Training a multilayer network with low-memory kernel-and-range projection.
Journal of the Franklin Institute(2020)
摘要
•A low-memory formulation is proposed to address the high memory demand of a recently proposed kernel-and-range method.•The proposed formulation is proved to be mathematically equivalent to the original kernel-and-range method.•A regularization technique is introduced to cope with the existing rounding-error effect.•Experiments including synthetic and real-world datasets are provided.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要