Learning Representations from Knowledge Graphs – We propose a novel framework for learning structured models of action-action interactions. The framework is based on the recent work of Yap and Chiao (2009) of learning structured models of action-action interfaces. In a supervised domain, a deep network is trained to be able to model the interaction of user-defined actions and objects, and then the model is extended to learn actions or objects independently. This framework learns the interactions of multiple users and interactions, and the interaction model is then modeled on the interaction space of the user-defined actions and objects. We show how to use the framework for learning structured action models from action spaces in a setting where the user has limited amount of knowledge. As a case study, we experimentally demonstrate the usefulness of the proposed system. The method is able to learn an agent from a knowledge graph, and the knowledge graph is then used to model the interaction between the agent and the user model.

A novel algorithm for the problem of learning a graph from a large corpus of texts is presented. Given $n$ sentences in English and English-German texts, the resulting graph is drawn from a large corpus of texts and labeled by the word level semantic similarity (SLE) method. In this paper, we formulate the graph-learning problem as two two-fold optimization problem: one is a sparse-sum solution problem, whereas the other is a sum problem to solve efficiently. This leads to a simple and efficient, flexible and accurate algorithm that is capable of solving both problems in the same round. The algorithm is based on a new approach for the SLE problem which addresses the main problem in this paper. Our algorithm shows good results, outperforming the previous two techniques.

Deep Predictive Models and Neural Networks

# Learning Representations from Knowledge Graphs

Improving Recurrent Neural Network with Contextual Dependence

On the Utility of the Maximum Entropy Principle for Modeling the Math of Concept ReuseA novel algorithm for the problem of learning a graph from a large corpus of texts is presented. Given $n$ sentences in English and English-German texts, the resulting graph is drawn from a large corpus of texts and labeled by the word level semantic similarity (SLE) method. In this paper, we formulate the graph-learning problem as two two-fold optimization problem: one is a sparse-sum solution problem, whereas the other is a sum problem to solve efficiently. This leads to a simple and efficient, flexible and accurate algorithm that is capable of solving both problems in the same round. The algorithm is based on a new approach for the SLE problem which addresses the main problem in this paper. Our algorithm shows good results, outperforming the previous two techniques.