When a transition probability matrix is represented as the product of two stochastic matrices, one can swap the factors of the multiplication to obtain another transition matrix that retains some fundamental characteristics of the original. Since the derived matrix can be much smaller than its precursor, this property can be exploited to create a compact version of a Markov decision process (MDP), and hence to reduce the computational cost of dynamic programming. Building on this idea, this paper presents an approximate policy iteration algorithm called policy iteration based on stochastic factorization, or PISF for short. In terms of computational complexity, PISF replaces standard policy iteration's cubic dependence on the size of the MDP with a function that grows only linearly with the number of states in the model. The proposed algorithm also enjoys nice theoretical properties: it always terminates after a finite number of iterations and returns a decision policy whose performance only depends on the quality of the stochastic factorization. In particular, if the approximation error in the factorization is sufficiently small, PISF computes the optimal value function of the MDP. The paper also discusses practical ways of factoring an MDP and illustrates the usefulness of the proposed algorithm with an application involving a large-scale decision problem of real economical interest.

While labeled data is expensive to prepare, ever increasing amounts of unlabeled data is becoming widely available. In order to adapt to this phenomenon, several semi-supervised learning (SSL) algorithms, which learn from labeled as well as unlabeled data, have been developed. In a separate line of work, researchers have started to realize that graphs provide a natural way to represent data in a variety of domains. Graph-based SSL algorithms, which bring together these two lines of work, have been shown to outperform the state-of-the-art in many applications in speech processing, computer vision, natural language processing, and other areas of Artificial Intelligence. Recognizing this promising and emerging area of research, this synthesis lecture focuses on graph-based SSL algorithms (e.g., label propagation methods).

Editing faces in videos is a popular yet challenging aspect of computer vision and graphics, which encompasses several applications including facial attractiveness enhancement, makeup transfer, face replacement, and expression manipulation. Simply applying image-based warping algorithms to video-based face editing produces temporal incoherence in the synthesized videos because it is impossible to consistently localize facial features in two frames representing two different faces in two different videos (or even two consecutive frames representing the same face in one video). Therefore, high performance face editing usually requires significant manual manipulation. In this paper we propose a novel temporal-spatial-smooth warping (TSSW) algorithm to effectively exploit the temporal information in two consecutive frames, as well as the spatial smoothness within each frame. TSSW precisely estimates two control lattices in the horizontal and vertical directions respectively from the corresponding control lattices in the previous frame, by minimizing a novel energy function that unifies a data-driven term, a smoothness term, and feature point constraints. Corresponding warping surfaces then precisely map source frames to the target frames. Experimental testing on facial attractiveness enhancement, makeup transfer, face replacement, and expression manipulation demonstrates that the proposed approaches can effectively preserve spatial smoothness and temporal coherence in editing facial geometry, skin detail, identity, and expression, which outperform the existing face editing methods. In particular, TSSW is robust to subtly inaccurate localization of feature points and is a vast improvement over image-based warping methods.

Ordinal data is omnipresent in almost all multiuser-generated feedback - questionnaires, preferences etc. This paper investigates modelling of ordinal data with Gaussian restricted Boltzmann machines (RBMs). In particular, we present the model architecture, learning and inference procedures for both vector-variate and matrix-variate ordinal data. We show that our model is able to capture latent opinion profile of citizens around the world, and is competitive against state-of-art collaborative filtering techniques on large-scale public datasets. The model thus has the potential to extend application of RBMs to diverse domains such as recommendation systems, product reviews and expert assessments.

Collaborative robots that operate in the same immediate environment as human workers have the potential to improve their co-workers' efficiency and quality of work. In this paper we present a taxonomy of assistive behavior types alongside methods that enable a robot to learn assistive behaviors from interactions with a human collaborator during live activity completion. We begin with a brief survey of the state of the art in human-robot collaboration. We proceed to focus on the challenges and issues surrounding the online development of assistive robot behaviors. Finally, we describe approaches for learning when and how to apply these behaviors, as well as for integrating them into a full end-to-end system utilizing techniques derived from the learning from demonstration, policy iteration, and task network communities.

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.

Reviews keep playing an increasingly important role in the decision process of buying products and booking hotels. However, the large amount of available information can be confusing to users. A more succinct interface, gathering only the most helpful reviews, can reduce information processing time and save effort. To create such an interface in real time, we need reliable prediction algorithms to classify and predict new reviews which have not been voted but are potentially helpful. So far such helpfulness prediction algorithms have benefited from structural aspects, such as the length and readability score. Since emotional words are at the heart of our written communication and are powerful to trigger listeners' attention, we believe that emotional words can serve as important parameters for predicting helpfulness of review text. Using GALC, a general lexicon of emotional words associated with a model representing 20 different categories, we extracted the emotionality from the review text and applied supervised classification method to derive the emotion-based helpful review prediction. As the second contribution, we propose an evaluation framework comparing three different real-world datasets extracted from the most well-known product review websites. This framework shows that emotion-based methods are outperforming the structure-based approach, by up to 9%.

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.

The dynamics of belief and knowledge is one of the major components of any autonomous system that should be able to incorporate new pieces of information. In order to apply the rationality result of belief dynamics theory to various practical problems, it should be generalized in two respects: first it should allow a certain part of belief to be declared as immutable; and second, the belief state need not be deductively closed. Such a generalization of belief dynamics, referred to as base dynamics, is presented in this paper, along with the concept of a generalized revision algorithm for knowledge bases (Horn or Horn logic with stratified negation). We show that knowledge base dynamics has an interesting connection with kernel change via hitting set and abduction. In this paper, we show how techniques from disjunctive logic programming can be used for efficient (deductive) database updates. The key idea is to transform the given database together with the update request into a disjunctive (datalog) logic program and apply disjunctive techniques (such as minimal model reasoning) to solve the original update problem. The approach extends and integrates standard techniques for efficient query answering and integrity checking. The generation of a hitting set is carried out through a hyper tableaux calculus and magic set that is focused on the goal of minimality.

We study agnostic active learning, where the goal is to learn a classifier in a pre-specified hypothesis class interactively with as few label queries as possible, while making no assumptions on the true function generating the labels. The main algorithms for this problem are {\em{disagreement-based active learning}}, which has a high label requirement, and {\em{margin-based active learning}}, which only applies to fairly restricted settings. A major challenge is to find an algorithm which achieves better label complexity, is consistent in an agnostic setting, and applies to general classification problems. In this paper, we provide such an algorithm. Our solution is based on two novel contributions -- a reduction from consistent active learning to confidence-rated prediction with guaranteed error, and a novel confidence-rated predictor.