A Fully Tensorized Recurrent Neural Network

TL;DR

This paper introduces a fully tensorized RNN architecture using tensor-train factorization to significantly reduce model size and improve training stability without sacrificing performance.

Contribution

It presents a novel tensorized RNN design that encodes weight matrices with tensor-train factorization, enabling efficient, compact models for sequential tasks.

Findings

Reduces model size by several orders of magnitude.

Maintains or improves performance on classification and verification tasks.

Enhances inference speed and training stability.

Abstract

Recurrent neural networks (RNNs) are powerful tools for sequential modeling, but typically require significant overparameterization and regularization to achieve optimal performance. This leads to difficulties in the deployment of large RNNs in resource-limited settings, while also introducing complications in hyperparameter selection and training. To address these issues, we introduce a "fully tensorized" RNN architecture which jointly encodes the separate weight matrices within each recurrent cell using a lightweight tensor-train (TT) factorization. This approach represents a novel form of weight sharing which reduces model size by several orders of magnitude, while still maintaining similar or better performance compared to standard RNNs. Experiments on image classification and speaker verification tasks demonstrate further benefits for reducing inference times and stabilizing model training and hyperparameter selection.