2017 (modified: 11 Nov 2022)ICML 2017Readers: Everyone
Abstract:We approach structured output prediction by optimizing a deep value network (DVN) to precisely estimate the task loss on different output configurations for a given input. Once the model is trained...