In many multi-agent applications such as distributed sensor nets, a network of agents act collaboratively under uncertainty and local interactions. Networked Distributed POMDP (ND-POMDP) provides a framework to model such cooperative multi-agent decision making. Existing work on ND-POMDPs has focused on offline techniques that require accurate models, which are usually costly to obtain in practice. This paper presents a model-free, scalable learning approach that synthesizes multi-agent reinforcement learning (MARL) and distributed constraint optimization (DCOP). By exploiting structured interaction in ND-POMDPs, our approach distributes the learning of the joint policy and employs DCOP techniques to coordinate distributed learning to ensure the global learning performance. Our approach can learn a globally optimal policy for ND-POMDPs with a property called groupwise observability. Experimental results show that, with communication during learning and execution, our approach significantly outperforms the nearly-optimal non-communication policies computed offline.

The behavior and dynamics of complex systems are in focus of many research fields. The complexity of such systems comes not only from the number of their elements, but also from the unavoidable emergence of new properties of the system, which are not just a simple summation of the properties of its elements. The behavior of complex systems can be fitted with a number of well developed models, which, however, do not incorporate the modularity and the evolution of a system simultaneously. In this work, we propose a generalized model that addresses this issue. Our model is developed within the Random Set Theory’s framework and allows for reconstructing the stochastic evolution diagrams of complex systems.

Recently, a number of researchers have proposed spectral algorithms for learning models of dynamical systems — for example, Hidden Markov Models (HMMs), Partially Observable Markov Decision Processes (POMDPs), and Transformed Predictive State Representations (TPSRs). These algorithms are attractive since they are statistically consistent and not subject to local optima. However, they are batch methods: they need to store their entire training data set in memory at once and operate on it as a large matrix, and so they cannot scale to extremely large data sets (either many examples or many features per example). In turn, this restriction limits their ability to learn accurate models of complex systems. To overcome these limitations, we propose a new online spectral algorithm, which uses tricks such as incremental Singular Value Decomposition (SVD) and random projections to scale to much larger data sets and more complex systems than previous methods. We demonstrate the new method on an inertial measurement prediction task and a high-bandwidth video mapping task and we illustrate desirable behaviors such as "closing the loop," where the latent state representation changes suddenly as the learner recognizes that it has returned to a previously known place.

Inspired by “GoogleTM Sets” and Bayesian sets, we consider the problem of retrieving complex objects and relations among them, i.e., ground atoms from a logical concept, given a query consisting of a few atoms from that concept. We formulate this as a within-network relational learning problem using few labels only and describe an algorithm that ranks atoms using a score based on random walks with restart (RWR): the probability that a random surfer hits an atom starting from the query atoms. Specifically, we compute an initial ranking using personalized PageRank. Then, we find paths of atoms that are connected via their arguments, variablize the ground atoms in each path, in order to create features for the query. These features are used to re-personalize the original RWR and to finally compute the set completion, based on Label Propagation. Moreover, we exploit that RWR techniques can naturally be lifted and show that lifted inference for label propagation is possible. We evaluate our algorithm on a realworld relational dataset by finding completions of sets of objects describing the Roman city of Pompeii. We compare to Bayesian sets and show that our approach gives very reasonable set completions.

We focus on a class of models used for representing the dynamics between a discrete set of probabilistic events in a continuous-time setting. The proposed framework offers tractable learning and inference procedures and provides compact state representations for processes which exhibit variable delays between events. The approach is applied to a heart sound labeling task that exhibits long-range dependencies on previous events, and in which explicit modeling of the rhythm timings is justifiable by cardiological principles.

Plan recognition is a form of abductive reasoning that involves inferring plans that best explain sets of observed actions. Most existing approaches to plan recognition and other abductive tasks employ either purely logical methods that donot handle uncertainty, or purely probabilistic methods thatdo not handle structured representations. To overcome these limitations, this paper introduces an approach to abductive reasoning using a first-order probabilistic logic, specifically Markov Logic Networks (MLNs). It introduces several novel techniques for making MLNs efficient and effective for abduction. Experiments on three plan recognition datasets showthe benefit of our approach over existing methods.

Various operational control systems (OCS) are naturally modeled as Markov Decision Processes. OCS often enjoy access to predictions of future events that have substantial impact on their operations. For example, reliable forecasts of extreme weather conditions are widely available, and such events can affect typical request patterns for customer response management systems, the flight and service time of airplanes, or the supply and demand patterns for electricity. The space of exogenous events impacting OCS can be very large, prohibiting their modeling within the MDP; moreover, for many of these exogenous events there is no useful predictive, probabilistic model. Realtime predictions, however, possibly with a short lead-time, are often available. In this work we motivate a model which combines offline MDP infinite horizon planning with realtime adjustments given specific predictions of future exogenous events, and suggest a framework in which such predictions are captured and trigger real-time planning problems. We propose a number of variants of existing MDP solution algorithms, adapted to this context, and evaluate them empirically.

Since learning in Boltzmann machines is typically quite slow, there is a need to restrict connections within hidden layers. However, theresulting states of hidden units exhibit statistical dependencies. Based on this observation, we propose using l1/l2 regularization upon the activation probabilities of hidden units in restricted Boltzmann machines to capture the local dependencies among hidden units. This regularization not only encourages hidden units of many groups to be inactive given observed data but also makes hidden units within a group compete with each other for modeling observed data. Thus, the l1/l2 regularization on RBMs yields sparsity at both the group and the hidden unit levels. We call RBMs trained with the regularizer sparse group RBMs (SGRBMs). The proposed SGRBMs are appliedto model patches of natural images, handwritten digits and OCR English letters. Then to emphasize that SGRBMs can learn more discriminative features we applied SGRBMs to pretrain deep networks for classification tasks. Furthermore, we illustrate the regularizer can also be applied to deep Boltzmann machines, which lead to sparse group deep Boltzmann machines. When adapted to the MNIST data set, a two-layer sparse group Boltzmann machine achieves an error rate of 0.84%, which is, to our knowledge, the best published result on the permutation-invariant version of the MNIST task.

This paper presents a generative model of eye-hand coordination. We use numerical optimization to solve for the joint behavior of an eye and two hands, deriving a predicted motion pattern from first principles, without imposing heuristics. We model the planar scene as a POMDP with 17 continuous state dimensions. Belief-space optimization is facilitated by using a nominal-belief heuristic, whereby we assume (during planning) that the maximum likelihood observation is always obtained. Since a globally-optimal solution for such a high-dimensional domain is computationally intractable, we employ local optimization in the belief domain. By solving for a locally-optimal plan through belief space, we generate a motion pattern of mutual coordination between hands and eye: the eye's saccades disambiguate the scene in a task-relevant manner, and the hands' motions anticipate the eye's saccades. Finally, the model is validated through a behavioral experiment, in which human subjects perform the same eye-hand coordination task. We show how simulation is congruent with the experimental results.

Planning under uncertainty for multiagent systems can be formalized as a decentralized partially observable Markov decision process. We advance the state of the art for optimal solution of this model, building on the Multiagent A* heuristic search method. A key insight is that we can avoid the full expansion of a search node that generates a number of children that is doubly exponential in the node's depth. Instead, we incrementally expand the children only when a next child might have the highest heuristic value. We target a subsequent bottleneck by introducing a more memory-efficient representation for our heuristic functions. Proof is given that the resulting algorithm is correct and experiments demonstrate a significant speedup over the state of the art, allowing for optimal solutions over longer horizons for many benchmark problems.