a noisy gradient estimate computed from a subset of the data. In this paper, we explore the properties of such a stochastic gradient HMC approach. Surprisingly, the natural implementation of the stochastic approximation can be arbitrarily bad. To address this problem we introduce a variant that uses second-order Langevin dynamics with a friction term that counteracts the effects of the noisy gradient, maintaining the desired target distribution as the invariant distribution. Results on simulated data validate our theory. We also provide an application of our methods to a classification task using neural networks and to online Bayesian matrix factorization.
1402.4157
Conservative collision prediction and avoidance for stochastic trajectories in continuous time and space
cs.AI cs.MA cs.RO
Existing work in multi-agent collision prediction and avoidance typically assumes discrete-time trajectories with Gaussian uncertainty or that are completely deterministic. We propose an approach that allows detection of collisions even between continuous, stochastic trajectories with the only restriction that means and variances can be computed. To this end, we employ probabilistic bounds to derive criterion functions whose negative sign provably is indicative of probable collisions. For criterion functions that are Lipschitz, an algorithm is provided to rapidly find negative values or prove their absence. We propose an iterative policy-search approach that avoids prior discretisations and yields collision-free trajectories with adjustably high certainty. We test our method with both fixed-priority and auction-based protocols for coordinating the iterative planning process. Results are provided in collision-avoidance simulations of feedback controlled plants.
1402.4159
Application of Pseudo-Transient Continuation Method in Dynamic Stability Analysis
cs.SY
In this paper, pseudo-transient continuation method has been modified and implemented in power system long-term stability analysis. This method is a middle ground between integration and steady state calculation, thus is a good compromise between accuracy and efficiency. Pseudo-transient continuation method can be applied in the long-term stability model directly to accelerate simulation speed and can also be implemented in the QSS model to overcome numerical difficulties. Numerical examples show that pseudo-transient continuation method can provide correct approximations for the long-term stability model in terms of trajectories and stability assessment.
1402.4179
Network robustness assessed within a dual connectivity perspective
physics.soc-ph cs.SI
Network robustness against attacks has been widely studied in fields as diverse as the Internet, power grids and human societies. Typically, in these studies, robustness is assessed only in terms of the connectivity of the nodes unaffected by the attack. Here we put forward the idea that the connectivity of the affected nodes can play a crucial role in properly evaluating the overall network robustness and its future recovery from the attack. Specifically, we propose a dual perspective approach wherein at any instant in the network evolution under attack, two distinct networks are defined: (i) the Active Network (AN) composed of the unaffected nodes and (ii) the Idle Network (IN) composed of the affected nodes. The proposed robustness metric considers both the efficiency of destroying the AN and the efficiency of building-up the IN. We show, via analysis of both prototype networks and real world data, that trade-offs between the efficiency of Active and Idle network dynamics give rise to surprising crossovers and re-ranking of different attack strategies, pointing to significant implications for decision making.
1402.4225
Information Theory of Matrix Completion
cs.IT math.IT
Matrix completion is a fundamental problem that comes up in a variety of applications like the Netflix problem, collaborative filtering, computer vision, and crowdsourcing. The goal of the problem is to recover a k-by-n unknown matrix from a subset of its noiseless (or noisy) entries. We define an information-theoretic notion of completion capacity C that quantifies the maximum number of entries that one observation of an entry can resolve. This number provides the minimum number m of entries required for reliable reconstruction: m=kn/C. Translating the problem into a distributed joint source-channel coding problem with encoder restriction, we characterize the completion capacity for a wide class of stochastic models of the unknown matrix and the observation process. Our achievability proof is inspired by that of the Slepian-Wolf theorem. For an arbitrary stochastic matrix, we derive an upper bound on the completion capacity.
1402.4238
Downlink and Uplink Energy Minimization Through User Association and Beamforming in Cloud RAN
cs.IT math.IT
The cloud radio access network (C-RAN) concept, in which densely deployed access points (APs) are empowered by cloud computing to cooperatively support mobile users (MUs), to improve mobile data rates, has been recently proposed. However, the high density of active ("on") APs results in severe interference and also inefficient energy consumption. Moreover, the growing popularity of highly interactive applications with stringent uplink (UL) requirements, e.g. network gaming and real-time broadcasting by wireless users, means that the UL transmission is becoming more crucial and requires special attention. Therefore in this paper, we propose a joint downlink (DL) and UL MU-AP association and beamforming design to coordinate interference in the C-RAN for energy minimization, a problem which is shown to be NP hard. Due to the new consideration of UL transmission, it is shown that the two state-of-the-art approaches for finding computationally efficient solutions of joint MU-AP association and beamforming considering only the DL, i.e., group-sparse optimization and relaxed-integer programming, cannot be modified in a straightforward way to solve our problem. Leveraging on the celebrated UL-DL duality result, we show that by establishing a virtual DL transmission for the original UL transmission, the joint DL and UL optimization problem can be converted to an equivalent DL problem in C-RAN with two inter-related subproblems for the original and virtual DL transmissions, respectively. Based on this transformation, two efficient algorithms for joint DL and UL MU-AP association and beamforming design are proposed, whose performances are evaluated and compared with other benchmarking schemes through extensive simulations.
1402.4246
Precoding by Priority: A UEP Scheme for RaptorQ Codes
cs.IT math.IT
Raptor codes are the first class of fountain codes with linear time encoding and decoding. These codes are recommended in standards such as Third Generation Partnership Project (3GPP) and digital video broadcasting. RaptorQ codes are an extension to Raptor codes, having better coding efficiency and flexibility. Standard Raptor and RaptorQ codes are systematic with equal error protection of the data. However, in many applications such as MPEG transmission, there is a need for Unequal Error Protection (UEP): namely, some data symbols require higher error correction capabilities compared to others. We propose an approach that we call Priority Based Precode Ratio (PBPR) to achieve UEP for systematic RaptorQ and Raptor codes. Our UEP assumes that all symbols in a source block belong to the same importance class. The UEP is achieved by changing the number of precode symbols depending on the priority of the information symbols in the source block. PBPR provides UEP with the same number of decoding overhead symbols for source blocks with different importance classes. We demonstrate consistent improvements in the error correction capability of higher importance class compared to the lower importance class across the entire range of channel erasure probabilities. We also show that PBPR does not result in a significant increase in decoding and encoding times compared to the standard implementation.
1402.4259
Extracting Networks of Characters and Places from Written Works with CHAPLIN
cs.CY cs.CL
We are proposing a tool able to gather information on social networks from narrative texts. Its name is CHAPLIN, CHAracters and PLaces Interaction Network, implemented in VB.NET. Characters and places of the narrative works are extracted in a list of raw words. Aided by the interface, the user selects names out of them. After this choice, the tool allows the user to enter some parameters, and, according to them, creates a network where the nodes are the characters and places, and the edges their interactions. Edges are labelled by performances. The output is a GV file, written in the DOT graph scripting language, which is rendered by means of the free open source software Graphviz.
