Planning with sensing actions under partial observability is a computationally challenging problem that is fundamental to the realization of AI tasks in areas as diverse as robotics, game playing, and diagnostic problem solving. Recent work on generating plans for partially observable domains has advocated for online planning, claiming that offline plans are often too large to generate. Here we push the envelope on this challenging problem, proposing a technique for generating conditional (aka contingent) plans offline. The key to our planner's success is the reliance on state-of-the-art techniques for fully observable non-deterministic (FOND) planning. In particular, we use an existing compilation for converting a planning problem under partial observability and sensing to a FOND planning problem. With a modified FOND planner in hand, we are able to scale beyond previous techniques for generating conditional plans with solutions that are orders of magnitude smaller than previously possible in some domains.

The problem of on-line planning in partially observable settings involves two problems: keeping track of beliefs about the environment and selecting actions for achieving goals. While the two problems are computationally intractable in the worst case, significant progress has been achieved in recent years through the use of suitable reductions. In particular, the state-of-the-art CLG planner is based on a translation that maps deterministic partially observable problems into fully observable non-deterministic ones. The translation, which is quadratic in the number of problem fluents and gets rid of the belief tracking problem, is adequate for most benchmarks, and it is in fact complete for problems that have width 1. The more recent K-replanner uses translations that are linear, one for keeping track of beliefs and the other for selecting actions using off-the-shelf classical planners. As a result, the K-replanner scales up better but it is not as general. In this work, we combine the benefits of the two approaches - the scope of the CLG planner and the efficiency of the Kreplanner. The new planner, called LW1, is based on a translation that is linear but complete for width-1 problems. The scope and scalability of the new planner is evaluated experimentally by considering the existing benchmarks and new problems.

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.

Multi-instance learning deals with tasks where each example is a bag of instances, and the bag labels of training data are known whereas instance labels are unknown. Most previous studies on multi-instance learning assumed that the training and testing data are from the same distribution; however, this assumption is often violated in real tasks. In this paper, we present possibly the first study on multi-instance learning with distribution change. We propose the MICS approach by considering both bag-level and instance-level distribution change. Experiments show that MICS is almost always significantly better than many state-of-the-art multi-instance learning algorithms when distribution change occurs; and even when there is no distribution change, their performances are still comparable.

Generalized fused lasso (GFL) penalizes variables with L1 norms based both on the variables and their pairwise differences. GFL is useful when applied to data where prior information is expressed using a graph over the variables. However, the existing GFL algorithms incur high computational costs and they do not scale to high-dimensional problems. In this study, we propose a fast and scalable algorithm for GFL. Based on the fact that fusion penalty is the Lov'asz extension of a cut function, we show that the key building block of the optimization is equivalent to recursively solving parametric graph-cut problems. Thus, we use a parametric flow algorithm to solve GFL in an efficient manner. Runtime comparisons demonstrated a significant speed-up compared with the existing GFL algorithms. By exploiting the scalability of the proposed algorithm, we formulated the diagnosis of Alzheimer's disease as GFL. Our experimental evaluations demonstrated that the diagnosis performance was promising and that the selected critical voxels were well structured i.e., connected, consistent according to cross-validation and in agreement with prior clinical knowledge.

Infinite SVM (iSVM) is a Dirichlet process (DP) mixture of large-margin classifiers. Though flexible in learning nonlinear classifiers and discovering latent clustering structures, iSVM has a difficult inference task and existing methods could hinder its applicability to large-scale problems. This paper presents a small-variance asymptotic analysis to derive a simple and efficient algorithm, which monotonically optimizes a max-margin DP-means (M2DPM) problem, an extension of DP-means for both predictive learning and descriptive clustering. Our analysis is built on Gibbs infinite SVMs, an alternative DP mixture of large-margin machines, which admits a partially collapsed Gibbs sampler without truncation by exploring data augmentation techniques. Experimental results show that M2DPM runs much faster than similar algorithms without sacrificing prediction accuracies.

Mutual information (MI) based approaches are a popular feature selection paradigm. Although the stated goal of MI-based feature selection is to identify a subset of features that share the highest mutual information with the class variable, most current MI-based techniques are greedy methods that make use of low dimensional MI quantities. The reason for using low dimensional approximation has been mostly attributed to the difficulty associated with estimating the high dimensional MI from limited samples. In this paper, we argue a different viewpoint that, given a very large amount of data, the high dimensional MI objective is still problematic to be employed as a meaningful optimization criterion, due to its overfitting nature: the MI almost always increases as more features are added, thus leading to a trivial solution which includes all features. We propose a novel approach to the MI-based feature selection problem, in which the overfitting phenomenon is controlled rigourously by means of a statistical test. We develop local and global optimization algorithms for this new feature selection model, and demonstrate its effectiveness in the applications of explaining variables and objects.

In many practical cases, we need to generalize a model trained in a source domain to a new target domain.However, the distribution of these two domains may differ very significantly, especially sometimes some crucial target features may not have support in the source domain.This paper proposes a novel locality preserving projection method for domain adaptation task,which can find a linear mapping preserving the 'intrinsic structure' for both source and target domains.We first construct two graphs encoding the neighborhood information for source and target domains separately.We then find linear projection coefficients which have the property of locality preserving for each graph.Instead of combing the two objective terms under compatibility assumption and requiring the user to decide the importance of each objective function,we propose a multi-objective formulation for this problem and solve it simultaneously using Pareto optimization.The Pareto frontier captures all possible good linear projection coefficients that are preferred by one or more objectives.The effectiveness of our approach is justified by both theoretical analysis and empirical results on real world data sets.The new feature representation shows better prediction accuracy as our experiments demonstrate.

A common bottleneck in deploying supervised learning systems is collecting human-annotated examples. In many domains, annotators form an opinion about the label of an example incrementally -- e.g., each additional word read from a document or each additional minute spent inspecting a video helps inform the annotation. In this paper, we investigate whether we can train learning systems more efficiently by requesting an annotation before inspection is fully complete -- e.g., after reading only 25 words of a document. While doing so may reduce the overall annotation time, it also introduces the risk that the annotator might not be able to provide a label if interrupted too early. We propose an anytime active learning approach that optimizes the annotation time and response rate simultaneously. We conduct user studies on two document classification datasets and develop simulated annotators that mimic the users. Our simulated experiments show that anytime active learning outperforms several baselines on these two datasets. For example, with an annotation budget of one hour, training a classifier by annotating the first 25 words of each document reduces classification error by 17% over annotating the first 100 words of each document.

Sequential learning for classification tasks is an effective tool in the machine learning community. In sequential learning settings, algorithms sometimes make incorrect predictions on data that were correctly classified in the past. This paper explicitly deals with such inconsistent prediction behavior. Our main contributions are 1) to experimentally show its effect for user utilities as a human cognitive bias, 2) to formalize a new framework by internalizing this bias into the optimization problem, 3) to develop new algorithms without memorization of the past prediction history, and 4) to show some theoretical guarantees of our derived algorithm for both online and stochastic learning settings. Our experimental results show the superiority of the derived algorithm for problems involving human cognition.