A Clockwork RNN

TL;DR

The paper introduces the Clockwork RNN, a modified architecture that partitions the hidden layer into modules operating at different clock rates, enhancing long-term dependency modeling, reducing parameters, and improving performance in sequence tasks.

Contribution

It proposes a novel RNN architecture that improves training efficiency and performance by processing different parts of the network at separate temporal scales.

Findings

Outperforms standard RNNs and LSTMs in sequence tasks

Reduces number of parameters compared to standard RNNs

Speeds up network evaluation

Abstract

Sequence prediction and classification are ubiquitous and challenging problems in machine learning that can require identifying complex dependencies between temporally distant inputs. Recurrent Neural Networks (RNNs) have the ability, in theory, to cope with these temporal dependencies by virtue of the short-term memory implemented by their recurrent (feedback) connections. However, in practice they are difficult to train successfully when the long-term memory is required. This paper introduces a simple, yet powerful modification to the standard RNN architecture, the Clockwork RNN (CW-RNN), in which the hidden layer is partitioned into separate modules, each processing inputs at its own temporal granularity, making computations only at its prescribed clock rate. Rather than making the standard RNN models more complex, CW-RNN reduces the number of RNN parameters, improves the performance significantly in the tasks tested, and speeds up the network evaluation. The network is demonstrated in preliminary experiments involving two tasks: audio signal generation and TIMIT spoken word classification, where it outperforms both RNN and LSTM networks.