This beginning graduate textbook describes both recent achievements and classical results of computational complexity theory. Requiring essentially no background apart from mathematical maturity, the book can be used as a reference for self-study for anyone interested in complexity, including physicists, mathematicians, and other scientists, as well as a textbook for a variety of courses and seminars. More than 300 exercises are included with a selected hint set.

/pdf/computational-complexity-a-modern-approach-5gtk8urev9.pdf

Computational Complexity: A Modern Approach

Recent years have witnessed significant advances in reinforcement learning (RL), which has registered tremendous success in solving various sequential decision-making problems in machine learning. Most of the successful RL applications, e.g., the games of Go and Poker, robotics, and autonomous driving, involve the participation of more than one single agent, which naturally fall into the realm of multi-agent RL (MARL), a domain with a relatively long history, and has recently re-emerged due to advances in single-agent RL techniques. Though empirically successful, theoretical foundations for MARL are relatively lacking in the literature. In this chapter, we provide a selective overview of MARL, with focus on algorithms backed by theoretical analysis. More specifically, we review the theoretical results of MARL algorithms mainly within two representative frameworks, Markov/stochastic games and extensive-form games, in accordance with the types of tasks they address, i.e., fully cooperative, fully competitive, and a mix of the two. We also introduce several significant but challenging applications of these algorithms. Orthogonal to the existing reviews on MARL, we highlight several new angles and taxonomies of MARL theory, including learning in extensive-form games, decentralized MARL with networked agents, MARL in the mean-field regime, (non-)convergence of policy-based methods for learning in games, etc. Some of the new angles extrapolate from our own research endeavors and interests. Our overall goal with this chapter is, beyond providing an assessment of the current state of the field on the mark, to identify fruitful future research directions on theoretical studies of MARL. We expect this chapter to serve as continuing stimulus for researchers interested in working on this exciting while challenging topic.

Multi-Agent Reinforcement Learning: A Selective Overview of Theories and Algorithms

The nearest neighbor classifier is one of the most used and well-known techniques for performing recognition tasks. It has also demonstrated itself to be one of the most useful algorithms in data mining in spite of its simplicity. However, the nearest neighbor classifier suffers from several drawbacks such as high storage requirements, low efficiency in classification response, and low noise tolerance. These weaknesses have been the subject of study for many researchers and many solutions have been proposed. Among them, one of the most promising solutions consists of reducing the data used for establishing a classification rule (training data) by means of selecting relevant prototypes. Many prototype selection methods exist in the literature and the research in this area is still advancing. Different properties could be observed in the definition of them, but no formal categorization has been established yet. This paper provides a survey of the prototype selection methods proposed in the literature from a theoretical and empirical point of view. Considering a theoretical point of view, we propose a taxonomy based on the main characteristics presented in prototype selection and we analyze their advantages and drawbacks. Empirically, we conduct an experimental study involving different sizes of data sets for measuring their performance in terms of accuracy, reduction capabilities, and runtime. The results obtained by all the methods studied have been verified by nonparametric statistical tests. Several remarks, guidelines, and recommendations are made for the use of prototype selection for nearest neighbor classification.

/pdf/prototype-selection-for-nearest-neighbor-classification-3c75s4xkiz.pdf

Prototype Selection for Nearest Neighbor Classification: Taxonomy and Empirical Study

Preliminary Definitions, Results and Motivation Stable Matching Problems: The Stable Marriage Problem: An Update SM and HR with Indifference The Stable Roommates Problem Further Stable Matching Problems Other Optimal Matching Problems: Pareto Optimal Matchings Popular Matchings Profile-Based Optimal Matchings.

/pdf/algorithmics-of-matching-under-preferences-2l3a9z5euq.pdf

Algorithmics of Matching Under Preferences

We formalize a new cryptographic primitive that we call Message-Locked Encryption (MLE), where the key under which encryption and decryption are performed is itself derived from the message. MLE provides a way to achieve secure deduplication (space-efficient secure outsourced storage), a goal currently targeted by numerous cloudstorage providers. We provide definitions both for privacy and for a form of integrity that we call tag consistency. Based on this foundation, we make both practical and theoretical contributions. On the practical side, we provide ROM security analyses of a natural family of MLE schemes that includes deployed schemes. On the theoretical side the challenge is standard model solutions, and we make connections with deterministic encryption, hash functions secure on correlated inputs and the sample-then-extract paradigm to deliver schemes under different assumptions and for different classes of message sources. Our work shows that MLE is a primitive of both practical and theoretical interest.

