Neural Information Processing Systems http://nips.cc/

We argue that regularizing terms in standard regression methods not only help against overfitting finite data, but sometimes also help in getting bettercausal models.

Although synchronous AR training is designed to have higher training efficiency,asynchronous PStraining would beabetter choice for training speed when there are stragglers (slow workers) in the shared cluster, especially under limited computing resources.

Neural Information Processing Systems http://nips.cc/

Does progress on ImageNet transfer to real-world datasets?