SEM217: Robert Anderson, UC Berkeley: Sparse Low Rank Dictionary Learning

Tuesday, September 5th @ 11:00-12:30 PM (639 Evans Hall)

Sparse Low-Rank Dictionary Learning

Robert Anderson, UC Berkeley

Sparse Dictionary Learning (SDL) can be used to extract narrow factors driving stock returns from a stock returns matrix, provided the returns are generated by sparse factors alone.  We describe progress on a variant called Sparse Low-Rank Dictionary Learning (SLRDL), designed to simultaneously extract broad and narrow factors for the returns matrix, when the returns are generated by both types of factors.