/pdf/message-locked-encryption-and-secure-deduplication-45mml7x0tj.pdf

Message-Locked Encryption and Secure Deduplication

A linear-time component-labeling algorithm using contour tracing technique

We study the online convex optimization problem, in which an online algorithm has to make repeated decisions with convex loss functions and hopes to achieve a small regret. We consider a natural restriction of this problem in which the loss functions have a small deviation, measured by the sum of the distances between every two consecutive loss functions, according to some distance metrics. We show that for the linear and general smooth convex loss functions, an online algorithm modified from the gradient descend algorithm can achieve a regret which only scales as the square root of the deviation. For the closely related problem of prediction with expert advice, we show that an online algorithm modified from the multiplicative update algorithm can also achieve a similar regret bound for a different measure of deviation. Finally, for loss functions which are strictly convex, we show that an online algorithm modified from the online Newton step algorithm can achieve a regret which is only logarithmic in terms of the deviation, and as an application, we can also have such a logarithmic regret for the portfolio management problem.

/pdf/online-optimization-with-gradual-variations-rbdo2robym.pdf

Online Optimization with Gradual Variations

This paper provides the first explicit construction of extractors which are simultaneously optimal up to constant factors in both seed length and output length. More precisely, for every n,k, our extractor uses a random seed of length O(log n) to transform any random source on n bits with (min-)entropy k, into a distribution on (1-α)k bits that is e-close to uniform. Here α and e can be taken to be any positive constants. (In fact, e can be almost polynomially small.Our improvements are obtained via three new techniques, each of which may be of independent interest. The first is a general construction of mergers [22] from locally decodable error-correcting codes. The second introduces new condensers that have constant seed length (and retain a constant fraction of the min-entropy in the random source). The third is a way to augment the "win-win repeated condensing" paradigm of [17] with error reduction techniques like [15] so that the our constant seed-length condensers can be used without error accumulation.

/pdf/extractors-optimal-up-to-constant-factors-w967r2hrzp.pdf

Extractors: optimal up to constant factors

We study conditional computational entropy: the amount of randomness a distribution appears to have to a computationally bounded observer who is given some correlated information. By considering conditional versions of HILL entropy (based on indistinguishability from truly random distributions) and Yao entropy (based on incompressibility), we obtain:


a separation between conditional HILL and Yao entropies (which can be viewed as a separation between the traditional HILL and Yao entropies in the shared random string model, improving on Wee's 2004 separation in the random oracle model);
the first demonstration of a distribution from which extraction techniques based on Yao entropy produce more pseudorandom bits than appears possible by the traditional HILL-entropy-based techniques;
a new, natural notion of unpredictability entropy, which implies conditional Yao entropy and thus allows for known extraction and hardcore bit results to be stated and used more generally.

/pdf/conditional-computational-entropy-or-toward-separating-szdqx2617q.pdf

Conditional Computational Entropy, or Toward Separating Pseudoentropy from Compressibility

To handle problems created by large data sets, we propose a method that uses a decision tree to decompose a given data space and trains SVMs on the decomposed regions. Although there are other means of decomposing a data space, we show that the decision tree has several merits for large-scale SVM training. First, it can classify some data points by its own means, thereby reducing the cost of SVM training applied to the remaining data points. Second, it is efficient for seeking the parameter values that maximize the validation accuracy, which helps maintain good test accuracy. For experiment data sets whose size can be handled by current non-linear, or kernel-based, SVM training techniques, the proposed method can speed up the training by a factor of thousands, and still achieve comparable test accuracy.

/pdf/tree-decomposition-for-large-scale-svm-problems-1qy5uuma8m.pdf

Chi-Jen Lu

Papers

A linear-time component-labeling algorithm using contour tracing technique

Online Optimization with Gradual Variations

Extractors: optimal up to constant factors

Conditional Computational Entropy, or Toward Separating Pseudoentropy from Compressibility

Tree Decomposition for Large-Scale SVM Problems