AISTATS 2017 Papers

In this paper we obtain sufficient and necessary conditions on the number of samples required for exact recovery of the pure-strategy Nash equilibria (PSNE) set of a graphical game from noisy observations of joint actions. We consider sparse linear influence games — a parametric class of graphical games with linear payoffs, and represented by directed graphs of n nodes (players) and in-degree of at most k. We show that one can efficiently recover the PSNE set of a linear influence game with $O(k^2 \log n)$ samples, under very general observation models. On the other hand, we show that $Ω(k \log n)$ samples are necessary for any procedure to recover the PSNE set from observations of joint actions.

Decision makers, such as doctors and judges, make crucial decisions such as recommending treatments to patients, and granting bails to defendants on a daily basis. Such decisions typically involve weighting the potential benefits of taking an action against the costs involved. In this work, we aim to automate this task of learning cost-effective, interpretable and actionable treatment regimes. We formulate this as a problem of learning a decision list – a sequence of if-then-else rules – which maps characteristics of subjects (eg., diagnostic test results of patients) to treatments. We propose a novel objective to construct a decision list which maximizes outcomes for the population, and minimizes overall costs. Since we do not observe the outcomes corresponding to counterfactual scenarios, we use techniques from causal inference literature to infer them. We model the problem of learning the decision list as a Markov Decision Process (MDP) and employ a variant of the Upper Confidence Bound for Trees (UCT) strategy which leverages customized checks for pruning the search space effectively. Experimental results on real world observational data capturing judicial bail decisions and treatment recommendations for asthma patients demonstrate the effectiveness of our approach.

Learning a computationally efficient kernel from data is an important machine learning problem. The majority of kernels in the literature do not leverage the geometry of the data, and those that do are computationally infeasible for contemporary datasets. Recent advances in approximation techniques have expanded the applicability of the kernel methodology to scale linearly with the data size. Data-dependent kernels, which could leverage this computational advantage, have however not yet seen the benefit. Here we derive an approximate large-scale learning procedure for data-dependent kernels that is efficient and performs well in practice. We provide a Lemma that can be used to derive the asymptotic convergence of the approximation in the limit of infinite random features, and, under certain conditions, an estimate of the convergence speed. We empirically prove that our construction represents a valid, yet efficient approximation of the data-dependent kernel. For large-scale datasets of millions of datapoints, where the proposed method is now applicable for the first time, we notice a significant performance boost over both baselines consisting of data independent kernels and of kernel approximations, at comparable computational cost.

When assigning labels to a test instance, most multilabel and multiclass classifiers systematically evaluate every single label to decide whether it is relevant or not. This linear scan over labels becomes prohibitive when the number of labels is very large. To alleviate this problem we propose a two step approach where computationally efficient label filters pre-select a small set of candidate labels before the base multiclass or multilabel classifier is applied. The label filters select candidate labels by projecting a test instance on a filtering line, and retaining only the labels that have training instances in the vicinity of this projection. The filter parameters are learned directly from data by solving a constraint optimization problem, and are independent of the base multilabel classifier. The proposed label filters can be used in conjunction with any multiclass or multilabel classifier that requires a linear scan over the labels, and speed up prediction by orders of magnitude without significant impact on performance.

We consider the problem of consistently matching multiple sets of elements to each other, which is a common task in fields such as computer vision. To solve the underlying NP-hard objective, existing methods often relax or approximate it, but end up with unsatisfying empirical performance due to a misaligned objective. We propose a coordinate update algorithm that directly optimizes the target objective. By using pairwise alignment information to build an undirected graph and initializing the permutation matrices along the edges of its Maximum Spanning Tree, our algorithm successfully avoids bad local optima. Theoretically, with high probability our algorithm guarantees an optimal solution under reasonable noise assumptions. Empirically, our algorithm consistently and significantly outperforms existing methods on several benchmark tasks on real datasets.

