-Handwritten-Digit-Recognition-using-Machine-Learning-and-Deep-Learning-Python-3.5-MINIST-Tensorfl

IMPLEMENTED and COMPARED the performance of the following different classifiers and implemented it on handwritten digits of the MINIST dataset • Linear Classifier • K-nearest neighbor classifier • Radial Basis Function (RBF) neural network • One-hidden layer fully connected multilayer neural network • Two-hidden layer fully connected multilayer neural network
All above mentioned algorithms are successfully applied on handwritten digit’s Mnist dataset. These results appeared to be state of art in handwritten digit recognition. Architecture of each algorithm on dataset fits well and the accuracy of each classifier is show below in the table. Every algorithm provides impressive performance comparatively but when you consider accuracy of each classifier KNN comes out as superior. If you Consider time taken by each classifier as well as accuracy into account then Random Forest fits well as compared to KNN.