
Olivier BousquetGoogle Inc. | Google · Research Department
Olivier Bousquet
PhD, Ecole Polytechnique
About
116
Publications
90,013
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
24,928
Citations
Introduction
Additional affiliations
May 2002 - May 2004
Education
September 1999 - May 2002
Publications
Publications (116)
We study deep neural networks (DNNs) trained on natural image data with entirely random labels. Despite its popularity in the literature, where it is often used to study memorization, generalization, and other phenomena, little is known about what DNNs learn in this setting. In this paper, we show analytically for convolutional and fully connected...
Recent progress in the field of reinforcement learning has been accelerated by virtual learning environments such as video games, where novel algorithms and ideas can be quickly tested in a safe and reproducible manner. We introduce the Google Research Football Environment, a new reinforcement learning environment where agents are trained to play f...
We study the prediction of the accuracy of a neural network given only its weights with the goal of better understanding network training and performance. To do so, we propose a formal setting which frames this task and connects to previous work in this area. We collect (and release) a large dataset of almost 80k convolutional neural networks train...
Representation learning promises to unlock deep learning for the long tail of vision tasks without expansive labelled datasets. Yet, the absence of a unified yardstick to evaluate general visual representations hinders progress. Many sub-fields promise representations, but each has different evaluation protocols that are either too constrained (lin...
Recent progress in the field of reinforcement learning has been accelerated by virtual learning environments such as video games, where novel algorithms and ideas can be quickly tested in a safe and reproducible manner. We introduce the Google Research Football Environment, a new reinforcement learning environment where agents are trained to play f...
In semi-supervised classification, one is given access both to labeled and unlabeled data. As unlabeled data is typically cheaper to acquire than labeled data, this setup becomes advantageous as soon as one can exploit the unlabeled data in order to produce a better classifier than with labeled data alone. However, the conditions under which such a...
The estimation of an f-divergence between two probability distributions based on samples is a fundamental problem in statistics and machine learning. Most works study this problem under very weak assumptions, in which case it is provably hard. We consider the case of stronger structural assumptions that are commonly satisfied in modern machine lear...
Despite the tremendous progress in the estimation of generative models, the development of tools for diagnosing their failures and assessing their performance has advanced at a much slower pace. Recent developments have investigated metrics that quantify which parts of the true distribution are modeled well, and, on the contrary, what the model fai...
Recent advances in generative modeling have led to an increased interest in the study of statistical divergences as means of model comparison. Commonly used evaluation methods, such as Fr\'echet Inception Distance (FID), correlate well with the perceived quality of samples and are sensitive to mode dropping. However, these metrics are unable to dis...
Generative adversarial networks (GAN) are a powerful subclass of generative models. Despite a very rich research activity leading to numerous interesting GAN algorithms, it is still very hard to assess which algorithm(s) perform better than others. We conduct a neutral, multi-faceted large-scale empirical study on state-of-the art models and evalua...
We propose the Wasserstein Auto-Encoder (WAE)---a new algorithm for building a generative model of the data distribution. WAE minimizes a penalized form of the Wasserstein distance between the model distribution and the target distribution, which leads to a different regularizer than the one used by the Variational Auto-Encoder (VAE). This regulari...
This paper aims at one-shot learning of deep neural nets, where a highly parallel setting is considered to address the algorithm calibration problem - selecting the best neural architecture and learning hyper-parameter values depending on the dataset at hand. The notoriously expensive calibration problem is optimally reduced by detecting and early...
The selection of hyper-parameters is critical in Deep Learning. Because of the long training time of complex models and the availability of compute resources in the cloud, "one-shot" optimization schemes - where the sets of hyper-parameters are selected in advance (e.g. on a grid or in a random manner) and the training is executed in parallel - are...
Generative adversarial networks (GAN) approximate a target data distribution by jointly optimizing an objective function through a "two-player game" between a generator and a discriminator. Despite their empirical success, however, two very basic questions on how well they can approximate the target distribution remain unanswered. First, it is not...
Generic text embeddings are successfully used in a variety of tasks. However, they are often learnt by capturing the co-occurrence structure from pure text corpora, resulting in limitations of their ability to generalize. In this paper, we explore models that incorporate visual information into the text representation. Based on comprehensive ablati...