In this paper, we study the information-theoretic limits of learning the structure of Bayesian networks (BNs), on discrete as well as continuous random variables, from a finite number of samples. We show that the minimum number of samples required by any procedure to recover the correct structure grows as $Ω(m)$ and $Ω(k \log m + (k^2)/m)$ for non-sparse and sparse BNs respectively, where m is the number of variables and k is the maximum number of parents per node. We provide a simple recipe, based on an extension of the Fano’s inequality, to obtain information-theoretic limits of structure recovery for any exponential family BN. We instantiate our result for specific conditional distributions in the exponential family to characterize the fundamental limits of learning various commonly used BNs, such as conditional probability table based networks, Gaussian BNs, noisy-OR networks, and logistic regression networks. En route to obtaining our main results, we obtain tight bounds on the number of sparse and non-sparse essential-DAGs. Finally, as a byproduct, we recover the information-theoretic limits of sparse variable selection for logistic regression.

Approximate inference via information projection has been recently introduced as a general-purpose technique for efficient probabilistic inference given sparse variables. This manuscript goes beyond classical sparsity by proposing efficient algorithms for approximate inference via information projection that are applicable to any structure on the set of variables that admits enumeration using matroid or knapsack constraints. Further, leveraging recent advances in submodular optimization, we provide an efficient greedy algorithm with strong optimization-theoretic guarantees. The class of probabilistic models that can be expressed in this way is quite broad and, as we show, includes group sparse regression, group sparse principal components analysis and sparse collective matrix factorization, among others. Empirical results on simulated data and high dimensional neuroimaging data highlight the superior performance of the information projection approach as compared to established baselines for a range of probabilistic models.

We introduce a method for using deep neural networks to amortize the cost of inference in models from the family induced by universal probabilistic programming languages, establishing a framework that combines the strengths of probabilistic programming and deep learning methods. We call what we do “compilation of inference” because our method transforms a denotational specification of an inference problem in the form of a probabilistic program written in a universal programming language into a trained neural network denoted in a neural network specification language. When at test time this neural network is fed observational data and executed, it performs approximate inference in the original model specified by the probabilistic program. Our training objective and learning procedure are designed to allow the trained neural network to be used as a proposal distribution in a sequential importance sampling inference engine. We illustrate our method on mixture models and Captcha solving and show significant speedups in the efficiency of inference.

This paper describes a new parameter-free online learning algorithm for changing environments. In comparing against algorithms with the same time complexity as ours, we obtain a strongly adaptive regret bound that is a factor of at least $\sqrt\log(T)$ better, where $T$ is the time horizon. Empirical results show that our algorithm outperforms state-of-the-art methods in learning with expert advice and metric learning scenarios.

The failure of LASSO to identify groups of correlated predictors in linear regression has sparked significant research interest. Recently, various norms were proposed, which can be best described as instances of ordered weighted $\ell_1$ norms (OWL), as an alternative to $\ell_1$ regularization used in LASSO. OWL can identify groups of correlated variables but it forces the model to be constant within a group. This artifact induces unnecessary bias in the model estimation. In this paper we take a submodular perspective and show that OWL can be posed as the Lovász extension of a suitably defined submodular function. The submodular perspective not only explains the group-wise constant behavior of OWL, but also suggests alternatives. The main contribution of this paper is smoothed OWL (SOWL), a new family of norms, which not only identifies the groups but also allows the model to be flexible inside a group. We establish several algorithmic and theoretical properties of SOWL including group identification and model consistency. We also provide algorithmic tools to compute the SOWL norm and its proximal operator, whose computational complexity $O(d\log d)$ is significantly better than that of general purpose solvers in $O(d^2\log d)$. In our experiments, SOWL compares favorably with respect to OWL in the regimes of interest.

