CUDA_SVM

• step 1: Implement SVM based on LIBSVM, in python;

• step 2: Impelement sequential SVM, Linear kernel; (Using the same data input and output processing)

• step 3: Implement CUDA based SVM, Linear kernel; (Using the same processing other than the optimzation part in step 3)

GPU Implementation of SVM

• GPUSVM: https://github.com/niitsuma/gpusvm
• SVM CUDA implementation: https://github.com/Site1997/SVM-Cuda
• LibSVM CUDA: https://mklab.iti.gr/results/gpu-accelerated-libsvm/

Sequential Minimal Optimization

• Original Paper (1998): https://www.microsoft.com/en-us/research/publication/sequential-minimal-optimization-a-fast-algorithm-for-training-support-vector-machines/
• Textbook (1999): [Platt J.C., Fast training of support vector machines using sequential minimal optimization, In: Advances in kernel methods – support vector learning, Sch lkopf, B. and Burges, C.J.C and Smola, A.J. (Eds.), 185-208, MIT Press, Cambridge, MA, USA, 1999](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/smo-book.pdf>
• Python Implementation: <https://jonchar.net/notebooks/SVM/)
• C++ Implementation: https://medium.com/@dr.sunhongyu/machine-learning-c-svm-support-vector-machine-simple-example-deff5d55d43e

Dataset

• UCI Machine Learning Repository: https://archive.ics.uci.edu/ml/datasets.php
  • Baseline: Glass，iris, wine, sonar, breast-cancer, adult
  • Statlog Data: heart, letter, shuttle
• Usps: https://git-disl.github.io/GTDLBench/datasets/usps_dataset/
• Web: https://archive.ics.uci.edu/ml/datasets/Anonymous+Microsoft+Web+Data
• Mnist: https://git-disl.github.io/GTDLBench/datasets/mnist_datasets/