Temporal-related Convolutional-Restricted-Boltzmann-Machine capable of learning relational order via reinforcement learning procedure?

TL;DR

This paper introduces an extension of convolutional-Restricted-Boltzmann-Machine that learns complex temporal relations among multiple input maps using multiplicative units and reinforcement learning for optimal relational-order detection.

Contribution

It proposes a novel extension with multiplicative units and a reinforcement learning method to learn relational order among multiple temporal input maps.

Findings

Successfully extends CRBM to handle multiple temporal relations

Develops a reinforcement learning approach for optimal relational-order learning

Proves the optimality of the training method

Abstract

In this article, we extend the conventional framework of convolutional-Restricted-Boltzmann-Machine to learn highly abstract features among abitrary number of time related input maps by constructing a layer of multiplicative units, which capture the relations among inputs. In many cases, more than two maps are strongly related, so it is wise to make multiplicative unit learn relations among more input maps, in other words, to find the optimal relational-order of each unit. In order to enable our machine to learn relational order, we developed a reinforcement-learning method whose optimality is proven to train the network.