1402.4279
A Bayesian Model of node interaction in networks
cs.LG stat.ME stat.ML
We are concerned with modeling the strength of links in networks by taking into account how often those links are used. Link usage is a strong indicator of how closely two nodes are related, but existing network models in Bayesian Statistics and Machine Learning are able to predict only wether a link exists at all. As priors for latent attributes of network nodes we explore the Chinese Restaurant Process (CRP) and a multivariate Gaussian with fixed dimensionality. The model is applied to a social network dataset and a word coocurrence dataset.
1402.4283
Discretization of Temporal Data: A Survey
cs.DB cs.LG
In real world, the huge amount of temporal data is to be processed in many application areas such as scientific, financial, network monitoring, sensor data analysis. Data mining techniques are primarily oriented to handle discrete features. In the case of temporal data the time plays an important role on the characteristics of data. To consider this effect, the data discretization techniques have to consider the time while processing to resolve the issue by finding the intervals of data which are more concise and precise with respect to time. Here, this research is reviewing different data discretization techniques used in temporal data applications according to the inclusion or exclusion of: class label, temporal order of the data and handling of stream data to open the research direction for temporal data discretization to improve the performance of data mining technique.
1402.4293
The Random Forest Kernel and other kernels for big data from random partitions
stat.ML cs.LG
We present Random Partition Kernels, a new class of kernels derived by demonstrating a natural connection between random partitions of objects and kernels between those objects. We show how the construction can be used to create kernels from methods that would not normally be viewed as random partitions, such as Random Forest. To demonstrate the potential of this method, we propose two new kernels, the Random Forest Kernel and the Fast Cluster Kernel, and show that these kernels consistently outperform standard kernels on problems involving real-world datasets. Finally, we show how the form of these kernels lend themselves to a natural approximation that is appropriate for certain big data problems, allowing $O(N)$ inference in methods such as Gaussian Processes, Support Vector Machines and Kernel PCA.
1402.4303
Finding Preference Profiles of Condorcet Dimension $k$ via SAT
cs.MA cs.AI cs.LO
Condorcet winning sets are a set-valued generalization of the well-known concept of a Condorcet winner. As supersets of Condorcet winning sets are always Condorcet winning sets themselves, an interesting property of preference profiles is the size of the smallest Condorcet winning set they admit. This smallest size is called the Condorcet dimension of a preference profile. Since little is known about profiles that have a certain Condorcet dimension, we show in this paper how the problem of finding a preference profile that has a given Condorcet dimension can be encoded as a satisfiability problem and solved by a SAT solver. Initial results include a minimal example of a preference profile of Condorcet dimension 3, improving previously known examples both in terms of the number of agents as well as alternatives. Due to the high complexity of such problems it remains open whether a preference profile of Condorcet dimension 4 exists.
1402.4304
Automatic Construction and Natural-Language Description of Nonparametric Regression Models
stat.ML cs.LG
This paper presents the beginnings of an automatic statistician, focusing on regression problems. Our system explores an open-ended space of statistical models to discover a good explanation of a data set, and then produces a detailed report with figures and natural-language text. Our approach treats unknown regression functions nonparametrically using Gaussian processes, which has two important consequences. First, Gaussian processes can model functions in terms of high-level properties (e.g. smoothness, trends, periodicity, changepoints). Taken together with the compositional structure of our language of models this allows us to automatically describe functions in simple terms. Second, the use of flexible nonparametric models and a rich language for composing them in an open-ended manner also results in state-of-the-art extrapolation performance evaluated over 13 real time series data sets from various domains.
1402.4306
Student-t Processes as Alternatives to Gaussian Processes
stat.ML cs.AI cs.LG stat.ME
We investigate the Student-t process as an alternative to the Gaussian process as a nonparametric prior over functions. We derive closed form expressions for the marginal likelihood and predictive distribution of a Student-t process, by integrating away an inverse Wishart process prior over the covariance kernel of a Gaussian process model. We show surprising equivalences between different hierarchical Gaussian process models leading to Student-t processes, and derive a new sampling scheme for the inverse Wishart process, which helps elucidate these equivalences. Overall, we show that a Student-t process can retain the attractive properties of a Gaussian process -- a nonparametric representation, analytic marginal and predictive distributions, and easy model selection through covariance kernels -- but has enhanced flexibility, and predictive covariances that, unlike a Gaussian process, explicitly depend on the values of training observations. We verify empirically that a Student-t process is especially useful in situations where there are changes in covariance structure, or in applications like Bayesian optimization, where accurate predictive covariances are critical for good performance. These advantages come at no additional computational cost over Gaussian processes.
1402.4308
Lossy Source Coding with Reconstruction Privacy
cs.IT math.IT
We consider the problem of lossy source coding with side information under a privacy constraint that the reconstruction sequence at a decoder should be kept secret to a certain extent from another terminal such as an eavesdropper, a sender, or a helper. We are interested in how the reconstruction privacy constraint at a particular terminal affects the rate-distortion tradeoff. In this work, we allow the decoder to use a random mapping, and give inner and outer bounds to the rate-distortion-equivocation region for different cases where the side information is available non-causally and causally at the decoder. In the special case where each reconstruction symbol depends only on the source description and current side information symbol, the complete rate-distortion-equivocation region is provided. A binary example illustrating a new tradeoff due to the new privacy constraint, and a gain from the use of a stochastic decoder is given.
1402.4310
Distributed Storage over Unidirectional Ring Networks
cs.IT math.IT
In this paper, we study distributed storage problems over unidirectional ring networks, whose storage nodes form a directed ring and data is transmitted along the same direction. The original data is distributed to store on these nodes. Each user can connect one and only one storage node to download the total data. A lower bound on the reconstructing bandwidth to recover the original data for each user is proposed, and it is achievable for arbitrary parameters. If a distributed storage scheme can achieve this lower bound with equality for every user, we say it an optimal reconstructing distributed storage scheme (ORDSS). Furthermore, the repair problem for a failed storage node in ORDSSes is under consideration and a tight lower bound on the repair bandwidth is obtained. In particular, we indicate the fact that for any ORDSS, every storage node can be repaired with repair bandwidth achieving the lower bound with equality. In addition, we present two constructions for ORDSSes of arbitrary parameters, called MDS construction and ED construction, respectively. Particularly, ED construction, by using the concept of Euclidean division, is more efficient by our analysis in detail.
1402.4322
On the properties of $\alpha$-unchaining single linkage hierarchical clustering
cs.LG
In the election of a hierarchical clustering method, theoretic properties may give some insight to determine which method is the most suitable to treat a clustering problem. Herein, we study some basic properties of two hierarchical clustering methods: $\alpha$-unchaining single linkage or $SL(\alpha)$ and a modified version of this one, $SL^*(\alpha)$. We compare the results with the properties satisfied by the classical linkage-based hierarchical clustering methods.
