Trimmed Density Ratio Estimation
Song Liu · Akiko Takeda · Taiji Suzuki · Kenji Fukumizu

Mon Dec 4th 06:30 -- 10:30 PM @ Pacific Ballroom #22 #None

Density ratio estimation is a vital tool in both machine learning and statistical community. However, due to the unbounded nature of density ratio, the estimation proceudre can be vulnerable to corrupted data points, which often pushes the estimated ratio toward infinity. In this paper, we present a robust estimator which automatically identifies and trims outliers. The proposed estimator has a convex formulation, and the global optimum can be obtained via subgradient descent. We analyze the parameter estimation error of this estimator under high-dimensional settings. Experiments are conducted to verify the effectiveness of the estimator.

Author Information

Song Liu (University of Bristol)

I am a lecturer in University of Bristol. Before, I was a Project Assistant Professor in The Institute of Statistical Mathematics, Japan. I got my Doctor of Engineering degree from Tokyo Institute of Technology supervised by Prof. [Masashi Sugiyama]( and was awarded The DC2 Fellowship from Japan Society for the Promotion of Science.

Akiko Takeda (The University of Tokyo / RIKEN)
Taiji Suzuki (
Kenji Fukumizu (Institute of Statistical Mathematics)

More from the Same Authors