Abstract:Understanding the dynamics of a system is important in many scientific and engineering domains. This problem can be approached by learning state transition rules from observations using machine learning techniques. Such observed time-series data often consist of sequences of many continuous variables with noise and ambiguity, but we often need rules of dynamics that can be modeled with a few essential variables. In this work, we propose a method for extracting a small number of essential hidden variables from high-dimensional time-series data and for learning state transition rules between these hidden variables. The proposed method is based on the Restricted Boltzmann Machine (RBM), which treats observable data in the visible layer and latent features in the hidden layer. However, real-world data, such as video and audio, include both discrete and continuous variables, and these variables have temporal relationships. Therefore, we propose Recurrent Temporal GaussianBernoulli Restricted Boltzmann Machine (RTGB-RBM), which combines Gaussian-Bernoulli Restricted Boltzmann Machine (GB-RBM) to handle continuous visible variables, and Recurrent Temporal Restricted Boltzmann Machine (RT-RBM) to capture time dependence between discrete hidden variables. We also propose a rule-based method that extracts essential information as hidden variables and represents state transition rules in interpretable form. We conduct experiments on Bouncing Ball and Moving MNIST datasets to evaluate our proposed method. Experimental results show that our method can learn the dynamics of those physical systems as state transition rules between hidden variables and can predict unobserved future states from observed state transitions.

Sequence Classification Restricted Boltzmann Machines With Gated Units

Linear-Time Sequence Classification using Restricted Boltzmann Machines

Residual Recurrent Neural Networks for Learning Sequential Representations.

Temporal Attention-Gated Model for Robust Sequence Classification

Generative and Discriminative Infinite Restricted Boltzmann Machine Training

Recurrently Controlled Recurrent Networks

A Precise Method for RBMs Training Using Phased Curricula

A Spiking Neural Network System for Robust Sequence Recognition

Dynamic temporal residual network for sequence modeling

Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

A Novel Restricted Boltzmann Machine Training Algorithm with Fast Gibbs Sampling Policy

Learning State Transition Rules from Hidden Layers of Restricted Boltzmann Machines

Subspace Restricted Boltzmann Machine

On extended long short-term memory and dependent bidirectional recurrent neural network

Sparse Group Restricted Boltzmann Machines

A New Variant of Restricted Boltzmann Machine with Horizontal Connections

The Recurrent Temporal Restricted Boltzmann Machine Captures Neural Assembly Dynamics in Whole-brain Activity

Hierarchically Gated Recurrent Neural Network for Sequence Modeling

Gaussian Cardinality Restricted Boltzmann Machines.

Deep Gate Recurrent Neural Network

Gated Tree-Structured RecurNN for Detecting Biomedical Event Trigger