Predictive State Recurrent Neural Networks
arXiv:1705.09353
Abstract
We present a new model, Predictive State Recurrent Neural Networks (PSRNNs), for filtering and prediction in dynamical systems. PSRNNs draw on insights from both Recurrent Neural Networks (RNNs) and Predictive State Representations (PSRs), and inherit advantages from both types of models. Like many successful RNN architectures, PSRNNs use (potentially deeply composed) bilinear transfer functions to combine information from multiple sources. We show that such bilinear functions arise naturally from state updates in Bayes filters like PSRs, in which observations can be viewed as gating belief states. We also show that PSRNNs can be learned effectively by combining Backpropogation Through Time (BPTT) with an initialization derived from a statistically consistent learning algorithm for PSRs called two-stage regression (2SR). Finally, we show that PSRNNs can be factorized using tensor decomposition, reducing model size and suggesting interesting connections to existing multiplicative architectures such as LSTMs. We applied PSRNNs to 4 datasets, and showed that we outperform several popular alternative approaches to modeling dynamical systems in all cases.
References in corpus (3)
Cited by in corpus (11)
- Deep Reinforcement Learning: An Overview
- Evaluation of deep learning models for multi-step ahead time series prediction
- Incorporating Symmetry into Deep Dynamics Models for Improved Generalization
- Connecting Weighted Automata and Recurrent Neural Networks through Spectral Learning
- General Value Function Networks
- Learning and Inference in Hilbert Space with Quantum Graphical Models
- Recurrent Neural Processes
- Dynamical Systems as Temporal Feature Spaces
- On Kernel Derivative Approximation with Random Fourier Features
- Learning Transition Models with Time-delayed Causal Relations
- A Unified Off-Policy Evaluation Approach for General Value Function