In this paper, we propose a novel supervised extension of GPLVM, called Gaussian process latent random field (GPLRF), by enforcing the latent variables to be a Gaussian Markov random field with respect to a graph constructed from the supervisory information.

Many problems in information extraction, text mining, natural language processing and other fields exhibit the same property: multiple prediction tasks are related in the sense that their outputs (labels) satisfy certain constraints. In this paper, we propose an active learning framework exploiting such relations among tasks. Intuitively, with task outputs coupled by constraints, active learning can utilize not only the uncertainty of the prediction in a single task but also the inconsistency of predictions across tasks. We formalize this idea as a cross-task value of information criteria, in which the reward of a labeling assignment is propagated and measured over all relevant tasks reachable through constraints. A specific example of our framework leads to the cross entropy measure on the predictions of coupled tasks, which generalizes the entropy in the classical single-task uncertain sampling. We conduct experiments on two real-world problems: web information extraction and document classification. Empirical results demonstrate the effectiveness of our framework in actively collecting labeled examples for multiple related tasks.

Graph-based semi-supervised learning methods are based on some smoothness assumption about the data. As a discrete approximation of the data manifold, the graph plays a crucial role in the success of such graph-based methods. In most existing methods, graph construction makes use of a predefined weighting function without utilizing label information even when it is available. In this work, by incorporating label information, we seek to enhance the performance of graph-based semi-supervised learning by learning the graph and label inference simultaneously. In particular, we consider a particular setting of semi-supervised learning called transductive learning. Using the LogDet divergence to define the objective function, we propose an iterative algorithm to solve the optimization problem which has closed-form solution in each step. We perform experiments on both synthetic and real data to demonstrate improvement in the graph and in terms of classification accuracy.

Traditional clustering methods deal with a single clustering task on a single data set. However, in some newly emerging applications, multiple similar clustering tasks are involved simultaneously. In this case, we not only desire a partition for each task, but also want to discover the relationship among clusters of different tasks. It's also expected that the learnt relationship among tasks can improve performance of each single task. In this paper, we propose a general framework for this problem and further suggest a specific approach. In our approach, we alternatively update clusters and learn relationship between clusters of different tasks, and the two phases boost each other. Our approach is based on the general Bregman divergence, hence it's suitable for a large family of assumptions on data distributions and divergences. Empirical results on several benchmark data sets validate the approach.

Over the past few years, a large family of manifold learning algorithms have been proposed, and applied to various applications. While designing new manifold learning algorithms has attracted much research attention, fewer research efforts have been focused on out-of-sample extrapolation of learned manifold. In this paper, we propose a novel algorithm of manifold learning. The proposed algorithm, namely Local and Global Regressive Mapping (LGRM), employs local regression models to grasp the manifold structure. We additionally impose a global regression term as regularization to learn a model for out-of-sample data extrapolation. Based on the algorithm, we propose a new manifold learning framework. Our framework can be applied to any manifold learning algorithms to simultaneously learn the low dimensional embedding of the training data and a model which provides explicit mapping of the out-of-sample data to the learned manifold. Experiments demonstrate that the proposed framework uncover the manifold structure precisely and can be freely applied to unseen data.

The discovery of non-linear causal relationship under additive non-Gaussian noise models has attracted considerable attention recently because of their high flexibility. In this paper, we propose a novel causal inference algorithm called least-squares independence regression (LSIR). LSIR learns the additive noise model through minimization of an estimator of the squared-loss mutual information between inputs and residuals. A notable advantage of LSIR over existing approaches is that tuning parameters such as the kernel width and the regularization parameter can be naturally optimized by cross-validation, allowing us to avoid overfitting in a data-dependent fashion. Through experiments with real-world datasets, we show that LSIR compares favorably with the state-of-the-art causal inference method.

The current methods used to mine and analyze temporal social network data make two assumptions: all edges have the same strength, and all parameters are time-homogeneous. We show that those assumptions may not hold for social networks and propose an alternative model with two novel aspects: (1) the modeling of edges as multi-valued variables that can change in intensity, and (2) the use of a curved exponential family framework to capture time-inhomogeneous properties while retaining a parsimonious and interpretable model. We show that our model outperforms traditional models on two real-world social network data sets.

Bayesian inference is an appealing approach for leveraging prior knowledge in reinforcement learning (RL). In this paper we describe an algorithm for discovering different classes of roles for agents via Bayesian inference. In particular, we develop a Bayesian policy search approach for Multi-Agent RL (MARL), which is model-free and allows for priors on policy parameters. We present a novel optimization algorithm based on hybrid MCMC, which leverages both the prior and gradient information estimated from trajectories. Our experiments in a complex real-time strategy game demonstrate the effective discovery of roles from supervised trajectories, the use of discovered roles for successful transfer to similar tasks, and the discovery of roles through reinforcement learning.

Many real life applications brought by modern technologies often have multiple data sources, which are usually characterized by both attributes and pairwise similarities at the same time. For example in webpage ranking, a webpage is usually represented by a vector of term values, and meanwhile the internet linkages induce pairwise similarities among the webpages. Although both attributes and pairwise similarities are useful for class membership inference, many traditional embedding algorithms only deal with one type of input data. In order to make use of the both types of data simultaneously, in this work, we propose a novel Discriminant Laplacian Embedding (DLE) approach. Supervision information from training data are integrated into DLE to improve the discriminativity of the resulted embedding space. By solving the ambiguity problem in computing the scatter matrices caused by data points with multiple labels, we successfully extend the proposed DLE to multi-label classification. In addition, through incorporating the label correlations, the classification performance using multi-label DLE is further enhanced. Promising experimental results in extensive empirical evaluations have demonstrated the effectiveness of our approaches.

Recent advancements in model-based reinforcement learning have shown that the dynamics of many structured domains (e.g. DBNs) can be learned with tractable sample complexity, despite their exponentially large state spaces. Unfortunately, these algorithms all require access to a planner that computes a near optimal policy, and while many traditional MDP algorithms make this guarantee, their computation time grows with the number of states. We show how to replace these over-matched planners with a class of sample-based planners — whose computation time is independent of the number of states — without sacrificing the sample-efficiency guarantees of the overall learning algorithms. To do so, we define sufficient criteria for a sample-based planner to be used in such a learning system and analyze two popular sample-based approaches from the literature. We also introduce our own sample-based planner, which combines the strategies from these algorithms and still meets the criteria for integration into our learning system. In doing so, we define the first complete RL solution for compactly represented (exponentially sized) state spaces with efficiently learnable dynamics that is both sample efficient and whose computation time does not grow rapidly with the number of states.