arXiv subjects

Naftali Tishby

Publications and source records attributed to Naftali Tishby.

At least 37 records · Page 2Linked to original sources

arXiv2017

Mixing Complexity and its Applications to Neural Networks

We suggest analyzing neural networks through the prism of space constraints. We observe that most training algorithms applied in practice use bounded memory, which enables us to use a new notion introduced in the study of space-time tradeoffs that we call mixing complexity. This notion was devised in order to measure the (in)ability to learn using a bounded-memory algorithm. In this paper we describe how we use mixing complexity to obtain new results on what can and cannot be learned using neural networks.

cs.LG↗

arXiv2017

Control Capacity of Partially Observable Dynamic Systems in Continuous Time

Stochastic dynamic control systems relate in a prob- abilistic fashion the space of control signals to the space of corresponding future states. Consequently, stochastic dynamic systems can be interpreted as an information channel between the control space and the state space. In this work we study this control-to-state informartion capacity of stochastic dynamic systems in continuous-time, when the states are observed only partially. The control-to-state capacity, known as empowerment, was shown in the past to be useful in solving various Artificial Intelligence & Control benchmarks, and was used to replace problem-specific utilities. The higher the value of empowerment is, the more optional future states an agent may reach by using its controls inside a given time horizon. The contribution of this work is that we derive an efficient solution for computing the control-to-state information capacity for a linear, partially-observed Gaussian dynamic control system in continuous time, and discover new relationships between control-theoretic and information-theoretic properties of dynamic systems. Particularly, using the derived method, we demonstrate that the capacity between the control signal and the system output does not grow without limits with the length of the control signal. This means that only the near-past window of the control signal contributes effectively to the control-to-state capacity, while most of the information beyond this window is irrelevant for the future state of the dynamic system. We show that empowerment depends on a time constant of a dynamic system.

cs.IT↗

arXiv2016

Memory shapes time perception and intertemporal choices

There is a consensus that human and non-human subjects experience temporal distortions in many stages of their perceptual and decision-making systems. Similarly, intertemporal choice research has shown that decision-makers undervalue future outcomes relative to immediate ones. Here we combine techniques from information theory and artificial intelligence to show how both temporal distortions and intertemporal choice preferences can be explained as a consequence of the coding efficiency of sensorimotor representation. In particular, the model implies that interactions that constrain future behavior are perceived as being both longer in duration and more valuable. Furthermore, using simulations of artificial agents, we investigate how memory constraints enforce a renormalization of the perceived timescales. Our results show that qualitatively different discount functions, such as exponential and hyperbolic discounting, arise as a consequence of an agent's probabilistic model of the world.

q-bio.NC↗

arXiv2015

Optimal Selective Attention in Reactive Agents

In POMDPs, information about the hidden state, delivered through observations, is both valuable to the agent, allowing it to base its actions on better informed internal states, and a "curse", exploding the size and diversity of the internal state space. One attempt to deal with this is to focus on reactive policies, that only base their actions on the most recent observation. However, even reactive policies can be demanding on resources, and agents need to pay selective attention to only some of the information available to them in observations. In this report we present the minimum-information principle for selective attention in reactive agents. We further motivate this approach by reducing the general problem of optimal control in POMDPs, to reactive control with complex observations. Lastly, we explore a newly discovered phenomenon of this optimization process - period doubling bifurcations. This necessitates periodic policies, and raises many more questions regarding stability, periodicity and chaos in optimal control.

cs.LG↗

arXiv2015

Information-Theoretic Bounded Rationality

Bounded rationality, that is, decision-making and planning under resource limitations, is widely regarded as an important open problem in artificial intelligence, reinforcement learning, computational neuroscience and economics. This paper offers a consolidated presentation of a theory of bounded rationality based on information-theoretic ideas. We provide a conceptual justification for using the free energy functional as the objective function for characterizing bounded-rational decisions. This functional possesses three crucial properties: it controls the size of the solution space; it has Monte Carlo planners that are exact, yet bypass the need for exhaustive search; and it captures model uncertainty arising from lack of evidence or from interacting with other agents having unknown intentions. We discuss the single-step decision-making case, and show how to extend it to sequential decisions using equivalence transformations. This extension yields a very general class of decision problems that encompass classical decision rules (e.g. EXPECTIMAX and MINIMAX) as limit cases, as well as trust- and risk-sensitive planning.

stat.ML↗

arXiv2015

Deep Learning and the Information Bottleneck Principle

Deep Neural Networks (DNNs) are analyzed via the theoretical framework of the information bottleneck (IB) principle. We first show that any DNN can be quantified by the mutual information between the layers and the input and output variables. Using this representation we can calculate the optimal information theoretic limits of the DNN and obtain finite sample generalization bounds. The advantage of getting closer to the theoretical limit is quantifiable both by the generalization bound and by the network's simplicity. We argue that both the optimal architecture, number of layers and features/connections at each layer, are related to the bifurcation points of the information bottleneck tradeoff, namely, relevant compression of the input layer with respect to the output layer. The hierarchical representations at the layered network naturally correspond to the structural phase transitions along the information curve. We believe that this new insight can lead to new optimality bounds and deep learning algorithms.

