Decentralized partially observable Markov decision processes (Dec-POMDPs) offer a powerful modeling technique for realistic multi-agent coordination problems under uncertainty. Prevalent solution techniques are centralized and assume prior knowledge of the model. We propose a distributed reinforcement learning approach, where agents take turns to learn best responses to each other’s policies. This promotes decentralization of the policy computation problem, and relaxes reliance on the full knowledge of the problem parameters. We derive the relation between the sample complexity of best response learning and error tolerance. Our key contribution is to show that sample complexity could grow exponentially with the problem horizon. We show empirically that even if the sample requirement is set lower than what theory demands, our learning approach can produce (near) optimal policies in some benchmark Dec-POMDP problems.

Decentralized partially observable Markov decision processes (Dec-POMDPs) offer a formal model for planning in cooperative multiagent systems where agents operate with noisy sensors and actuators, and local information. Prevalent Dec-POMDP solution techniques have mostly been centralized and have assumed knowledge of the model. In real world scenarios, however, solving centrally may not be an option and model parameters maybe unknown. To address this, we propose a distributed, model-free algorithm for learning Dec-POMDP policies, in which agents take turns learning, with each agent not currently learning following a static policy. For agents that have not yet learned a policy, this static policy must be initialized. We propose a principled method for learning such initial policies through interaction with the environment. We show that by using such informed initial policies, our alternate learning algorithm can find near-optimal policies for two benchmark problems.

Common spatial patterns (CSP) is a popular feature extraction method for discriminating between positive andnegative classes in electroencephalography (EEG) data.Two probabilistic models for CSP were recently developed: probabilistic CSP (PCSP), which is trained by expectation maximization (EM), and variational BayesianCSP (VBCSP) which is learned by variational approx-imation. Parameter expansion methods use auxiliaryparameters to speed up the convergence of EM or thedeterministic approximation of the target distributionin variational inference. In this paper, we describethe development of parameter-expanded algorithms forPCSP and VBCSP, leading to PCSP-PX and VBCSP-PX, whose convergence speed-up and high performanceare emphasized. The convergence speed-up in PCSP-PX and VBCSP-PX is a direct consequence of parame-ter expansion methods. The contribution of this study is the performance improvement in the case of CSP,which is a novel development. Numerical experimentson the BCI competition datasets, III IV a and IV 2ademonstrate the high performance and fast convergenceof PCSP-PX and VBCSP-PX, as compared to PCSP andVBCSP.

Inferring probabilistic networks from data is a notoriously difficult task. Under various goodness-of-fit measures, finding an optimal network is NP-hard, even if restricted to polytrees of bounded in-degree. Polynomial-time algorithms are known only for rare special cases, perhaps most notably for branchings, that is, polytrees in which the in-degree of every node is at most one. Here, we study the complexity of finding an optimal polytree that can be turned into a branching by deleting some number of arcs or nodes, treated as a parameter. We show that the problem can be solved via a matroid intersection formulation in polynomial time if the number of deleted arcs is bounded by a constant. The order of the polynomial time bound depends on this constant, hence the algorithm does not establish fixed-parameter tractability when parameterized by the number of deleted arcs. We show that a restricted version of the problem allows fixed-parameter tractability and hence scales well with the parameter. We contrast this positive result by showing that if we parameterize by the number of deleted nodes, a somewhat more powerful parameter, the problem is not fixed-parameter tractable, subject to a complexity-theoretic assumption.

Many first-order probabilistic models can be represented much more compactly using aggregation operations such as counting. While traditional statistical relational representations share factors across sets of interchangeable random variables, representations that explicitly model aggregations also exploit interchangeability of random variables within factors. This is especially useful in decision making settings, where an agent might need to reason about counts of the different types of objects it interacts with. Previous work on counting formulas in statistical relational representations has mostly focused on the problem of exact inference on an existing model. The problem of learning such models is largely unexplored. In this paper, we introduce Counting Markov Logic Networks (C-MLNs), an extension of Markov logic networks that can compactly represent complex counting formulas. We present a structure learning algorithm for C-MLNs; we apply this algorithm to the novel problem of generalizing natural language instructions, and to relational reinforcement learning in the Crossblock domain, in which standard MLN learning algorithms fail to find any useful structure. The C-MLN policies learned from natural language instructions are compact and intuitive, and, despite requiring no instructions on test games, win 20% more Crossblock games than a state-of-the-art algorithm for following natural language instructions.

