We propose a new approach for metric learning by framing it as learning a sparse combination of locally discriminative metrics that are inexpensive to generate from the training data. This flexible framework allows us to naturally derive formulations for global, multi-task and local metric learning. The resulting algorithms have several advantages over existing methods in the literature: a much smaller number of parameters to be estimated and a principled way to generalize learned metrics to new testing data points. To analyze the approach theoretically, we derive a generalization bound that justifies the sparse combination. Empirically, we evaluate our algorithms on several datasets against state-of-the-art metric learning methods. The results are consistent with our theoretical findings and demonstrate the superiority of our approach in terms of classification performance and scalability.

Dropout and other feature noising schemes have shown promising results in controlling over-fitting by artificially corrupting the training data. Though extensive theoretical and empirical studies have been performed for generalized linear models, little work has been done for support vector machines (SVMs), one of the most successful approaches for supervised learning. This paper presents dropout training for linear SVMs. To deal with the intractable expectation of the non-smooth hinge loss under corrupting distributions, we develop an iteratively re-weighted least square (IRLS) algorithm by exploring data augmentation techniques. Our algorithm iteratively minimizes the expectation of a re-weighted least square problem, where the re-weights have closed-form solutions. The similar ideas are applied to develop a new IRLS algorithm for the expected logistic loss under corrupting distributions. Our algorithms offer insights on the connection and difference between the hinge loss and logistic loss in dropout training. Empirical results on several real datasets demonstrate the effectiveness of dropout training on significantly boosting the classification accuracy of linear SVMs.

During the past decade, machine learning algorithms have become commonplace in large-scale real-world industrial applications. In these settings, the computation time to train and test machine learning algorithms is a key consideration. At training-time the algorithms must scale to very large data set sizes.At testing-time, the cost of feature extraction can dominate the CPU runtime. Recently, a promising method was proposed to account for the feature extraction cost at testing time, called Cost-sensitive Tree of Classifiers (CSTC). Although the CSTC problem is NP-hard, the authors suggest an approximation through a mixed-norm relaxation across many classifiers. This relaxation is slow to train and requires involved optimization hyperparameter tuning. We propose a different relaxation using approximate submodularity, called Approximately Submodular Tree of Classifiers (ASTC). ASTC is much simpler to implement, yields equivalent results but requires no optimization hyperparameter tuning and is up to two orders of magnitude faster to train.

Sparse modeling has been highly successful in many real-world applications. While a lot of interests have been on convex regularization, recent studies show that nonconvexregularizers can outperform their convex counterparts in many situations.However, the resulting nonconvex optimization problems are often challenging, especiallyfor composite regularizers such as the nonconvex overlapping group lasso. In thispaper, byusing a recent mathematical tool known as the proximal average,we propose a novel proximal gradient descent method for optimization with a wide class of nonconvex and composite regularizers.Instead of directlysolving the proximal stepassociated with a composite regularizer, we average thesolutions from the proximal problems of the constituent regularizers. This simple strategy has guaranteed convergenceand low per-iteration complexity.Experimental results on a number of synthetic andreal-world data sets demonstrate the effectiveness and efficiency of theproposed optimization algorithm, and also the improved classification performanceresulting from thenonconvex regularizers.

A Chinese character embedded in different compound words may carry different meanings. In this paper, we aim at semantical clustering of a given family of morphologically related Chinese words. In Experiment 1, we employed linguistic features at the word, syntactic, semantic, and contextual levels in aggregated computational linguistics methods to handle the clustering task. In Experiment 2, we recruited adults and children to perform the clustering task. Experimental results indicate that our computational model achieved a similar level of performance as children.