cs.LG↗

arXiv2013

Distribution-Dependent Sample Complexity of Large Margin Learning

We obtain a tight distribution-specific characterization of the sample complexity of large-margin classification with L2 regularization: We introduce the margin-adapted dimension, which is a simple function of the second order statistics of the data distribution, and show distribution-specific upper and lower bounds on the sample complexity, both governed by the margin-adapted dimension of the data distribution. The upper bounds are universal, and the lower bounds hold for the rich family of sub-Gaussian distributions with independent features. We conclude that this new quantity tightly characterizes the true sample complexity of large-margin classification. To prove the lower bound, we develop several new tools of independent interest. These include new connections between shattering and hardness of learning, new properties of shattering with linear classifiers, and a new lower bound on the smallest eigenvalue of a random Gram matrix generated by sub-Gaussian variables. Our results can be used to quantitatively compare large margin learning to other learning rules, and to improve the effectiveness of methods that use sample complexity bounds, such as active learning.

stat.ML↗

arXiv2013

Multivariate Information Bottleneck

The Information bottleneck method is an unsupervised non-parametric data organization technique. Given a joint distribution P(A,B), this method constructs a new variable T that extracts partitions, or clusters, over the values of A that are informative about B. The information bottleneck has already been applied to document classification, gene expression, neural code, and spectral analysis. In this paper, we introduce a general principled framework for multivariate extensions of the information bottleneck method. This allows us to consider multiple systems of data partitions that are inter-related. Our approach utilizes Bayesian networks for specifying the systems of clusters and what information each captures. We show that this construction provides insight about bottleneck variations and enables us to characterize solutions of these variations. We also present a general framework for iterative algorithms for constructing solutions, and apply it to several examples.

cs.LG↗

arXiv2012

Sufficient Dimensionality Reduction with Irrelevant Statistics

The problem of finding a reduced dimensionality representation of categorical variables while preserving their most relevant characteristics is fundamental for the analysis of complex data. Specifically, given a co-occurrence matrix of two variables, one often seeks a compact representation of one variable which preserves information about the other variable. We have recently introduced ``Sufficient Dimensionality Reduction' [GT-2003], a method that extracts continuous reduced dimensional features whose measurements (i.e., expectation values) capture maximal mutual information among the variables. However, such measurements often capture information that is irrelevant for a given task. Widely known examples are illumination conditions, which are irrelevant as features for face recognition, writing style which is irrelevant as a feature for content classification, and intonation which is irrelevant as a feature for speech recognition. Such irrelevance cannot be deduced apriori, since it depends on the details of the task, and is thus inherently ill defined in the purely unsupervised case. Separating relevant from irrelevant features can be achieved using additional side data that contains such irrelevant structures. This approach was taken in [CT-2002], extending the information bottleneck method, which uses clustering to compress the data. Here we use this side-information framework to identify features whose measurements are maximally informative for the original data set, but carry as little information as possible on a side data set. In statistical terms this can be understood as extracting statistics which are maximally sufficient for the original dataset, while simultaneously maximally ancillary for the side dataset. We formulate this tradeoff as a constrained optimization problem and characterize its solutions. We then derive a gradient descent algorithm for this problem, which is based on the Generalized Iterative Scaling method for finding maximum entropy distributions. The method is demonstrated on synthetic data, as well as on real face recognition datasets, and is shown to outperform standard methods such as oriented PCA.

cs.LG↗

arXiv2012

Multi-Instance Learning with Any Hypothesis Class

In the supervised learning setting termed Multiple-Instance Learning (MIL), the examples are bags of instances, and the bag label is a function of the labels of its instances. Typically, this function is the Boolean OR. The learner observes a sample of bags and the bag labels, but not the instance labels that determine the bag labels. The learner is then required to emit a classification rule for bags based on the sample. MIL has numerous applications, and many heuristic algorithms have been used successfully on this problem, each adapted to specific settings or applications. In this work we provide a unified theoretical analysis for MIL, which holds for any underlying hypothesis class, regardless of a specific application or problem domain. We show that the sample complexity of MIL is only poly-logarithmically dependent on the size of the bag, for any underlying hypothesis class. In addition, we introduce a new PAC-learning algorithm for MIL, which uses a regular supervised learning algorithm as an oracle. We prove that efficient PAC-learning for MIL can be generated from any efficient non-MIL supervised learning algorithm that handles one-sided error. The computational complexity of the resulting algorithm is only polynomially dependent on the bag size.

cs.LG↗

arXiv2012

The Minimum Information Principle for Discriminative Learning

Exponential models of distributions are widely used in machine learning for classiffication and modelling. It is well known that they can be interpreted as maximum entropy models under empirical expectation constraints. In this work, we argue that for classiffication tasks, mutual information is a more suitable information theoretic measure to be optimized. We show how the principle of minimum mutual information generalizes that of maximum entropy, and provides a comprehensive framework for building discriminative classiffiers. A game theoretic interpretation of our approach is then given, and several generalization bounds provided. We present iterative algorithms for solving the minimum information problem and its convex dual, and demonstrate their performance on various classiffication tasks. The results show that minimum information classiffiers outperform the corresponding maximum entropy models.

