Lyn-L / SMFH

Supervised Matrix Factorization Hashing

Paper : Supervised Matrix Factorization for Cross-Modality Hashing
Author : Hong Liu, Rongrong Ji, Yongjian Wu, Gang Hua
Published in IJCAI 2016
Contact : [e-mail]

This package contains cleaned up codes for the IJCAI paper, including:
SMFH.m: function to compute the hash code and hash function
normalize1.m: function to normalize the feature points
testWiki.m: example script to run the code on the Wiki dataset

p.s. The parameters $\mu$ and $\alpha$ should be set to 100 and 10 to achieve well performance, when the feature dimensional is high. And the number of sampling should be increased for the large-scale training set, such as 5000~10000.