Transfer learning considers related but distinct tasks defined on heterogenous domains and tries to transfer knowledge between these tasks to improve generalization performance. It is particularly useful when we do not have sufficient amount of labeled training data in some tasks, which may be very costly, laborious, or even infeasible to obtain. Instead, learning the tasks jointly enables us to effectively increase the amount of labeled training data. In this paper, we formulate a kernelized Bayesian transfer learning framework that is a principled combination of kernel-based dimensionality reduction models with task-specific projection matrices to find a shared subspace and a coupled classification model for all of the tasks in this subspace. Our two main contributions are: (i) two novel probabilistic models for binary and multiclass classification, and (ii) very efficient variational approximation procedures for these models. We illustrate the generalization performance of our algorithms on two different applications. In computer vision experiments, our method outperforms the state-of-the-art algorithms on nine out of 12 benchmark supervised domain adaptation experiments defined on two object recognition data sets. In cancer biology experiments, we use our algorithm to predict mutation status of important cancer genes from gene expression profiles using two distinct cancer populations, namely, patient-derived primary tumor data and in-vitro-derived cancer cell line data. We show that we can increase our generalization performance on primary tumors using cell lines as an auxiliary data source.

We present a novel approach to low-dimensional neighbor embedding for visualization, based on formulating an information retrieval based neighborhood preservation cost function as Maximum satisfiability on a discretized output display. The method has a rigorous interpretation as optimal visualization based on the cost function. Unlike previous low-dimensional neighbor embedding methods, our formulation is guaranteed to yield globally optimal visualizations, and does so reasonably fast. Unlike previous manifold learning methods yielding global optima of their cost functions, our cost function and method are designed for low-dimensional visualization where evaluation and minimization of visualization errors are crucial. Our method performs well in experiments, yielding clean embeddings of datasets where a state-of-the-art comparison method yields poor arrangements. In a real-world case study for semi-supervised WLAN signal mapping in buildings we outperform state-of-the-art methods.

Predicting human activities is important for improving recommender systems or analyzing social relationships among users. Those human activities are usually repre- sented as multi-object relationships (e.g. user’s tagging activities for items or user’s tweeting activities at some locations). Since multi-object relationships are naturally represented as a tensor, tensor factorization is becom- ing more important for predicting users’ possible ac- tivities. However, its prediction accuracy is weak for ambiguous and/or sparsely observed objects. Our so- lution, Semantic data Representation for Tensor Fac- torization (SRTF), tackles these problems by incorpo- rating semantics into tensor factorization based on the following ideas: (1) It first links objects to vocabu- laries/taxonomies and resolves the ambiguity caused by objects that can be used for multiple purposes. (2) It next links objects to composite classes that merge classes in different kinds of vocabularies/taxonomies (e.g. classes in vocabularies for movie genres and those for directors) to avoid low prediction accuracy caused by rough-grained semantics. (3) It then lifts sparsely observed objects into their classes to solve the sparsity problem for rarely observed objects. To the best of our knowledge, this is the first study that leverages seman- tics to inject expert knowledge into tensor factorization. Experiments show that SRTF achieves up to 10% higher accuracy than state-of-the-art methods.

Imitation Learning (IL) is a popular approach for teaching behavior policies to agents by demonstrating the desired target policy. While the approach has lead to many successes, IL often requires a large set of demonstrations to achieve robust learning, which can be expensive for the teacher. In this paper, we consider a novel approach to improve the learning efficiency of IL by providing a shaping reward function in addition to the usual demonstrations. Shaping rewards are numeric functions of states (and possibly actions) that are generally easily specified, and capture general principles of desired behavior, without necessarily completely specifying the behavior. Shaping rewards have been used extensively in reinforcement learning, but have been seldom considered for IL, though they are often easy to specify. Our main contribution is to propose an IL approach that learns from both shaping rewards and demonstrations. We demonstrate the effectiveness of the approach across several IL problems, even when the shaping reward is not fully consistent with the demonstrations.

As robots become more ubiquitous, it is increasingly important for untrained users to be able to interact with them intuitively. In this work, we investigate how people refer to objects in the world during relatively unstructured communication with robots. We collect a corpus of deictic interactions from users describing objects, which we use to train language and gesture models that allow our robot to determine what objects are being indicated. We introduce a temporal extension to state-of-the-art hierarchical matching pursuit features to support gesture understanding, and demonstrate that combining multiple communication modalities more effectively captures user intent than relying on a single type of input. Finally, we present initial interactions with a robot that uses the learned models to follow commands while continuing to learn from user input.