We study unsupervised generative modeling in terms of the optimal transport (OT) problem between true (but unknown) data distribution $P_X$ and the latent variable model distribution $P_G$. We show that the OT problem can be equivalently written in terms of probabilistic encoders, which are constrained to match the posterior and prior distributions...
Generative Adversarial Networks (GAN) (Goodfellow et al., 2014) are an effective method for training generative models of complex data such as natural images. However, they are notoriously hard to train and can suffer from the problem of missing modes where the model is not able to produce examples in certain regions of the space. We propose an ite...
A server may identify a first address stored in a first search index; determine one or more first identifiers associated with the first address; identify a second address stored in a second search index; determine one or more second identifiers associated with the second address; map the first address to the second address based on a first identifi...
Methods, systems, and articles of manufacture for detecting placeholder images are disclosed. These include, accessing a collection of digital images, clustering the digital images to generate at least one of a plurality of exact-duplicate image clusters and a plurality of near-duplicate image clusters, and selecting one or more placeholder image c...
We shed light on the discrimination between patterns belonging to two different classes by casting this decoding problem into a generalized prototype framework. The discrimination process is then separated into two stages: a projection stage that reduces the dimensionality of the data by projecting it on a line and a threshold stage where the distr...
We shed light on the discrimination between patterns belonging to two different classes by casting this decoding problem into a generalized prototype framework. The discrimination process is then separated into two stages: a projection stage that reduces the dimensionality of the data by projecting it on a line and a threshold stage where the distr...
Consistency is a key property of all statistical procedures analyzing randomly sampled data. Surprisingly, despite decades of work, little is known about consistency of most clustering algorithms. In this paper we investigate consistency of the popular family of spectral clustering algorithms, which clusters the data with the help of eigenvectors o...
The support vector machine (SVM) algorithm is well known to the computer learning community for its very good practical results. The goal of the present paper is to study this algorithm from a statistical perspective, using tools of concentration theory and empirical processes. Our main result builds on the observation made by other authors that th...
This contribution develops a theoretical framework that takes into ac- count the effect of approximate optimization on learning algorithms. The anal- ysis shows distinct tradeoffs for the case of small-scale and large-scale learning problems. Small-scale learning problems are subject to the usual approximation- estimation tradeoff. Large-scale lear...
There exist many different generalization error bounds in statistical learning theory. Each of these bounds contains an improvement over the others for certain situations or algorithms. Our goal is, first, to underline the links between these bounds, and second, to combine the different improvements into a single bound. In particular we combine the...
This statistical learning theory frameworks that are used when the examples are not provided sequentially are discussed. A new PAC Bound framework for Intersection-Closed Concepts Classes is introduced that will provide a number of examples that are needed for an algorithm to achieve a given accuracy in its prediction. The framework provide an impr...
The main goal of this paper is to prove inequalities on the reconstruction error for kernel principal component analysis. With respect to previous work on this topic, our contribution is twofold: (1) we give bounds that explicitly take into account the empirical centering step in this algorithm, and (2) we show that a “localized” approach allows to...
Comment on ``Support Vector Machines with Applications'' [math.ST/0612817]
This contribution develops a theoretical framework that takes into account the effect of approximate optimization on learning algorithms. The analysis shows distinct tradeoffs for the case of small-scale and large-sc ale learning problems. Small-scale learning problems are subject to the usual approximation-estimation tradeoff. Large-scale learning...
This paper is about the evaluation of the results of clustering algorithms, and the comparison of such algorithms. We propose a new method based on the enrichment of a set of independent labeled datasets by the results of clustering, and the use of a supervised method to evaluate the interest of adding such new information to the datasets.
We thus...
This Chapter presents the PASCAL Evaluating Predictive Uncertainty Challenge, introduces the contributed Chapters by the participants
who obtained outstanding results, and provides a discussion with some lessons to be learnt. The Challenge was set up to evaluate
the ability of Machine Learning algorithms to provide good “probabilistic predictions”,...
Cet article se place dans le cadre de l'évaluation des résultats d'algorithmes de clustering et de la comparaison de tels algorithmes. Nous proposons une nouvelle méthode basée sur l'enrichissement d'un ensemble de jeux de données étiquetés indépendants par les résultats des algorithmes de clustering considérés, et sur l'utilisation d'un algorithme...
