/deepSVMnet

Training deep support vector machine (SVM) networks using a novel back sign signal propagation (BSSP) learning algorithm

Primary LanguagePythonMIT LicenseMIT

deepSVMnet

Training deep rectified linear support vector machine (SVM) networks using a novel back sign signal propagation (BSSP) learning algorithm For more check my post here https://chombabupe.quora.com/Deep-Support-Vector-Machine-Networks?srid=TwSi You need TensorFlow to access the MNIST dataset or you can provide the MNIST loader yourself. The code is built from ground up with Python and Numpy. Run the svm_net_xor.py for testing the algorithm on the XOR problem Run mnist_recognizer.py to train on the MNIST dataset