The gang of bandits (GOB) model [7] is a recent contextual bandits framework that shares information between a set of bandit problems, related by a known (possibly noisy) graph. This model is useful in problems like recommender systems where the large number of users makes it vital to transfer information between users. Despite its effectiveness, the existing GOB model can only be applied to small problems due to its quadratic time-dependence on the number of nodes. Existing solutions to combat the scalability issue require an often-unrealistic clustering assumption. By exploiting a connection to Gaussian Markov random fields (GMRFs), we show that the GOB model can be made to scale to much larger graphs without additional assumptions. In addition, we propose a Thompson sampling algorithm which uses the recent GMRF sampling-by-perturbation technique, allowing it to scale to even larger problems (leading to a “horde” of bandits). We give regret bounds and experimental results for GOB with Thompson sampling and epoch-greedy algorithms, indicating that these methods are as good as or significantly better than ignoring the graph or adopting a clustering-based approach. Finally, when an existing graph is not available, we propose a heuristic for learning it on the fly and show promising results.

We propose the Hit-and-Run algorithm for planning and sampling problems in non- convex spaces. For sampling, we show the first analysis of the Hit-and-Run algorithm in non-convex spaces and show that it mixes fast as long as certain smoothness conditions are satisfied. In particular, our analysis reveals an intriguing connection between fast mixing and the existence of smooth measure-preserving mappings from a convex space to the non-convex space. For planning, we show advantages of Hit-and- Run compared to state-of-the-art planning methods such as Rapidly-Exploring Random Trees.

The key to time series clustering is how to characterize the similarity between any two time series. In this paper, we explore a new similarity metric called “cross-predictability”: the degree to which a future value in each time series is predicted by past values of the others. However, it is challenging to estimate such cross-predictability among time series in the high-dimensional regime, where the number of time series is much larger than the length of each time series. We address this challenge with a sparsity assumption: only time series in the same cluster have significant cross-predictability with each other. We demonstrate that this approach is computationally attractive, and provide a theoretical proof that the proposed algorithm will identify the correct clustering structure with high probability under certain conditions. To the best of our knowledge, this is the first practical high-dimensional time series clustering algorithm with a provable guarantee. We evaluate with experiments on both synthetic data and real-world data, and results indicate that our method can achieve more than 80% clustering accuracy on real-world data, which is 20% higher than the state-of-art baselines.

The Gaussian process (GP) is a simple yet powerful probabilistic framework for various machine learning tasks. However, exact algorithms for learning and prediction are prohibitive to be applied to large datasets due to inherent computational complexity. To overcome this main limitation, various techniques have been proposed, and in particular, local GP algorithms that scales “truly linearly” with respect to the dataset size. In this paper, we introduce a hierarchical model based on local GP for large-scale datasets, which stacks inducing points over inducing points in layers. By using different kernels in each layer, the overall model becomes multi-scale and is able to capture both long- and short-range dependencies. We demonstrate the effectiveness of our model by speed-accuracy performance on challenging real-world datasets.

Submodular continuous functions are a category of (generally) non-convex/non-concave functions with a wide spectrum of applications. We characterize these functions and demonstrate that they can be maximized efficiently with approximation guarantees. Specifically, i) We introduce the weak DR property that gives a unified characterization of submodularity for all set, integer-lattice and continuous functions; ii) for maximizing monotone DR-submodular continuous functions under general down-closed convex constraints, we propose a Frank-Wolfe variant with (1-1/e) approximation guarantee, and sub-linear convergence rate; iii) for maximizing general non-monotone submodular continuous functions subject to box constraints, we propose a DoubleGreedy algorithm with 1/3 approximation guarantee. Submodular continuous functions naturally find applications in various real-world settings, including influence and revenue maximization with continuous assignments, sensor energy management, facility location, etc. Experimental results show that the proposed algorithms efficiently generate superior solutions compared to baseline algorithms.