Subspace clustering is an extension of traditional clustering that seeks to find clusters embedded in different subspaces within a dataset. This is a particularly important challenge with high dimensional data where the curse of dimensionality occurs. It also has the benefit of providing smaller descriptions of the clusters found. In this field, we sh...
We introduce two new functionals, the constrained covariance and the kernel mutual information, to measure the degree of independence of random variables. These quantities are both based on the covariance between functions of the random variables in reproducing kernel Hilbert spaces (RKHSs). We prove that when the RKHSs are universal, both function...
We propose an independence criterion based on the eigenspectrum of covariance operators in reproducing kernel Hilbert spaces (RKHSs), consisting of an empirical estimate of the Hilbert-Schmidt norm of the cross-covariance operator (we term this a Hilbert-Schmidt Independence Criterion, or HSIC). This approach has several advantages, compared with p...
In this article we construct a maximal margin classification algorithm for arbitrary metric spaces. At first we show that the Support Vector Machine (SVM) is a maximal margin algorithm for the class of metric spaces where the negative squared distance is conditionally posi- tive definite (CPD). This means that the metric space can be isometri- call...
We propose new bounds on the error of learning algorithms in terms of a data-dependent notion of complexity. The estimates we establish give optimal rates and are based on a local and empirical version of Rademacher averages, in the sense that the Rademacher averages are computed from the data, on a subset of functions with small empirical error. W...
Subspace clustering is an extension of traditional clustering that seeks to find clusters in different subspaces within a dataset. This is a particularly important challenge with high dimensional data where the curse of dimensionality occurs. It has also the benefit of providing smaller descriptions of the clusters found.
Existing methods only cons...
We develop a methodology for solving high dimensional dependency estimation problems between pairs of data types, which is viable in the case where the output of interest has very high dimension, e.g., thousands of dimensions. This is achieved by mapping the objects into continuous or discrete spaces, using joint kernels. Known correlations between...
The last few years have witnessed important new developments in
the theory and practice of pattern classification. We intend to
survey some of the main new ideas that have led to these
recent results.
A general method for obtaining moment inequalities for functions of independent random variables is presented. It is a generalization of the entropy method which has been used to derive concentration inequalities for such functions [Boucheron, Lugosi and Massart Ann. Probab. 31 (2003) 1583-1614], and is based on a generalized tensorization inequali...
A general method for obtaining moment inequalities for functions of independent random variables is presented. It is a generalization of the entropy method which has been used to derive concentration inequalities for such functions [Boucheron, Lugosi and Massart Ann. Probab. 31 (2003) 1583–1614], and is based on a generalized tensorization inequali...
We introduce two new functionals, the constrained covariance and the kernel mutual information, to measure the degree of independence of random variables. These quantities are both based on the covariance between functions of the random variables in reproducing kernel Hilbert spaces (RKHSs). We prove that when the RKHSs are universal, both function...
We develop a methodology for solving high dimensional dependency estimation problems between pairs of data types, which is viable in the case where the output of interest has very high dimension, e.g., thousands of dimensions. This is achieved by mapping the objects into continuous or discrete spaces, using joint kernels. Known correlations between...
The last few years have witnessed important new developments in the theory and practice of pattern classification. We intend to survey some of the main new ideas that have lead to these important recent developments.
In this paper we investigate connections between statistical learning theory and data compression on the basis of support vector machine (SVM) model selection. Inspired by several generalization bounds we construct "compression coefficients" for SVMs which measure the amount by which the training labels can be compressed by a code built from the se...
We give a basic introduction to Gaussian Process regression models. We focus on understanding the role of the stochastic process and how it is used to define a distribution over functions. We present the simple equations for incorporating training data and examine how to learn the hyperparameters using the marginal likelihood. We explain the practi...
We study the properties of the eigenvalues of Gram matrices in a non-asymptotic setting. Using local Rademacher averages, we provide data-dependent and tight bounds for their convergence towards eigenvalues of the corresponding kernel operator. We perform these computations in a functional analytic framework which allows to deal implicitly with rep...
The goal of this article is to investigate the field of Hilbertian metrics on probability measures. Since they are very versatile and can therefore be applied in various problems they are of great interest in kernel methods. Quit recently Tops{o}e and Fuglede introduced a family of Hilbertian metrics on probability measures. We give basic propertie...