1402.4325
Rich-cores in networks
physics.soc-ph cs.SI
A core is said to be a group of central and densely connected nodes which governs the overall behavior of a network. Profiling this meso--scale structure currently relies on a limited number of methods which are often complex, and have scalability issues when dealing with very large networks. As a result, we are yet to fully understand its impact on network properties and dynamics. Here we introduce a simple method to profile this structure by combining the concepts of core/periphery and rich-club. The key challenge in addressing such association of the two concepts is to establish a way to define the membership of the core. The notion of a "rich-club" describes nodes which are essentially the hub of a network, as they play a dominating role in structural and functional properties. Interestingly, the definition of a rich-club naturally emphasizes high degree nodes and divides a network into two subgroups. Our approach theoretically couples the underlying principle of a rich-club with the escape time of a random walker, and a rich-core is defined by examining changes in the associated persistence probability. The method is fast and scalable to large networks. In particular, we successfully show that the evolution of the core in \emph{C. elegans} and World Trade networks correspond to key development stages and responses to historical events respectively.
1402.4353
Communication and Interference Coordination
cs.IT math.IT
We study the problem of controlling the interference created to an external observer by a communication processes. We model the interference in terms of its type (empirical distribution), and we analyze the consequences of placing constraints on the admissible type. Considering a single interfering link, we characterize the communication-interference capacity region. Then, we look at a scenario where the interference is jointly created by two users allowed to coordinate their actions prior to transmission. In this case, the trade-off involves communication and interference as well as coordination. We establish an achievable communication-interference region and show that efficiency is significantly improved by coordination.
1402.4354
Hybrid SRL with Optimization Modulo Theories
cs.LG stat.ML
Generally speaking, the goal of constructive learning could be seen as, given an example set of structured objects, to generate novel objects with similar properties. From a statistical-relational learning (SRL) viewpoint, the task can be interpreted as a constraint satisfaction problem, i.e. the generated objects must obey a set of soft constraints, whose weights are estimated from the data. Traditional SRL approaches rely on (finite) First-Order Logic (FOL) as a description language, and on MAX-SAT solvers to perform inference. Alas, FOL is unsuited for con- structive problems where the objects contain a mixture of Boolean and numerical variables. It is in fact difficult to implement, e.g. linear arithmetic constraints within the language of FOL. In this paper we propose a novel class of hybrid SRL methods that rely on Satisfiability Modulo Theories, an alternative class of for- mal languages that allow to describe, and reason over, mixed Boolean-numerical objects and constraints. The resulting methods, which we call Learning Mod- ulo Theories, are formulated within the structured output SVM framework, and employ a weighted SMT solver as an optimization oracle to perform efficient in- ference and discriminative max margin weight learning. We also present a few examples of constructive learning applications enabled by our method.
1402.4360
An Elementary Completeness Proof for Secure Two-Party Computation Primitives
cs.CR cs.IT math.IT
In the secure two-party computation problem, two parties wish to compute a (possibly randomized) function of their inputs via an interactive protocol, while ensuring that neither party learns more than what can be inferred from only their own input and output. For semi-honest parties and information-theoretic security guarantees, it is well-known that, if only noiseless communication is available, only a limited set of functions can be securely computed; however, if interaction is also allowed over general communication primitives (multi-input/output channels), there are "complete" primitives that enable any function to be securely computed. The general set of complete primitives was characterized recently by Maji, Prabhakaran, and Rosulek leveraging an earlier specialized characterization by Kilian. Our contribution in this paper is a simple, self-contained, alternative derivation using elementary information-theoretic tools.
1402.4371
A convergence proof of the split Bregman method for regularized least-squares problems
math.OC cs.LG stat.ML
The split Bregman (SB) method [T. Goldstein and S. Osher, SIAM J. Imaging Sci., 2 (2009), pp. 323-43] is a fast splitting-based algorithm that solves image reconstruction problems with general l1, e.g., total-variation (TV) and compressed sensing (CS), regularizations by introducing a single variable split to decouple the data-fitting term and the regularization term, yielding simple subproblems that are separable (or partially separable) and easy to minimize. Several convergence proofs have been proposed, and these proofs either impose a "full column rank" assumption to the split or assume exact updates in all subproblems. However, these assumptions are impractical in many applications such as the X-ray computed tomography (CT) image reconstructions, where the inner least-squares problem usually cannot be solved efficiently due to the highly shift-variant Hessian. In this paper, we show that when the data-fitting term is quadratic, the SB method is a convergent alternating direction method of multipliers (ADMM), and a straightforward convergence proof with inexact updates is given using [J. Eckstein and D. P. Bertsekas, Mathematical Programming, 55 (1992), pp. 293-318, Theorem 8]. Furthermore, since the SB method is just a special case of an ADMM algorithm, it seems likely that the ADMM algorithm will be faster than the SB method if the augmented Largangian (AL) penalty parameters are selected appropriately. To have a concrete example, we conduct a convergence rate analysis of the ADMM algorithm using two splits for image restoration problems with quadratic data-fitting term and regularization term. According to our analysis, we can show that the two-split ADMM algorithm can be faster than the SB method if the AL penalty parameter of the SB method is suboptimal. Numerical experiments were conducted to verify our analysis.
1402.4380
A Comparative Study of Machine Learning Methods for Verbal Autopsy Text Classification
cs.CL
A Verbal Autopsy is the record of an interview about the circumstances of an uncertified death. In developing countries, if a death occurs away from health facilities, a field-worker interviews a relative of the deceased about the circumstances of the death; this Verbal Autopsy can be reviewed off-site. We report on a comparative study of the processes involved in Text Classification applied to classifying Cause of Death: feature value representation; machine learning classification algorithms; and feature reduction strategies in order to identify the suitable approaches applicable to the classification of Verbal Autopsy text. We demonstrate that normalised term frequency and the standard TFiDF achieve comparable performance across a number of classifiers. The results also show Support Vector Machine is superior to other classification algorithms employed in this research. Finally, we demonstrate the effectiveness of employing a "locally-semi-supervised" feature reduction strategy in order to increase performance accuracy.
1402.4381
Fast X-ray CT image reconstruction using the linearized augmented Lagrangian method with ordered subsets
math.OC cs.LG stat.ML
The augmented Lagrangian (AL) method that solves convex optimization problems with linear constraints has drawn more attention recently in imaging applications due to its decomposable structure for composite cost functions and empirical fast convergence rate under weak conditions. However, for problems such as X-ray computed tomography (CT) image reconstruction and large-scale sparse regression with "big data", where there is no efficient way to solve the inner least-squares problem, the AL method can be slow due to the inevitable iterative inner updates. In this paper, we focus on solving regularized (weighted) least-squares problems using a linearized variant of the AL method that replaces the quadratic AL penalty term in the scaled augmented Lagrangian with its separable quadratic surrogate (SQS) function, thus leading to a much simpler ordered-subsets (OS) accelerable splitting-based algorithm, OS-LALM, for X-ray CT image reconstruction. To further accelerate the proposed algorithm, we use a second-order recursive system analysis to design a deterministic downward continuation approach that avoids tedious parameter tuning and provides fast convergence. Experimental results show that the proposed algorithm significantly accelerates the "convergence" of X-ray CT image reconstruction with negligible overhead and greatly reduces the OS artifacts in the reconstructed image when using many subsets for OS acceleration.
