NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:4378
Title:Dynamic Local Regret for Non-convex Online Forecasting

The problem and formal model are interesting and general. The proposed objective is a non-straightforward modification of an existing criterion. It is subsequently optimized by adaptation of existing methods. The clarification provided in the author rebuttal (including the toy example) were very useful in appreciating the paper, and swung the reviewing process to accept the paper. Please incorporate the information from the rebuttal in the camera-ready version.