This paper gives a survey of results in the mathematical literature on positive definite kernels and their associated structures. We concentrate on properties which seem potentially relevant for Machine Learning and try to clarify some results that have been misused in the literature. Moreover we consider different lines of generalizations of posit...
Given a set of n randomly drawn sample points, spectral clustering in its simplest form uses the second eigenvector of the graph Laplacian matrix, constructed on the similarity graph between the sam- ple points, to obtain a partition of the sample. We are interested in the question how spectral clustering behaves for growing sample size n. In case...
We investigate measures of complexity of function classes based on continuity moduli of Gaussian and Rademacher processes. For Gaussian processes, we obtain bounds on the continuity modulus on the convex hull of a function class in terms of the same quantity for the class itself. We also obtain new bounds on generalization error in terms of localiz...
The notion of kernels, recently introduced, has drawn much interest as it allows one to obtain nonlinear algorithms from linear ones in a simple and elegant manner. This, in conjunction with the introduction of new linear classification methods such as the support vector machines (SVMs), has produced significant progress in machine learning and rel...
The Google search engine has enjoyed huge success with its web page ranking algorithm, which exploits global, rather than local, hyperlink structure of the web using random walks. Here we propose a simple universal ranking algorithm for data lying in the Euclidean space, such as text or image data. The core idea of our method is to rank the data wi...
We address in this paper the question of how the knowledge of the marginal distribution P (x) can be incorporated in a learning algorithm. We suggest three theoretical methods for taking into account this distribution for regularization and provide links to existing graph-based semi-supervised learning algorithms. We also propose practical implemen...
We consider the general problem of learning from labeled and unlabeled data, which is often called semi-supervised learning or transductive inference. A principled approach to semi-supervised learning is to design a classifying function which is sufficiently smooth with respect to the intrinsic structure collectively revealed by known labeled and u...
A general method for obtaining moment inequalities for functions of independent random variables is presented. It is a generalization of the entropy method which has been used to derive concentration inequalities for such functions [7], and is based on a generalized tensorization inequality due to Latala and Oleszkiewicz [25]. The new inequalities...
this paper is to provide such general-purpose inequalities. Our approach is based on a generalization of Ledoux's entropy method (see [26, 28]). Ledoux's method relies on abstract functional inequalities known as logarithmic Sobolev inequalities and provide a powerful tool for deriving exponential inequalities for functions of independent random va...
Concentration inequalities deal with deviations of functions of independent random variables from their expectation. In the last decade new tools have been introduced making it possible to establish simple and powerful inequalities. These inequalities are at the heart of the mathematical analysis of various problems in machine learning and made it...
An important aspect of clustering algorithms is whether the partitions constructed on finite samples converge to a useful clustering of the whole data space as the sample size increases. This paper investigates this question for normalized and unnormalized versions of the popular spec- tral clustering algorithm. Surprisingly, the convergence of unn...
The goal of this article is to develop a framework for large margin classification in metric spaces. We want to find a generalization of linear decision functions for metric spaces and define a corresponding notion of margin such that the decision function separates the training points with a large margin. It will turn out that using Lipschitz func...
We study the properties of the eigenvalues of Gram matrices in a non-asymptotic setting. Using local Rademacher averages, we provide data-dependent and tight bounds for their convergence towards eigenvalues of the corresponding kernel operator. We perform these computations in a functional analytic framework which allows to deal implicitly with rep...
We consider the general problem of learning from labeled and unlabeled data. Given a set of points, some of them are labeled, and the remaining points are unlabeled. The goal is to predict the labels of the unlabeled points. Any supervised learning algorithm can be applied to this problem, for instances, Support Vector Machines (SVMs). The problem...
The goal of this article is to develop a framework for large margin classification in metric spaces. We want to find a generalization of linear decision functions for metric spaces and define a corresponding notion of margin such that the decision function separates the training points with a large margin. It will turn out that using Lipschitz func...
We investigate data based procedures for selecting the kernel when learning with Support Vector Machines. We provide generalization error bounds by estimating the Rademacher complexities of the corresponding function classes. In particular we obtain a complexity bound for function classes induced by kernels with given eigenvectors, i.e., we allow t...