1402.4385
Estimating the resolution limit of the map equation in community detection
physics.soc-ph cs.SI
A community detection algorithm is considered to have a resolution limit if the scale of the smallest modules that can be resolved depends on the size of the analyzed subnetwork. The resolution limit is known to prevent some community detection algorithms from accurately identifying the modular structure of a network. In fact, any global objective function for measuring the quality of a two-level assignment of nodes into modules must have some sort of resolution limit or an external resolution parameter. However, it is yet unknown how the resolution limit affects the so-called map equation, which is known to be an efficient objective function for community detection. We derive an analytical estimate and conclude that the resolution limit of the map equation is set by the total number of links between modules instead of the total number of links in the full network as for modularity. This mechanism makes the resolution limit much less restrictive for the map equation than for modularity, and in practice orders of magnitudes smaller. Furthermore, we argue that the effect of the resolution limit often results from shoehorning multi-level modular structures into two-level descriptions. As we show, the hierarchical map equation effectively eliminates the resolution limit for networks with nested multi-level modular structures.
1402.4388
Automatic Detection of Font Size Straight from Run Length Compressed Text Documents
cs.CV
Automatic detection of font size finds many applications in the area of intelligent OCRing and document image analysis, which has been traditionally practiced over uncompressed documents, although in real life the documents exist in compressed form for efficient storage and transmission. It would be novel and intelligent if the task of font size detection could be carried out directly from the compressed data of these documents without decompressing, which would result in saving of considerable amount of processing time and space. Therefore, in this paper we present a novel idea of learning and detecting font size directly from run-length compressed text documents at line level using simple line height features, which paves the way for intelligent OCRing and document analysis directly from compressed documents. In the proposed model, the given mixed-case text documents of different font size are segmented into compressed text lines and the features extracted such as line height and ascender height are used to capture the pattern of font size in the form of a regression line, using which the automatic detection of font size is done during the recognition stage. The method is experimented with a dataset of 50 compressed documents consisting of 780 text lines of single font size and 375 text lines of mixed font size resulting in an overall accuracy of 99.67%.
1402.4413
Towards Ultra Rapid Restarts
cs.AI cs.LO
We observe a trend regarding restart strategies used in SAT solvers. A few years ago, most state-of-the-art solvers restarted on average after a few thousands of backtracks. Currently, restarting after a dozen backtracks results in much better performance. The main reason for this trend is that heuristics and data structures have become more restart-friendly. We expect further continuation of this trend, so future SAT solvers will restart even more rapidly. Additionally, we present experimental results to support our observations.
1402.4417
Incremental Entity Resolution from Linked Documents
cs.DB cs.IR
In many government applications we often find that information about entities, such as persons, are available in disparate data sources such as passports, driving licences, bank accounts, and income tax records. Similar scenarios are commonplace in large enterprises having multiple customer, supplier, or partner databases. Each data source maintains different aspects of an entity, and resolving entities based on these attributes is a well-studied problem. However, in many cases documents in one source reference those in others; e.g., a person may provide his driving-licence number while applying for a passport, or vice-versa. These links define relationships between documents of the same entity (as opposed to inter-entity relationships, which are also often used for resolution). In this paper we describe an algorithm to cluster documents that are highly likely to belong to the same entity by exploiting inter-document references in addition to attribute similarity. Our technique uses a combination of iterative graph-traversal, locality-sensitive hashing, iterative match-merge, and graph-clustering to discover unique entities based on a document corpus. A unique feature of our technique is that new sets of documents can be added incrementally while having to re-resolve only a small subset of a previously resolved entity-document collection. We present performance and quality results on two data-sets: a real-world database of companies and a large synthetically generated `population' database. We also demonstrate benefit of using inter-document references for clustering in the form of enhanced recall of documents for resolution.
1402.4419
Incremental Majorization-Minimization Optimization with Application to Large-Scale Machine Learning
math.OC cs.LG stat.ML
Majorization-minimization algorithms consist of successively minimizing a sequence of upper bounds of the objective function. These upper bounds are tight at the current estimate, and each iteration monotonically drives the objective function downhill. Such a simple principle is widely applicable and has been very popular in various scientific fields, especially in signal processing and statistics. In this paper, we propose an incremental majorization-minimization scheme for minimizing a large sum of continuous functions, a problem of utmost importance in machine learning. We present convergence guarantees for non-convex and convex optimization when the upper bounds approximate the objective up to a smooth error; we call such upper bounds "first-order surrogate functions". More precisely, we study asymptotic stationary point guarantees for non-convex problems, and for convex ones, we provide convergence rates for the expected objective function value. We apply our scheme to composite optimization and obtain a new incremental proximal gradient algorithm with linear convergence rate for strongly convex functions. In our experiments, we show that our method is competitive with the state of the art for solving machine learning problems such as logistic regression when the number of training samples is large enough, and we demonstrate its usefulness for sparse estimation with non-convex penalties.
1402.4423
New Method for Accurate Parameter Estimation of Induction Motors Based on Artificial Bee Colony Algorithm
cs.SY
This paper proposes an effective method for estimating the parameters of double-cage induction motors by using Artificial Bee Colony (ABC) algorithm. For this purpose the unknown parameters in the electrical model of asynchronous machine are calculated such that the sum of the square of differences between full load torques, starting torques, maximum torques, starting currents, full load currents, and nominal power factors obtained from model and provided by manufacturer is minimized. In order to confirm the efficiency of the proposed method the results are also compared with those achieved by using GA, PSO, and PAMP. The simulations show that in the problem under consideration ABC converges considerably faster than other algorithms and the results are as accurate as PAMP.
1402.4437
Learning the Irreducible Representations of Commutative Lie Groups
cs.LG
We present a new probabilistic model of compact commutative Lie groups that produces invariant-equivariant and disentangled representations of data. To define the notion of disentangling, we borrow a fundamental principle from physics that is used to derive the elementary particles of a system from its symmetries. Our model employs a newfound Bayesian conjugacy relation that enables fully tractable probabilistic inference over compact commutative Lie groups -- a class that includes the groups that describe the rotation and cyclic translation of images. We train the model on pairs of transformed image patches, and show that the learned invariant representation is highly effective for classification.
1402.4442
Artificial Mutation inspired Hyper-heuristic for Runtime Usage of Multi-objective Algorithms
cs.SE cs.NE
In the last years, multi-objective evolutionary algorithms (MOEA) have been applied to different software engineering problems where many conflicting objectives have to be optimized simultaneously. In theory, evolutionary algorithms feature a nice property for runtime optimization as they can provide a solution in any execution time. In practice, based on a Darwinian inspired natural selection, these evolutionary algorithms produce many deadborn solutions whose computation results in a computational resources wastage: natural selection is naturally slow. In this paper, we reconsider this founding analogy to accelerate convergence of MOEA, by looking at modern biology studies: artificial selection has been used to achieve an anticipated specific purpose instead of only relying on crossover and natural selection (i.e., Muller et al [18] research on artificial mutation of fruits with X-Ray). Putting aside the analogy with natural selection , the present paper proposes an hyper-heuristic for MOEA algorithms named Sputnik 1 that uses artificial selective mutation to improve the convergence speed of MOEA. Sputnik leverages the past history of mutation efficiency to select the most relevant mutations to perform. We evaluate Sputnik on a cloud-reasoning engine, which drives on-demand provisioning while considering conflicting performance and cost objectives. We have conducted experiments to highlight the significant performance improvement of Sputnik in terms of resolution time.
