Sarcasm Detection Using Deep Learning - January 2019 - May 2019
Technology/Languages: Neural Networks, Web Scraping, Python, Data Engineering
Led a team of 4 to build a deep learning model that accurately detected sarcasm in text-based input with an accuracy of over 70% by utilizing sentiment, punctuation, and emoticons as features. The model was trained using a dataset scraped from product reviews on Twitter.
MRI scan segmentation - August 2022
Technology/Languages: Python, Deep Learning, Jupyter Notebooks, Transfer Learning
Successfully utilized a U-Net Model with transfer learning to segment MRI scan images and detect brain tumors, resulting in a high accuracy of 97% and an Intersection over Union (IoU) score of 0.83.
Image Colorizer - November 2022
Technology/Languages: Python, Deep Learning, Jupyter Notebooks, Transfer Learning
Designed a cutting-edge Convolutional Neural Network to colorize black and white images, further enhancing the results through transfer learning to attain an SSIM value of 92%.
AI for games - January 2023
Technology/Languages: Python, Deep Learning, Jupyter Notebooks, reinforcement Learning
Used neural networks to devise an AI that could play tic tac toe and connect 4 with a win rate of 80%. The AI could go toe-to-toe with a human opponent by using reinforcement learning to help it make decisions at each stage.
Comparative Study on GANs and VAEs - April 2023
Technology/Languages: Python, Deep Learning, Jupyter Notebooks, Transfer Learning
Worked on a comparative study of the performance of different Generative Adversarial Networks (GANs) and Variational Auto Encoders (VAEs) on the MNIST dataset. The performance difference each model was calculated and visualized using various plots to complete the study.