Hang Li

US Patent:

20130325879, Dec 5, 2013

Filed:

Aug 9, 2013

Appl. No.:

13/963924

Inventors:

Hang Li - Los Angeles CA, US
Hua Zheng - Beijing, CN
Ting-hao Yang - Shanghai, CN
Richard W. Tom - Marina Del Rey CA, US

Assignee:

Hulu, LLC - Santa Monica CA

International Classification:

G06F 17/30

US Classification:

707748, 707758

Abstract:

A method and apparatus for recommending a media program of a set of media programs to a user of a set of users is disclosed. The method and apparatus computes a measure wof the implied similarity of a first media program (i) and a second media program (j) that corrects for the popularity of the media programs, thus resulting in a more accurate indication of the relatedness of the media programs.

US Patent:

20190313155, Oct 10, 2019

Filed:

Apr 4, 2018

Appl. No.:

15/945675

Inventors:

- Santa Monica CA, US
Colin Zhou - Los Angeles CA, US
Matthew Holcombe - Los Angeles CA, US
Atul Arun Phadnis - Duarte CA, US
Hang Li - Los Angeles CA, US
Sridhar Srinivasa Subramanian - Culver City CA, US
Kristen Huff - Los Angeles CA, US

International Classification:

H04N 21/466
H04N 21/442
G06N 3/08
G06N 3/04

Abstract:

In one embodiment, a method separates subscriber features generated from subscriber interaction with a video delivery service into feature dimensions and inputs the feature dimensions into a respective prediction network. Each prediction network is trained to output a respective dimension score. The method outputs dimension scores using parameters in the plurality of prediction networks that are trained using a variance term to control a variance of the plurality of feature dimensions and using a de-correlation term to control a correlation of the plurality of feature dimensions. The dimension scores are combined into a retention prediction score and an action is performed on the video delivery service for the subscriber based on the retention score.