Temporal-related Convolutional-Restricted-Boltzmann-Machine capable of learning relational order via reinforcement learning procedure?
This work addresses the challenge of modeling complex temporal relationships in data for machine learning applications, representing an incremental improvement over existing methods.
The paper tackles the problem of learning abstract features from multiple time-related input maps by extending convolutional Restricted Boltzmann Machines with multiplicative units to capture relations among more than two maps, and it develops a reinforcement learning method to optimize the relational order, with proven optimality for training.
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.