From Nodes to Networks: Evolving Recurrent Neural Networks
This work addresses the need for automated design of neural network structures to enhance performance in sequential processing tasks like language modeling, representing an incremental advancement over existing evolutionary methods.
The paper tackled the problem of designing better recurrent neural network nodes by proposing an evolutionary method using tree-based encoding, which discovered nodes with multiple recurrent paths and memory cells, leading to significant improvement in a standard language modeling benchmark task.
Gated recurrent networks such as those composed of Long Short-Term Memory (LSTM) nodes have recently been used to improve state of the art in many sequential processing tasks such as speech recognition and machine translation. However, the basic structure of the LSTM node is essentially the same as when it was first conceived 25 years ago. Recently, evolutionary and reinforcement learning mechanisms have been employed to create new variations of this structure. This paper proposes a new method, evolution of a tree-based encoding of the gated memory nodes, and shows that it makes it possible to explore new variations more effectively than other methods. The method discovers nodes with multiple recurrent paths and multiple memory cells, which lead to significant improvement in the standard language modeling benchmark task. The paper also shows how the search process can be speeded up by training an LSTM network to estimate performance of candidate structures, and by encouraging exploration of novel solutions. Thus, evolutionary design of complex neural network structures promises to improve performance of deep learning architectures beyond human ability to do so.