This is a digit recognizer that is used to classify handwritten digits (the classic MNIST dataset). It uses machine learning to train a small convolutional neural network to achieve the goal. I utilized techniques such as dataset augmentation and early stopping in training and achieved 99.4% accuracy on the testing data.
This project based on the cat-grass classifier developed using the Gaussian linear model. This project exploits possible vulnerabilities in the linear classifier, which can also apply to the deep neural network, and can be attacked easily if the parameters are known. At the end we are able to input a perturbed image that is visually identical to the input image but the output result is mostly grass.
This project develop a cat-grass two class classifier based on maximum-a-posteriori model (no neural network or deep learning). The entire classifier is implemented in python using numpy module mainly. The result accuracy for the test image is over 90% but it does not generalize to other images.
Hi, My Name is Binhan Xu (I go by Alex). I am a software engineer at Pure Storage working on developing automated testing infrasturcture in Python for Flash Array products.