1402.4455
Symbiosis of Search and Heuristics for Random 3-SAT
cs.DS cs.AI
When combined properly, search techniques can reveal the full potential of sophisticated branching heuristics. We demonstrate this observation on the well-known class of random 3-SAT formulae. First, a new branching heuristic is presented, which generalizes existing work on this class. Much smaller search trees can be constructed by using this heuristic. Second, we introduce a variant of discrepancy search, called ALDS. Theoretical and practical evidence support that ALDS traverses the search tree in a near-optimal order when combined with the new heuristic. Both techniques, search and heuristic, have been implemented in the look-ahead solver march. The SAT 2009 competition results show that march is by far the strongest complete solver on random k-SAT formulae.
1402.4465
Concurrent Cube-and-Conquer
cs.DS cs.AI
Recent work introduced the cube-and-conquer technique to solve hard SAT instances. It partitions the search space into cubes using a lookahead solver. Each cube is tackled by a conflict-driven clause learning (CDCL) solver. Crucial for strong performance is the cutoff heuristic that decides when to switch from lookahead to CDCL. Yet, this offline heuristic is far from ideal. In this paper, we present a novel hybrid solver that applies the cube and conquer steps simultaneously. A lookahead and a CDCL solver work together on each cube, while communication is restricted to synchronization. Our concurrent cube-and-conquer solver can solve many instances faster than pure lookahead, pure CDCL and offline cube-and-conquer, and can abort early in favor of a pure CDCL search if an instance is not suitable for cube-and-conquer techniques.
1402.4466
Compressed bitmap indexes: beyond unions and intersections
cs.DB cs.DS
Compressed bitmap indexes are used to speed up simple aggregate queries in databases. Indeed, set operations like intersections, unions and complements can be represented as logical operations (AND,OR,NOT) that are ideally suited for bitmaps. However, it is less obvious how to apply bitmaps to more advanced queries. For example, we might seek products in a store that meet some, but maybe not all, criteria. Such threshold queries generalize intersections and unions; they are often used in information-retrieval and data-mining applications. We introduce new algorithms that are sometimes three orders of magnitude faster than a naive approach. Our work shows that bitmap indexes are more broadly applicable than is commonly believed.
1402.4512
Classification with Sparse Overlapping Groups
cs.LG stat.ML
Classification with a sparsity constraint on the solution plays a central role in many high dimensional machine learning applications. In some cases, the features can be grouped together so that entire subsets of features can be selected or not selected. In many applications, however, this can be too restrictive. In this paper, we are interested in a less restrictive form of structured sparse feature selection: we assume that while features can be grouped according to some notion of similarity, not all features in a group need be selected for the task at hand. When the groups are comprised of disjoint sets of features, this is sometimes referred to as the "sparse group" lasso, and it allows for working with a richer class of models than traditional group lasso methods. Our framework generalizes conventional sparse group lasso further by allowing for overlapping groups, an additional flexiblity needed in many applications and one that presents further challenges. The main contribution of this paper is a new procedure called Sparse Overlapping Group (SOG) lasso, a convex optimization program that automatically selects similar features for classification in high dimensions. We establish model selection error bounds for SOGlasso classification problems under a fairly general setting. In particular, the error bounds are the first such results for classification using the sparse group lasso. Furthermore, the general SOGlasso bound specializes to results for the lasso and the group lasso, some known and some new. The SOGlasso is motivated by multi-subject fMRI studies in which functional activity is classified using brain voxels as features, source localization problems in Magnetoencephalography (MEG), and analyzing gene activation patterns in microarray data analysis. Experiments with real and synthetic data demonstrate the advantages of SOGlasso compared to the lasso and group lasso.
1402.4525
Off-Policy General Value Functions to Represent Dynamic Role Assignments in RoboCup 3D Soccer Simulation
cs.AI
Collecting and maintaining accurate world knowledge in a dynamic, complex, adversarial, and stochastic environment such as the RoboCup 3D Soccer Simulation is a challenging task. Knowledge should be learned in real-time with time constraints. We use recently introduced Off-Policy Gradient Descent algorithms within Reinforcement Learning that illustrate learnable knowledge representations for dynamic role assignments. The results show that the agents have learned competitive policies against the top teams from the RoboCup 2012 competitions for three vs three, five vs five, and seven vs seven agents. We have explicitly used subsets of agents to identify the dynamics and the semantics for which the agents learn to maximize their performance measures, and to gather knowledge about different objectives, so that all agents participate effectively and efficiently within the group.
1402.4540
A Unifying Framework for Measuring Weighted Rich Clubs
physics.soc-ph cs.SI
Network analysis can help uncover meaningful regularities in the organization of complex systems. Among these, rich clubs are a functionally important property of a variety of social, technological and biological networks. Rich clubs emerge when nodes that are somehow prominent or 'rich' (e.g., highly connected) interact preferentially with one another. The identification of rich clubs is non-trivial, especially in weighted networks, and to this end multiple distinct metrics have been proposed. Here we describe a unifying framework for detecting rich clubs which intuitively generalizes various metrics into a single integrated method. This generalization rests upon the explicit incorporation of randomized control networks into the measurement process. We apply this framework to real-life examples, and show that, depending on the selection of randomized controls, different kinds of rich-club structures can be detected, such as topological and weighted rich clubs.
1402.4542
Unsupervised Ranking of Multi-Attribute Objects Based on Principal Curves
cs.LG cs.AI stat.ML
Unsupervised ranking faces one critical challenge in evaluation applications, that is, no ground truth is available. When PageRank and its variants show a good solution in related subjects, they are applicable only for ranking from link-structure data. In this work, we focus on unsupervised ranking from multi-attribute data which is also common in evaluation tasks. To overcome the challenge, we propose five essential meta-rules for the design and assessment of unsupervised ranking approaches: scale and translation invariance, strict monotonicity, linear/nonlinear capacities, smoothness, and explicitness of parameter size. These meta-rules are regarded as high level knowledge for unsupervised ranking tasks. Inspired by the works in [8] and [14], we propose a ranking principal curve (RPC) model, which learns a one-dimensional manifold function to perform unsupervised ranking tasks on multi-attribute observations. Furthermore, the RPC is modeled to be a cubic B\'ezier curve with control points restricted in the interior of a hypercube, thereby complying with all the five meta-rules to infer a reasonable ranking list. With control points as the model parameters, one is able to understand the learned manifold and to interpret the ranking list semantically. Numerical experiments of the presented RPC model are conducted on two open datasets of different ranking applications. In comparison with the state-of-the-art approaches, the new model is able to show more reasonable ranking lists.
