Beyond Gradient Averaging in Parallel Optimization: Improved Robustness through Gradient Agreement Filtering

Open in new window