Luc Brogat-Motte (Télécom Paris): Reduced-rank Regression in Structured Prediction

Graduate Student - 02, LIKE22: videos of early stage researchers

We propose and analyse a reduced-rank method for solving surrogate problems in structured prediction under output regularity assumptions. We give learning bounds for our method, and show that statistical performance are improved in comparison to full-rank method. Our analysis extends the interest of reduced-rank regression beyond the standard low-rank setting to more general output regularity assumptions. We assess its benefits on two different problems: multi-label classification, and metabolite identification.