HyperNetworks

TL;DR

This paper investigates hypernetworks, which generate weights for other networks, demonstrating their effectiveness in deep convolutional and recurrent models, achieving competitive results with fewer parameters.

Contribution

It introduces end-to-end trainable hypernetworks for deep and recurrent networks, challenging traditional weight-sharing paradigms and demonstrating their practical benefits.

Findings

Hypernetworks can generate non-shared weights for LSTM.

Achieve near state-of-the-art results on sequence modeling tasks.

Perform well on image recognition with fewer parameters.

Abstract

This work explores hypernetworks: an approach of using a one network, also known as a hypernetwork, to generate the weights for another network. Hypernetworks provide an abstraction that is similar to what is found in nature: the relationship between a genotype - the hypernetwork - and a phenotype - the main network. Though they are also reminiscent of HyperNEAT in evolution, our hypernetworks are trained end-to-end with backpropagation and thus are usually faster. The focus of this work is to make hypernetworks useful for deep convolutional networks and long recurrent networks, where hypernetworks can be viewed as relaxed form of weight-sharing across layers. Our main result is that hypernetworks can generate non-shared weights for LSTM and achieve near state-of-the-art results on a variety of sequence modelling tasks including character-level language modelling, handwriting generation and neural machine translation, challenging the weight-sharing paradigm for recurrent networks. Our results also show that hypernetworks applied to convolutional networks still achieve respectable results for image recognition tasks compared to state-of-the-art baseline models while requiring fewer learnable parameters.