Machine Learning with Clos Networks
This work addresses efficiency in neural network design for resource-constrained applications, but appears incremental as it builds on existing concepts like Clos networks and ReLU.
The paper tackled improving accuracy of small neural networks by applying Clos network concepts to maximize expression with fewer parameters, showing that more layers are beneficial given the same parameter count and exploring ReLU effects in separable networks, with results on Cifar-10.
We present a new methodology for improving the accuracy of small neural networks by applying the concept of a clos network to achieve maximum expression in a smaller network. We explore the design space to show that more layers is beneficial, given the same number of parameters. We also present findings on how the relu nonlinearity ffects accuracy in separable networks. We present results on early work with Cifar-10 dataset.