Stochastic Block-ADMM for Training Deep Networks
This method addresses training challenges for deep networks with constraints, offering a solution for scenarios like non-differentiable layers, but it appears incremental as it builds on existing ADMM and stochastic optimization techniques.
The paper tackles the problem of training deep neural networks with non-differentiable constraints, where conventional backpropagation fails, by proposing Stochastic Block-ADMM, which splits networks into blocks and uses auxiliary variables for optimization. The result includes proven convergence and experimental validation in supervised and weakly-supervised settings, with applications like DeepFacto for feature disentangling.
In this paper, we propose Stochastic Block-ADMM as an approach to train deep neural networks in batch and online settings. Our method works by splitting neural networks into an arbitrary number of blocks and utilizes auxiliary variables to connect these blocks while optimizing with stochastic gradient descent. This allows training deep networks with non-differentiable constraints where conventional backpropagation is not applicable. An application of this is supervised feature disentangling, where our proposed DeepFacto inserts a non-negative matrix factorization (NMF) layer into the network. Since backpropagation only needs to be performed within each block, our approach alleviates vanishing gradients and provides potentials for parallelization. We prove the convergence of our proposed method and justify its capabilities through experiments in supervised and weakly-supervised settings.