Machine Learning for BioInformatics

Extracting principal features of sickness symptoms from prior data of patients, using probabilistic models (GMM, EM) to classify patients.

Brain tumor diagnosis

Tumor detection by CNN (VGG16 net) and visualizing essential factors of results by Grad-CAM.

Hearbeat Classification

Comparison between LSTM and CNN methods to classify heartbeats from MIT-BIH Arrhythmia Database, implementation the best clustering method.

Digit Generator by GAN (Generative Adversarial Networks)

Implementing Generative Adversarial Networks to generate artificial digit images which are barely distinguished with real handwritten digits.