1402.4543
Normalized Volume of Hyperball in Complex Grassmann Manifold and Its Application in Large-Scale MU-MIMO Communication Systems
cs.IT math.IT
This paper provides a solution to a critical issue in large-scale Multi-User Multiple-Input Multiple-Output (MU-MIMO) communication systems: how to estimate the Signal-to-Interference-plus-Noise-Ratios (SINRs) and their expectations in MU-MIMO mode at the Base Station (BS) side when only the Channel Quality Information (CQI) in Single-User MIMO (SU-MIMO) mode and non-ideal Channel State Information (CSI) are known? A solution to this problem would be very beneficial for the BS to predict the capacity of MU-MIMO and choose the proper modulation and channel coding for MU-MIMO. To that end, this paper derives a normalized volume formula of a hyperball based on the probability density function of the canonical angle between any two points in a complex Grassmann manifold, and shows that this formula provides a solution to the aforementioned issue. It enables the capability of a BS to predict the capacity loss due to non-ideal CSI, group users in MU-MIMO mode, choose the proper modulation and channel coding, and adaptively switch between SU-MIMO and MU-MIMO modes, as well as between Conjugate Beamforming (CB) and Zero-Forcing (ZF) precoding. Numerical results are provided to verify the validity and accuracy of the solution.
1402.4566
Transduction on Directed Graphs via Absorbing Random Walks
cs.CV cs.LG stat.ML
In this paper we consider the problem of graph-based transductive classification, and we are particularly interested in the directed graph scenario which is a natural form for many real world applications. Different from existing research efforts that either only deal with undirected graphs or circumvent directionality by means of symmetrization, we propose a novel random walk approach on directed graphs using absorbing Markov chains, which can be regarded as maximizing the accumulated expected number of visits from the unlabeled transient states. Our algorithm is simple, easy to implement, and works with large-scale graphs. In particular, it is capable of preserving the graph structure even when the input graph is sparse and changes over time, as well as retaining weak signals presented in the directed edges. We present its intimate connections to a number of existing methods, including graph kernels, graph Laplacian based methods, and interestingly, spanning forest of graphs. Its computational complexity and the generalization error are also studied. Empirically our algorithm is systematically evaluated on a wide range of applications, where it has shown to perform competitively comparing to a suite of state-of-the-art methods.
1402.4572
Caching and Coded Multicasting: Multiple Groupcast Index Coding
cs.IT math.IT
The capacity of caching networks has received considerable attention in the past few years. A particularly studied setting is the case of a single server (e.g., a base station) and multiple users, each of which caches segments of files in a finite library. Each user requests one (whole) file in the library and the server sends a common coded multicast message to satisfy all users at once. The problem consists of finding the smallest possible codeword length to satisfy such requests. In this paper we consider the generalization to the case where each user places $L \geq 1$ requests. The obvious naive scheme consists of applying $L$ times the order-optimal scheme for a single request, obtaining a linear in $L$ scaling of the multicast codeword length. We propose a new achievable scheme based on multiple groupcast index coding that achieves a significant gain over the naive scheme. Furthermore, through an information theoretic converse we find that the proposed scheme is approximately optimal within a constant factor of (at most) $18$.
1402.4576
On the Average Performance of Caching and Coded Multicasting with Random Demands
cs.IT cs.NI math.IT
For a network with one sender, $n$ receivers (users) and $m$ possible messages (files), caching side information at the users allows to satisfy arbitrary simultaneous demands by sending a common (multicast) coded message. In the worst-case demand setting, explicit deterministic and random caching strategies and explicit linear coding schemes have been shown to be order optimal. In this work, we consider the same scenario where the user demands are random i.i.d., according to a Zipf popularity distribution. In this case, we pose the problem in terms of the minimum average number of equivalent message transmissions. We present a novel decentralized random caching placement and a coded delivery scheme which are shown to achieve order-optimal performance. As a matter of fact, this is the first order-optimal result for the caching and coded multicasting problem in the case of random demands.
1402.4590
On the distinctness of binary sequences derived from $2$-adic expansion of m-sequences over finite prime fields
cs.IT math.IT
Let $p$ be an odd prime with $2$-adic expansion $\sum_{i=0}^kp_i\cdot2^i$. For a sequence $\underline{a}=(a(t))_{t\ge 0}$ over $\mathbb{F}_{p}$, each $a(t)$ belongs to $\{0,1,\ldots, p-1\}$ and has a unique $2$-adic expansion $$a(t)=a_0(t)+a_1(t)\cdot 2+\cdots+a_{k}(t)\cdot2^k,$$ with $a_i(t)\in\{0, 1\}$. Let $\underline{a_i}$ denote the binary sequence $(a_i(t))_{t\ge 0}$ for $0\le i\le k$. Assume $i_0$ is the smallest index $i$ such that $p_{i}=0$ and $\underline{a}$ and $\underline{b}$ are two different m-sequences generated by a same primitive characteristic polynomial over $\mathbb{F}_p$. We prove that for $i\neq i_0$ and $0\le i\le k$, $\underline{a_i}=\underline{b_i}$ if and only if $\underline{a}=\underline{b}$, and for $i=i_0$, $\underline{a_{i_0}}=\underline{b_{i_0}}$ if and only if $\underline{a}=\underline{b}$ or $\underline{a}=-\underline{b}$. Then the period of $\underline{a_i}$ is equal to the period of $\underline{a}$ if $i\ne i_0$ and half of the period of $\underline{a}$ if $i=i_0$. We also discuss a possible application of the binary sequences $\underline{a_i}$.
1402.4600
Ancillary Service to the Grid Using Intelligent Deferrable Loads
math.OC cs.SY
Renewable energy sources such as wind and solar power have a high degree of unpredictability and time-variation, which makes balancing demand and supply challenging. One possible way to address this challenge is to harness the inherent flexibility in demand of many types of loads. Introduced in this paper is a technique for decentralized control for automated demand response that can be used by grid operators as ancillary service for maintaining demand-supply balance. A Markovian Decision Process (MDP) model is introduced for an individual load. A randomized control architecture is proposed, motivated by the need for decentralized decision making, and the need to avoid synchronization that can lead to large and detrimental spikes in demand. An aggregate model for a large number of loads is then developed by examining the mean field limit. A key innovation is an LTI-system approximation of the aggregate nonlinear model, with a scalar signal as the input and a measure of the aggregate demand as the output. This makes the approximation particularly convenient for control design at the grid level. The second half of the paper contains a detailed application of these results to a network of residential pools. Simulations are provided to illustrate the accuracy of the approximations and effectiveness of the proposed control approach.
1402.4612
Power Allocation in Compressed Sensing of Non-uniformly Sparse Signals
cs.IT math.IT
This paper studies the problem of power allocation in compressed sensing when different components in the unknown sparse signal have different probability to be non-zero. Given the prior information of the non-uniform sparsity and the total power budget, we are interested in how to optimally allocate the power across the columns of a Gaussian random measurement matrix so that the mean squared reconstruction error is minimized. Based on the state evolution technique originated from the work by Donoho, Maleki, and Montanari, we revise the so called approximate message passing (AMP) algorithm for the reconstruction and quantify the MSE performance in the asymptotic regime. Then the closed form of the optimal power allocation is obtained. The results show that in the presence of measurement noise, uniform power allocation, which results in the commonly used Gaussian random matrix with i.i.d. entries, is not optimal for non-uniformly sparse signals. Empirical results are presented to demonstrate the performance gain.