Maximum-a-Posteriori (MAP) inference lies at the heart of Graphical Models and Structured Prediction. Despite the intractability of exact MAP inference, approximated methods based on LP relaxations have exhibited superior performance across a wide range of applications. Yet for problems involving large output domains (i.e., the state space for each variable is large), standard LP relaxations can easily give rise to a large number of variables and constraints which are beyond the limit of existing optimization algorithms. In this paper, we introduce an effective MAP inference method for problems with large output domains. The method builds upon alternating minimization of an Augmented Lagrangian that exploits the sparsity of messages through greedy optimization techniques. A key feature of our greedy approach is to introduce variables in an on-demand manner with a pre-built data structure over local factors. This results in a single-loop algorithm of sublinear cost per iteration and O(log(1/epsilon))-type iteration complexity to achieve epsilon sub-optimality. In addition, we introduce a variant of GDMM for binary MAP inference problems with a large number of factors. Empirically, the proposed algorithms demonstrate orders of magnitude speedup over state-of-the-art MAP inference techniques on MAP inference problems including Segmentation, Alignment, Protein Folding, Graph Matching, and Pairwise-Interacted Multilabel Prediction.

We develop an automated variational inference method for Bayesian structured prediction problems with Gaussian process (GP) priors and linear-chain likelihoods. Our approach does not need to know the details of the structured likelihood model and can scale up to a large number of observations. Furthermore, we show that the required expected likelihood term and its gradients in the variational objective (ELBO) can be estimated efficiently by using expectations over very low-dimensional Gaussian distributions. Optimization of the ELBO is fully parallelizable over sequences and amenable to stochastic optimization, which we use along with control variate techniques to make our framework useful in practice. Results on a set of natural language processing tasks show that our method can be as good as (and sometimes better than, in particular with respect to expected log-likelihood) hard-coded approaches including SVM-struct and CRF, and overcomes the scalability limitations of previous inference algorithms based on sampling. Overall, this is a fundamental step to developing automated inference methods for Bayesian structured prediction.

Boosting is a popular ensemble algorithm that generates more powerful learners by linearly combining base models from a simpler hypothesis class. In this work, we investigate the problem of adapting batch gradient boosting for minimizing convex loss functions to online setting where the loss at each iteration is i.i.d sampled from an unknown distribution. To generalize from batch to online, we first introduce the definition of online weak learning edge with which for strongly convex and smooth loss functions, we present an algorithm, Streaming Gradient Boosting (SGB) with exponential shrinkage guarantees in the number of weak learners. We further present an adaptation of SGB to optimize non-smooth loss functions, for which we derive a $O(\ln(N)/N)$ convergence rate. We also show that our analysis can extend to adversarial online learning setting under a stronger assumption that the online weak learning edge will hold in adversarial setting. We finally demonstrate experimental results showing that in practice our algorithms can achieve competitive results as classic gradient boosting while using less computation.

While there has been a significant amount of work studying gradient descent techniques for non-convex optimization problems over the last few years, all existing results establish either local convergence with good rates or global convergence with highly suboptimal rates, for many problems of interest. In this paper, we take the first step in getting the best of both worlds – establishing global convergence and obtaining a good rate of convergence for the problem of computing squareroot of a positive semidefinite (PSD) matrix, which is a widely studied problem in numerical linear algebra with applications in machine learning and statistics among others. Given a PSD matrix M and a PSD starting point $U_0$, we show that gradient descent with appropriately chosen step-size finds an epsilon-accurate squareroot of M in $O(α\log(||M-U_0||_F^2 / ε))$ iterations, where $α= (\max{||U_0||^2, ||M||} / \min{\sigma_min^2(U_0), \sigma_min(M)} )^3/2$. Our result is the first to establish global convergence for this problem and that it is robust to errors in each iteration. A key contribution of our work is the general proof technique which we believe should further excite research in understanding deterministic and stochastic variants of simple non-convex gradient descent algorithms with good global convergence rates for other problems in machine learning and numerical linear algebra.