Spectral Algorithm for Shared Low-rank Matrix Regressions
ID:54 View Protection:ATTENDEE Updated Time:2020-08-05 10:17:00 Hits:435 Oral Presentation

Start Time:2020-06-09 14:40(Asia/Shanghai)

Duration:20min

Session:R Regular Session » R04Computational and Optimization Techniques for Multi-Channel Processing

Video No Permission

Tips: Only the registered participant can access the file. Please sign in first.

Abstract
We consider multiple matrix regression tasks that share common weights in order to reduce sample complexity. For this purpose, we introduce the common mechanism regression model which assumes a shared right low-rank component across all tasks, but allows an individual per-task left low-rank component. We provide a closed form spectral algorithm for recovering the common component and derive a bound on its error as a function of the number of related tasks and the number of samples available for each of them. Both the algorithm and its analysis are natural extensions of known results in the context of phase retrieval and low rank reconstruction. We demonstrate the efficacy of our approach for the challenging task of remote river discharge estimation across multiple river sites, where data for each task is naturally scarce. In this scenario sharing a low-rank component between the tasks translates to a shared spectral reflection of the water, which is a true underlying physical model. We also show the benefit of the approach in the setting of image classification where the common component can be interpreted as the shared convolution filters.
Keywords
Phase Retrieval; Low-rank optimization; Multitask Learning
Speaker
Yotam Gigi
Google Research & Hebrew University (HUJI), Israel

Submission Author
Yotam Gigi Google Research & Hebrew University (HUJI), Israel
Sella Nevo Google Research, Israel
Gal Elidan Google Research and HUJI, Israel
Avinatan Hassidim Google, Israel
Yossi Matias Google, Israel
Ami Wiesel The Hebrew University of Jerusalem, Israel
Submit Comment
Verify Code Change Another
All Comments
Important Date
  • Conference Date

    Jun 08

    2020

    to

    Jun 11

    2020

  • Jan 12 2020

    Draft paper submission deadline

  • Apr 15 2020

    Early Bird Registration

  • Dec 31 2020

    Registration deadline

Sponsored By
IEEE Signal Processing Society
Contact Information