We present the concept of Bayesian Tool Affordances as a solution to estimate the suitable action for the given tool to realize the given novel effects to the robot. We define Tool affordances as the “awareness within robot about the different kind of effects it can create in the environment using a tool”. It incorporates understanding the bi-directional association of executed Action, functionally relevant features of the Tool and the resulting effects. We propose Bayesian leaning of Tool Affordances for prediction, inference and planning capabilities while dealing with uncertainty, redundancy and irrelevant information using limited learning samples. The estimation results are presented in this paper to validate the proposed concept of Bayesian Tool Affordances.

This paper presents a theoretical advance by which factored POSGs can be decomposed into local models. We formalize the interface between such local models as the influence agents can exert on one another; and we prove that this interface is sufficient for decoupling them. The resulting influence-based abstraction substantially generalizes previous work on exploiting weakly-coupled agent interaction structures. Therein lie several important contributions. First, our general formulation sheds new light on the theoretical relationships among previous approaches, and promotes future empirical comparisons that could come by extending them beyond the more specific problem contexts for which they were developed. More importantly, the influence-based approaches that we generalize have shown promising improvements in the scalability of planning for more restrictive models. Thus, our theoretical result here serves as the foundation for practical algorithms that we anticipate will bring similar improvements to more general planning contexts, and also into other domains such as approximate planning, decision-making in adversarial domains, and online learning.

We study the marginal-MAP problem on graphical models, and present a novel approximation method based on direct approximation of the sum operation. A primary difficulty of marginal-MAP problems lies in the non-commutativity of the sum and max operations, so that even in highly structured models, marginalization may produce a densely connected graph over the variables to be maximized, resulting in an intractable potential function with exponential size. We propose a chain decomposition approach for summing over the marginalized variables, in which we produce a structured approximation to the MAP component of the problem consisting of only pairwise potentials. We show that this approach is equivalent to the maximization of a specific variational free energy, and it provides an upper bound of the optimal probability. Finally, experimental results demonstrate that our method performs favorably compared to previous methods.

We consider symbolic dynamic programming (SDP) for solving Markov Decision Processes (MDP) with factored state and action spaces, where both states and actions are described by sets of discrete variables. Prior work on SDP has considered only the case of factored states and ignored structure in the action space, causing them to scale poorly in terms of the number of action variables. Our main contribution is to present the first SDP-based planning algorithm for leveraging both state and action space structure in order to compute compactly represented value functions and policies. Since our new algorithm can potentially require more space than when action structure is ignored, our second contribution is to describe an approach for smoothly trading-off space versus time via recursive conditioning. Finally, our third contribution is to introduce a novel SDP approximation that often significantly reduces planning time with little loss in quality by exploiting action structure in weakly coupled MDPs. We present empirical results in three domains with factored action spaces that show that our algorithms scale much better with the number of action variables as compared to state-of-the-art SDP algorithms.

In music, boundaries may occur because scientific domains, including artificial intelligence (Keogh of multiple changes, such as a change in instrumentation, 2011). Research on time series has a long tradition, but a change in harmony, or a change in tempo. The seminal its application to real-world datasets requires to cope with approach by Foote (2000) estimated these changes by new relevant issues, such as the multiple dimensionality of means of a so-called novelty curve, obtained by sliding a data or limited computational resources. Specifically, dealing short-time checkerboard kernel over the diagonal of a selfsimilarity with large-scale data, (1) algorithms must be efficient, matrix of pairwise sample comparisons. Works inspired i.e. they have to scale, (2) supervised approaches may become by Foote's approach explicitly make use of the concept unfeasible, and (3) solutions must use general techniques, of novelty curves (Paulus et al. 2010). Other musictargeted i.e. they should be as independent of the domain as approaches exploit homogeneities in a time series possible (see Mueen and Keogh 2010 for a more detailed by employing more refined techniques like hidden Markov discussion).