top of page

Projects

Deepfake Detection

Used computer vision and machine learning to classify deepfake videos

​

  • Developed a state-of-the-art pipeline using a convolution LSTM neural network and optical flow to detect deepfakes

  • Generated and preprocessed a large dataset for model training and validation

​

Paper

Github Link

deepfakecoffee_2-10_edited_edited_edited
opticalFlow_vector.PNG

NBA Rookie of the Year Prediction                                                                     

Used machine learning to predict the NBA ROY for 2020

  • Scraped player statistic from the web and determined the best predictor stats through feature engineering

  • Created logistic regression, KNN, and neural network models that accurately predicted the ROY for the season

​

Paper

Github Link

nba_edited.jpg

VAE Dog Image Classification

Using a variational autoencoder to classify images of dogs

  • Achieved 85.5% accuracy classify dog images vs. images of other animals

  • Able to train model without having to label data

​

Github Link

bottom of page