(original) (raw)

We thank the help from the following group members and users. Chih-Wei Hsu and Jen-Hao Lee give helpful comments in the early development. Ryszard Czerminski and Lily Tian give useful comments on LIBSVM implementation documents. Carl Staelin from HP labs initiated the development of the python interface. Li-lun Wang helps the integration of python interface. Hsuan-Tien Lin helps adding the decision value subroutines in libsvm 2.5. Ting-Fan Wu helps updating the python interface of libsvm 2.5 and 2.6. Li-lun Wang helps to test libsvm 2.6. David James (University of Toronto) helps to initiate the stratified cross validation code (in libsvm 2.7). Fang-Lan Huang helps the stratified CV implementation in Java. S V N Vishwanathan (Australian National University) helps to modify the python interface so input data with zero values are skipped. (libsvm 2.7) Rong-En Fan helps to test libsvm 2.7. Tzu-Hsiang Kao helps to develop the new faq (2004/8). Koen Tanghe from Ghent University, Belgium helps to point out an error on handling #SV=0 case in loading model. Rong-En Fan helps to modify many things from 2.71 to 2.8. Rong-En Fan helps to write part of the script subset.py for 2.81. Pei-Chin Wang helps to prepare a code of the precomputed kernel in libsvmtools. The code is improved and merged to libsvm 2.82. Ed Johnson from San Francisco State University helps to point out a memory leak in svm_binary_svc_probability(). Rong-En Fan helps to write the script `tools/checkdata.py' for checking input data (available after libsvm 2.85). Alain Lifchitz from LIP6, France suggests some useful changes from 2.85 to 2.86. Ding-Yi Chen of redhat and Hsiang-Fu Yu of NTU help to update svm-toy/qt from qt3 to qt4 (2.87) Rudi Cilibrasi and Chen-Tse Tsai help to prepare debian package of libsvm Chia-Hua Ho and Hsiang-Fu Yu of NTU help to update tools/*.py so that they can be run under python 3.0. Hsiang-Fu Yu helps to add the code for quiet mode (-q). Patrick Blitz at CMU proposes overloading svm_load_model to read from a BufferedReader directly. Hsiang-Fu Yu writes the new Python interface using ctypes (version 2.91). Christian J. Dietrich at University of Applied Sciences Gelsenkirchen proposes adding exit status for checkdata.py. Chia-Hua Ho helps to merge libsvm matlab interface to the standard libsvm. He also helps to handle many minor updates for version 3.1 Vebjorn Ljosa (MIT) and Chia-Hua Ho help to add subroutines in python interface for accessing SVs (version 3.1) Tse-Ju Lin and Chia-Hua Ho helped to add save/load regression data in svm-toy (version 3.12). Yong Zhuang helped to add the support of obtaining support vector indices and add -q in svm-predict.c (version 3.13) Hsiang-Fu Yu writes a new subset.py (version 3.13) Yu-Chin Juan, Hsiang-Fu Yu, Chia-Hua Ho helped to revise grid.py (version 3.15) Yong Zhuang, Yu-Chin Juan developed a regression test code for libsvm and Tzu-Ming Kuo helped to validate version 3.17 Wei-Cheng Chang and Tzu-Ming Kuo helped the release of 3.18 Bo-Wen Yuan and Yong Zhuang helped to fix bugs and prepare the release of 3.19 and 3.20. Hsiang-Fu Yu and Pin-Yen Lin helped to add scaling code in python for 3.23 Jui-Nan Yen and Yu-Sheng Li helped to create the installation through PyPI for version 3.25. Zhongyi Que added the functionality of one-class probabilistic outputs. Li-Chung Lin, Yaxu Liu, and Cheng-Hung Liu helped to prepare the release of version 3.3. He-Zhe Lin helped to update the java applet and Guan-Ting Chen helped to prepare the release of version 3.31. Kuan-Ting Chen helped to prepare the release of version 3.33.