1402.4618
Passive Dynamics in Mean Field Control
cs.SY
Mean-field models are a popular tool in a variety of fields. They provide an understanding of the impact of interactions among a large number of particles or people or other "self-interested agents", and are an increasingly popular tool in distributed control. This paper considers a particular randomized distributed control architecture introduced in our own recent work. In numerical results it was found that the associated mean-field model had attractive properties for purposes of control. In particular, when viewed as an input-output system, its linearization was found to be minimum phase. In this paper we take a closer look at the control model. The results are summarized as follows: (i) The Markov Decision Process framework of Todorov is extended to continuous time models, in which the "control cost" is based on relative entropy. This is the basis of the construction of a family of controlled Markovian generators. (ii) A decentralized control architecture is proposed in which each agent evolves as a controlled Markov process. A central authority broadcasts a common control signal to each agent. The central authority chooses this signal based on an aggregate scalar output of the Markovian agents. (iii) Provided the control-free system is a reversible Markov process, the following identity holds for the linearization, \[ \text{Real} (G(j\omega)) = \text{PSD}_Y(\omega)\ge 0, \quad \omega\in\Re, \] where the right hand side denotes the power spectral density for the output of any one of the individual (control-free) Markov processes.
1402.4645
A Survey on Semi-Supervised Learning Techniques
cs.LG
Semisupervised learning is a learning standard which deals with the study of how computers and natural systems such as human beings acquire knowledge in the presence of both labeled and unlabeled data. Semisupervised learning based methods are preferred when compared to the supervised and unsupervised learning because of the improved performance shown by the semisupervised approaches in the presence of large volumes of data. Labels are very hard to attain while unlabeled data are surplus, therefore semisupervised learning is a noble indication to shrink human labor and improve accuracy. There has been a large spectrum of ideas on semisupervised learning. In this paper we bring out some of the key approaches for semisupervised learning.
1402.4653
Retrieval of Experiments by Efficient Estimation of Marginal Likelihood
stat.ML cs.IR cs.LG
We study the task of retrieving relevant experiments given a query experiment. By experiment, we mean a collection of measurements from a set of `covariates' and the associated `outcomes'. While similar experiments can be retrieved by comparing available `annotations', this approach ignores the valuable information available in the measurements themselves. To incorporate this information in the retrieval task, we suggest employing a retrieval metric that utilizes probabilistic models learned from the measurements. We argue that such a metric is a sensible measure of similarity between two experiments since it permits inclusion of experiment-specific prior knowledge. However, accurate models are often not analytical, and one must resort to storing posterior samples which demands considerable resources. Therefore, we study strategies to select informative posterior samples to reduce the computational load while maintaining the retrieval performance. We demonstrate the efficacy of our approach on simulated data with simple linear regression as the models, and real world datasets.
1402.4662
Optimal Control of Applications for Hybrid Cloud Services
cs.DC cs.SY
Development of cloud computing enables to move Big Data in the hybrid cloud services. This requires research of all processing systems and data structures for provide QoS. Due to the fact that there are many bottlenecks requires monitoring and control system when performing a query. The models and optimization criteria for the design of systems in a hybrid cloud infrastructures are created. In this article suggested approaches and the results of this build.
1402.4663
Concept of Feedback in Future Computing Models to Cloud Systems
cs.DC cs.NI cs.SY
Currently, it is urgent to ensure QoS in distributed computing systems. This became especially important to the development and spread of cloud services. Big data structures become heavily distributed. Necessary to consider the communication channels and data transmission systems and virtualization and scalability in future design of computational models in problems of designing cloud systems, evaluating the effectiveness of the algorithms, the assessment of economic performance data centers. Requires not only the monitoring of data flows and computing resources, but also the operational management of these resources to QoS provide. Such a tool may be just the introduction of feedback in computational models. The article presents the main dynamic model with feedback as a basis for a new model of distributed computing processes. The research results are presented here. Formulated in this work can be used for other complex tasks - estimation of structural complexity of distributed databases, evaluation of dynamic characteristics of systems operating in the hybrid cloud, etc.
1402.4678
When Learners Surpass their Sources: Mathematical Modeling of Learning from an Inconsistent Source
cs.CL
We present a new algorithm to model and investigate the learning process of a learner mastering a set of grammatical rules from an inconsistent source. The compelling interest of human language acquisition is that the learning succeeds in virtually every case, despite the fact that the input data are formally inadequate to explain the success of learning. Our model explains how a learner can successfully learn from or even surpass its imperfect source without possessing any additional biases or constraints about the types of patterns that exist in the language. We use the data collected by Singleton and Newport (2004) on the performance of a 7-year boy Simon, who mastered the American Sign Language (ASL) by learning it from his parents, both of whom were imperfect speakers of ASL. We show that the algorithm possesses a frequency-boosting property, whereby the frequency of the most common form of the source is increased by the learner. We also explain several key features of Simon's ASL.
1402.4699
A Powerful Genetic Algorithm for Traveling Salesman Problem
cs.NE cs.AI
This paper presents a powerful genetic algorithm(GA) to solve the traveling salesman problem (TSP). To construct a powerful GA, I use edge swapping(ES) with a local search procedure to determine good combinations of building blocks of parent solutions for generating even better offspring solutions. Experimental results on well studied TSP benchmarks demonstrate that the proposed GA is competitive in finding very high quality solutions on instances with up to 16,862 cities.
1402.4729
On the Degrees-of-freedom of the 3-user MISO Broadcast Channel with Hybrid CSIT
cs.IT math.IT
The 3-user multiple-input single-output (MISO) broadcast channel (BC) with hybrid channel state information at the transmitter (CSIT) is considered. In this framework, there is perfect and instantaneous CSIT from a subset of users and delayed CSIT from the remaining users. We present new results on the degrees of freedom (DoF) of the 3-user MISO BC with hybrid CSIT. In particular, for the case of 2 transmit antennas, we show that with perfect CSIT from one user and delayed CSIT from the remaining two users, the optimal DoF is 5/3. For the case of 3 transmit antennas and the same hybrid CSIT setting, it is shown that a higher DoF of 9/5 is achievable and this result improves upon the best known bound. Furthermore, with 3 transmit antennas, and the hybrid CSIT setting in which there is perfect CSIT from two users and delayed CSIT from the third one, a novel scheme is presented which achieves 9/4 DoF. Our results also reveal new insights on how to utilize hybrid channel knowledge for multi-user scenarios.
1402.4732
Efficient Inference of Gaussian Process Modulated Renewal Processes with Application to Medical Event Data
stat.ML cs.LG stat.AP
The episodic, irregular and asynchronous nature of medical data render them difficult substrates for standard machine learning algorithms. We would like to abstract away this difficulty for the class of time-stamped categorical variables (or events) by modeling them as a renewal process and inferring a probability density over continuous, longitudinal, nonparametric intensity functions modulating that process. Several methods exist for inferring such a density over intensity functions, but either their constraints and assumptions prevent their use with our potentially bursty event streams, or their time complexity renders their use intractable on our long-duration observations of high-resolution events, or both. In this paper we present a new and efficient method for inferring a distribution over intensity functions that uses direct numeric integration and smooth interpolation over Gaussian processes. We demonstrate that our direct method is up to twice as accurate and two orders of magnitude more efficient than the best existing method (thinning). Importantly, the direct method can infer intensity functions over the full range of bursty to memoryless to regular events, which thinning and many other methods cannot. Finally, we apply the method to clinical event data and demonstrate the face-validity of the abstraction, which is now amenable to standard learning algorithms.
