Hogwild: A Lock-Free Approach to Parallelizing Stochastic Gradient Descent

Open in new window