Implicit dynamic regularization in deep networks

TitleImplicit dynamic regularization in deep networks
Publication TypeCBMM Memos
Year of Publication2020
AuthorsPoggio, TA, Liao, Q
Date Published08/2020
Abstract

Square loss has been observed to perform well in classification tasks, at least as well as crossentropy. However, a theoretical justification is lacking. Here we develop a theoretical analysis for the square loss that also complements the existing asymptotic analysis for the exponential loss.

DSpace@MIT

https://hdl.handle.net/1721.1/126653

Download:  PDF icon TPR_ver2.pdf
CBMM Memo No:  112

Associated Module: 

CBMM Relationship: 

  • CBMM Funded