1402.4738
A measure of compression gain for new symbols in data-compression
cs.IT math.IT
Huffman encoding is often improved by using block codes, for example a 3-block would be an alphabet consisting of each possible combination of three characters. We take the approach of starting with a base alphabet and expanding it to include frequently occurring aggregates of symbols. We prove that the change in compressed message length by the introduction of a new aggregate symbol can be expressed as the difference of two entropies, dependent only on the probabilities and length of the introduced symbol. The expression is independent of the probability of all other symbols in the alphabet. This measure of information gain, for a new symbol, can be applied in data compression methods. We also demonstrate that aggregate symbol alphabets, as opposed to mutually exclusive alphabets have the potential to provide good levels of compression, with a simple experiment. Finally, compression gain as defined in this paper may also be useful for feature selection.
1402.4741
A normative account of defeasible and probabilistic inference
cs.LO cs.AI
In this paper, we provide more evidence for the contention that logical consequence should be understood in normative terms. Hartry Field and John MacFarlane covered the classical case. We extend their work, examining what it means for an agent to be obliged to infer a conclusion when faced with uncertain information or reasoning within a non-monotonic, defeasible, logical framework (which allows e. g. for inference to be drawn from premises considered true unless evidence to the contrary is presented).
1402.4742
IVOA Recommendation: TAPRegExt: a VOResource Schema Extension for Describing TAP Services
astro-ph.IM cs.DB
This document describes an XML encoding standard for metadata about services implementing the table access protocol TAP [TAP], referred to as TAPRegExt. Instance documents are part of the service's registry record or can be obtained from the service itself. They deliver information to both humans and software on the languages, output formats, and upload methods supported by the service, as well as data models implemented by the exposed tables, optional language features, and certain limits enforced by the service.
1402.4746
Near-optimal-sample estimators for spherical Gaussian mixtures
cs.LG cs.DS cs.IT math.IT stat.ML
Statistical and machine-learning algorithms are frequently applied to high-dimensional data. In many of these applications data is scarce, and often much more costly than computation time. We provide the first sample-efficient polynomial-time estimator for high-dimensional spherical Gaussian mixtures. For mixtures of any $k$ $d$-dimensional spherical Gaussians, we derive an intuitive spectral-estimator that uses $\mathcal{O}_k\bigl(\frac{d\log^2d}{\epsilon^4}\bigr)$ samples and runs in time $\mathcal{O}_{k,\epsilon}(d^3\log^5 d)$, both significantly lower than previously known. The constant factor $\mathcal{O}_k$ is polynomial for sample complexity and is exponential for the time complexity, again much smaller than what was previously known. We also show that $\Omega_k\bigl(\frac{d}{\epsilon^2}\bigr)$ samples are needed for any algorithm. Hence the sample complexity is near-optimal in the number of dimensions. We also derive a simple estimator for one-dimensional mixtures that uses $\mathcal{O}\bigl(\frac{k \log \frac{k}{\epsilon} }{\epsilon^2} \bigr)$ samples and runs in time $\widetilde{\mathcal{O}}\left(\bigl(\frac{k}{\epsilon}\bigr)^{3k+1}\right)$. Our other technical contributions include a faster algorithm for choosing a density estimate from a set of distributions, that minimizes the $\ell_1$ distance to an unknown underlying distribution.
1402.4799
Multiple Access Channel with Common Message and Secrecy constraint
cs.IT math.IT
In this paper, we study the problem of secret communication over a multiple-access channel with a common message. Here, we assume that two transmitters have confidential messages, which must be kept secret from the wiretapper (the second receiver), and both of them have access to a common message which can be decoded by the two receivers. We call this setting as Multiple-Access Wiretap Channel with Common message (MAWC-CM). For this setting, we derive general inner and outer bounds on the secrecy capacity region for the discrete memoryless case and show that these bounds meet each other for a special case called the switch channel. As well, for a Gaussian version of MAWC-CM, we derive inner and outer bounds on the secrecy capacity region. Providing numerical results for the Gaussian case, we illustrate the comparison between the derived achievable rate region and the outer bound for the considered model and the capacity region of compound multiple access channel.
1402.4802
Ambiguity in language networks
physics.soc-ph cs.CL q-bio.NC
Human language defines the most complex outcomes of evolution. The emergence of such an elaborated form of communication allowed humans to create extremely structured societies and manage symbols at different levels including, among others, semantics. All linguistic levels have to deal with an astronomic combinatorial potential that stems from the recursive nature of languages. This recursiveness is indeed a key defining trait. However, not all words are equally combined nor frequent. In breaking the symmetry between less and more often used and between less and more meaning-bearing units, universal scaling laws arise. Such laws, common to all human languages, appear on different stages from word inventories to networks of interacting words. Among these seemingly universal traits exhibited by language networks, ambiguity appears to be a specially relevant component. Ambiguity is avoided in most computational approaches to language processing, and yet it seems to be a crucial element of language architecture. Here we review the evidence both from language network architecture and from theoretical reasonings based on a least effort argument. Ambiguity is shown to play an essential role in providing a source of language efficiency, and is likely to be an inevitable byproduct of network growth.
1402.4834
The Application of Imperialist Competitive Algorithm for Fuzzy Random Portfolio Selection Problem
math.OC cs.AI
This paper presents an implementation of the Imperialist Competitive Algorithm (ICA) for solving the fuzzy random portfolio selection problem where the asset returns are represented by fuzzy random variables. Portfolio Optimization is an important research field in modern finance. By using the necessity-based model, fuzzy random variables reformulate to the linear programming and ICA will be designed to find the optimum solution. To show the efficiency of the proposed method, a numerical example illustrates the whole idea on implementation of ICA for fuzzy random portfolio selection problem.
1402.4844
Subspace Learning with Partial Information
cs.LG stat.ML
The goal of subspace learning is to find a $k$-dimensional subspace of $\mathbb{R}^d$, such that the expected squared distance between instance vectors and the subspace is as small as possible. In this paper we study subspace learning in a partial information setting, in which the learner can only observe $r \le d$ attributes from each instance vector. We propose several efficient algorithms for this task, and analyze their sample complexity
1402.4845
Diffusion Least Mean Square: Simulations
cs.LG cs.MA
In this technical report we analyse the performance of diffusion strategies applied to the Least-Mean-Square adaptive filter. We configure a network of cooperative agents running adaptive filters and discuss their behaviour when compared with a non-cooperative agent which represents the average of the network. The analysis provides conditions under which diversity in the filter parameters is beneficial in terms of convergence and stability. Simulations drive and support the analysis.