Exploring Weight Symmetry in Deep Neural Networks
This addresses parameter redundancy for practitioners in computer vision and NLP, offering a way to deploy more efficient models without significant performance drops, though it is incremental as it builds on existing architectures.
The paper tackles the problem of parameter inefficiency in deep neural networks by imposing weight symmetry, which reduces parameters by up to 25% with minimal accuracy loss, such as only 0.2% on ImageNet for ResNet-101.
We propose to impose symmetry in neural network parameters to improve parameter usage and make use of dedicated convolution and matrix multiplication routines. Due to significant reduction in the number of parameters as a result of the symmetry constraints, one would expect a dramatic drop in accuracy. Surprisingly, we show that this is not the case, and, depending on network size, symmetry can have little or no negative effect on network accuracy, especially in deep overparameterized networks. We propose several ways to impose local symmetry in recurrent and convolutional neural networks, and show that our symmetry parameterizations satisfy universal approximation property for single hidden layer networks. We extensively evaluate these parameterizations on CIFAR, ImageNet and language modeling datasets, showing significant benefits from the use of symmetry. For instance, our ResNet-101 with channel-wise symmetry has almost 25% less parameters and only 0.2% accuracy loss on ImageNet. Code for our experiments is available at https://github.com/hushell/deep-symmetry