Recently introduced in machine learning, the notion of kernels has drawn a lot of interest as it allows nonlinear algorithms to be obtained from linear ones in a simple and elegant manner. This, in conjunction with the introduction of new linear classification methods such as the support vector machines has produced significant progress. The succes...
Motivation:
In drug discovery a key task is to identify characteristics that separate active (binding) compounds from inactive (non-binding) ones. An automated prediction system can help reduce resources necessary to carry out this task.
Results:
Two methods for prediction of molecular bioactivity for drug design are introduced and shown to perf...
We present new tools from probability theory that can be applied to the analysis of learning algorithms. These tools allow to derive new bounds on the generalization performance of learning algorithms and to propose alternative measures of the complexity of the learning task, which in turn can be used to derive new learning algorithms.
We propose new bounds on the error of learning algorithms in terms of a data-dependent notion of complexity. The estimates we establish give optimal rates and are based on a local and empirical version of Rademacher averages, in the sense that the Rademacher averages are computed from the data, on a subset of functions with small empirical error. W...
The Google search engine has enjoyed huge success with its web page ranking algorithm, which exploits global, rather than local, hyperlink structure of the web using random walks. Here we propose a simple universal ranking algorithm for data lying in the Euclidean space, such as text or image data. The core idea of our method is to rank the data wi...
The goal of this article is to develop a framework for large margin classification in metric spaces. We want to find a generalization of linear decision functions for metric spaces and define a corresponding notion of margin such that the decision function separates the training points with a large margin. It will turn out that using Lipschitz func...
Concentration inequalities deal with deviations of functions of independent random variables from their expectation. In the last decade new tools have been introduced making it possible to establish simple and powerful inequalities. These inequalities are at the heart of the mathematical analysis of various problems in machine learning and made it...
In this short note, building on ideas of M. Herbster [2] we propose a method for automatically tuning the parameter of the FIXED-SHARE algorithm proposed by Herbster and Warmuth [3] in the context of on-line learning with shifting experts. We show that this can be done with a memory requirement of O(nT) and that the additional loss incurred by the...
We obtain exponential concentration inequalities for sub-additive functions of independent random variables under weak conditions on the increments of those functions, like the existence of exponential moments for these increments. As a consequence of these general inequalities, we obtain refinements of Talagrand’s inequality for empirical processe...
The goal of statistical learning theory is to study, in a sta- tistical framework, the properties of learning algorithms. In particular, most results take the form of so-called error bounds. This tutorial intro- duces the techniques that are used to obtain such results.
In order to apply the maximum margin method in arbitrary metric spaces, we suggest to embed the metric space into a Banach or Hilbert space and to perform linear classification in this space. We propose several embeddings and recall that an isometric embedding in a Banach space is always possible while an isometric embedding in a Hilbert space is o...
In this paper, we examine on-line learning problems in which the target concept is allowed to change over time. In each trial a master algorithm receives predictions from a large set of n experts. Its goal is to predict almost as well as the best sequence of such experts chosen o#-line by partitioning the training sequence into k + 1 sections and t...
We investigate measures of complexity of function classes based on continuity moduli of Gaussian and Rademacher processes.
For Gaussian processes, we obtain bounds on the continuity modulus on the convex hull of a function class in terms of the
same quantity for the class itself. We also obtain new bounds on generalization error in terms of localiz...
We define notions of stability for learning algorithms and derive bounds on their generalization error based on the empirical error and the leave-one-out error. We then study the stability properties of large classes of learning algorithms such as regularization based algorithms. In particular we focus on Hilbert space regularization and Kullback-L...
We de ne notions of stability for learning algorithms and show how to use these notions to derive generalization error bounds based on the empirical error and the leave-one-out error. The methods we use can be applied in the regression framework as well as in the classi cation one when the classi er is obtained by thresholding a real-valued functio...
In this article we investigate the behaviour of the global and local Rademacher averages. We present new error bounds which are based on the localized averages and indicate how data-dependent localized averages can be estimated without a priori knowledge of the class at hand.
We introduce new concentration inequalities for functions on product spaces They allow to obtain a Bennett type deviation bound for suprema of empirical processes indexed by upper bounded functions. The result is an improvement on Rio's version [6] of Talagrand's inequality [7] for equidistributed variables.