Cabrel Teguemne Fokam

Asynchronous stochastic gradient descent with decoupled backpropagation and layer-wise updates

Block-local learning with probabilistic latent representations