zj00377 / hsc

A efficient online dictionary learning package

Stochastic-Coordinate-Coding

1. Stochastic Gradient Descent. The dictionary is efficiently updated via stochastic gradient descent. Only the support of the dictionary is updated.

2. Coordinate Decent. The sparse code is first updated via one step of coordinate descent. Then only the nonzero entries of the sparse code are updated.

3. Adaptive Learning Rate. Each coordinate of the sparse code has its own learning rate. The learning rates are updated adaptively.

Please refer to the file Readme_SCC_1.0.txt to use the scc package.

Stochastic Coordinate Coding

Authors:

Qingyang Li, email: [email protected] (Primacy contact person)

Jie Zhang, email: [email protected]

The method has been published in several top conferences in the medical image field:

Binbin Lin, Qingyang Li, Qian Sun, Ming-Jun Lai, Ian Davidson, Wei Fan, Jieping Ye. "Stochastic coordinate coding and its application for drosophila gene expression pattern annotation", arXiv preprint arXiv:1407.8147.

Jie Zhang, Jie Shi, Cynthia Stonnington, Qingyang Li, Boris A. Gutman, Kewei Chen, Eric M. Reiman, Richard Caselli, Paul M. Thompson, Jieping Ye, Yalin Wang. “Hyperbolic Space Sparse Coding with Its Application on Prediction of Alzheimers Disease in Mild Cognitive Impairment”. MICCAI. 2016.

Jinglei Lv, Binbin Lin, Qingyang Li, Wei Zhang, Yu Zhao, Xi Jiang, Lei Guo, Junwei Han, Xintao Hu, Christine Guo, Jieping Ye, Tianming Liu. “Task FMRI data analysis based on supervised stochastic coordinate coding”. Medical Image Analysis.2017.