cs.LG↗

arXiv2012

Bounded Planning in Passive POMDPs

In Passive POMDPs actions do not affect the world state, but still incur costs. When the agent is bounded by information-processing constraints, it can only keep an approximation of the belief. We present a variational principle for the problem of maintaining the information which is most useful for minimizing the cost, and introduce an efficient and simple algorithm for finding an optimum.

cs.LG↗

arXiv2012

Tight Sample Complexity of Large-Margin Learning

We obtain a tight distribution-specific characterization of the sample complexity of large-margin classification with L_2 regularization: We introduce the γ-adapted-dimension, which is a simple function of the spectrum of a distribution's covariance matrix, and show distribution-specific upper and lower bounds on the sample complexity, both governed by the γ-adapted-dimension of the source distribution. We conclude that this new quantity tightly characterizes the true sample complexity of large-margin classification. The bounds hold for a rich family of sub-Gaussian distributions.

cs.LG↗

arXiv2007

Efficient representation as a design principle for neural coding and computation

Does the brain construct an efficient representation of the sensory world? We review progress on this question, focusing on a series of experiments in the last decade which use fly vision as a model system in which theory and experiment can confront each other. Although the idea of efficient representation has been productive, clearly it is incomplete since it doesn't tell us which bits of sensory information are most valuable to the organism. We suggest that an organism which maximizes the (biologically meaningful) adaptive value of its actions given fixed resources should have internal representations of the outside world that are optimal in a very specific information theoretic sense: they maximize the information about the future of sensory inputs at a fixed value of the information about their past. This principle contains as special cases computations which the brain seems to carry out, and it should be possible to test this optimization directly. We return to the fly visual system and report the results of preliminary experiments that are in encouraging agreement with theory.

q-bio.NC↗

arXiv2007

Information and fitness

The growth rate of organisms depends both on external conditions and on internal states, such as the expression levels of various genes. We show that to achieve a criterion mean growth rate over an ensemble of conditions, the internal variables must carry a minimum number of bits of information about those conditions. Evolutionary competition thus can select for cellular mechanisms that are more efficient in an abstract, information theoretic sense. Estimates based on recent experiments suggest that the minimum information required for reasonable growth rates is close to the maximum information that can be conveyed through biologically realistic regulatory mechanisms. These ideas are applicable most directly to unicellular organisms, but there are analogies to problems in higher organisms, and we suggest new experiments for both cases.

q-bio.PE↗

arXiv2003

Spike sorting in the frequency domain with overlap detection

This paper deals with the problem of extracting the activity of individual neurons from multi-electrode recordings. Important aspects of this work are: 1) the sorting is done in two stages - a statistical model of the spikes from different cells is built and only then are occurrences of these spikes in the data detected by scanning through the original data, 2) the spike sorting is done in the frequency domain, 3) strict statistical tests are applied to determine if and how a spike should be classiffed, 4) the statistical model for detecting overlaping spike events is proposed, 5) slow dynamics of spike shapes are tracked during long experiments. Results from the application of these techniques to data collected from the escape response system of the American cockroach, Periplaneta americana, are presented.

physics.data-an↗

arXiv2001

Complexity Through Nonextensivity

The problem of defining and studying complexity of a time series has interested people for years. In the context of dynamical systems, Grassberger has suggested that a slow approach of the entropy to its extensive asymptotic limit is a sign of complexity. We investigate this idea further by information theoretic and statistical mechanics techniques and show that these arguments can be made precise, and that they generalize many previous approaches to complexity, in particular unifying ideas from the physics literature with ideas from learning and coding theory; there are even connections of this statistical approach to algorithmic or Kolmogorov complexity. Moreover, a set of simple axioms similar to those used by Shannon in his development of information theory allows us to prove that the divergent part of the subextensive component of the entropy is a unique complexity measure. We classify time series by their complexities and demonstrate that beyond the `logarithmic' complexity classes widely anticipated in the literature there are qualitatively more complex, `power--law' classes which deserve more attention.

physics.data-an↗

arXiv2001

Predictability, complexity and learning

We define {\em predictive information} $I_{\rm pred} (T)$ as the mutual information between the past and the future of a time series. Three qualitatively different behaviors are found in the limit of large observation times $T$: $I_{\rm pred} (T)$ can remain finite, grow logarithmically, or grow as a fractional power law. If the time series allows us to learn a model with a finite number of parameters, then $I_{\rm pred} (T)$ grows logarithmically with a coefficient that counts the dimensionality of the model space. In contrast, power--law growth is associated, for example, with the learning of infinite parameter (or nonparametric) models such as continuous functions with smoothness constraints. There are connections between the predictive information and measures of complexity that have been defined both in learning theory and in the analysis of physical systems through statistical mechanics and dynamical systems theory. Further, in the same way that entropy provides the unique measure of available information consistent with some simple and plausible conditions, we argue that the divergent part of $I_{\rm pred} (T)$ provides the unique measure for the complexity of dynamics underlying a time series. Finally, we discuss how these ideas may be useful in different problems in physics, statistics, and biology.

physics.data-an↗