Multiagent Reinforcement Learning With Heterogeneous Graph Attention Network.

Most recent research on multiagent reinforcement learning (MARL) has explored how to deploy cooperative policies for homogeneous agents. However, realistic multiagent environments may contain heterogeneous agents that have different attributes or tasks. The heterogeneity of the agents and the diversity of relationships cause the learning of policy excessively tough. To tackle this difficulty, we present a novel method that employs a heterogeneous graph attention network to model the relationships between heterogeneous agents. The proposed method can generate an integrated feature representation for each agent by hierarchically aggregating latent feature information of neighbor agents, with the importance of the agent level and the relationship level being entirely considered. The method is agnostic to specific MARL methods and can be flexibly integrated with diverse value decomposition methods. We conduct experiments in predator-prey and StarCraft Multiagent Challenge (SMAC) environments, and the empirical results demonstrate that the performance of our method is superior to existing methods in several heterogeneous scenarios.

Triplex transfer learning: exploiting both shared and distinct concepts for text classification.

Transfer learning focuses on the learning scenarios when the test data from target domains and the training data from source domains are drawn from similar but different data distributions with respect to the raw features. Along this line, some recent studies revealed that the high-level concepts, such as word clusters, could help model the differences of data distributions, and thus are more appropriate for classification. In other words, these methods assume that all the data domains have the same set of shared concepts, which are used as the bridge for knowledge transfer. However, in addition to these shared concepts, each domain may have its own distinct concepts. In light of this, we systemically analyze the high-level concepts, and propose a general transfer learning framework based on nonnegative matrix trifactorization, which allows to explore both shared and distinct concepts among all the domains simultaneously. Since this model provides more flexibility in fitting the data, it can lead to better classification accuracy. Moreover, we propose to regularize the manifold structure in the target domains to improve the prediction performances. To solve the proposed optimization problem, we also develop an iterative algorithm and theoretically analyze its convergence properties. Finally, extensive experiments show that the proposed model can outperform the baseline methods with a significant margin. In particular, we show that our method works much better for the more challenging tasks when there are distinct concepts in the data.

A computational model for feature binding.

The "Binding Problem" is an important problem across many disciplines, including psychology, neuroscience, computational modeling, and even philosophy. In this work, we proposed a novel computational model, Bayesian Linking Field Model, for feature binding in visual perception, by combining the idea of noisy neuron model, Bayesian method, Linking Field Network and competitive mechanism. Simulation Experiments demonstrated that our model perfectly fulfilled the task of feature binding in visual perception and provided us some enlightening idea for future research.

The universal fuzzy logical framework of neural circuits and its application in modeling primary visual cortex.

Analytical study of large-scale nonlinear neural circuits is a difficult task. Here we analyze the function of neural systems by probing the fuzzy logical framework of the neural cells' dynamical equations. Although there is a close relation between the theories of fuzzy logical systems and neural systems and many papers investigate this subject, most investigations focus on finding new functions of neural systems by hybridizing fuzzy logical and neural system. In this paper, the fuzzy logical framework of neural cells is used to understand the nonlinear dynamic attributes of a common neural system by abstracting the fuzzy logical framework of a neural cell. Our analysis enables the educated design of network models for classes of computation. As an example, a recurrent network model of the primary visual cortex has been built and tested using this approach.