4 papers
Efficient Wasserstein Natural Gradients for Reinforcement Learning
Ted Moskovitz, Michael Arbel, Ferenc Huszar +1
A novel optimization approach is proposed for application to policy gradient methods and evolution strategies for reinforcement learning (RL). The procedure uses a computationally…
Amortised Learning by Wake-Sleep
Li K. Wenliang, Theodore Moskovitz, Heishiro Kanagawa +1
Models that employ latent variables to capture structure in observed data lie at the heart of many current unsupervised learning algorithms, but exact maximum-likelihood learning f…
First-Order Preconditioning via Hypergradient Descent
Ted Moskovitz, Rui Wang, Janice Lan +4
Standard gradient descent methods are susceptible to a range of issues that can impede training, such as high correlations and different scaling in parameter space.These difficulti…
Feedback alignment in deep convolutional networks
Theodore H. Moskovitz, Ashok Litwin-Kumar, L. F. Abbott
Ongoing studies have identified similarities between neural representations in biological networks and in deep artificial neural networks. This has led to renewed interest in devel…