Music Recommendation

Moody Lyrics
NEW

A web application that feteches song lyrics for a given song and assigns a predicted mood to it by analyzing the lyrics with BERT.

Lyric-Based Playlist Continuation with BERT

Studied the effectiveness of lyric-based features in music recommendation with mood classification of song lyrics using BERT, and recommendation of tracks using a two-stage approach with matrix factorization and XGBoost

Neural Collaborative Filtering for Playlist Continuation

Developed a neural collaborative filtering-based recommender system leveraging the Spotify Million Playlist Dataset, incorporating playlist-track interactions and track metadata to deliver ranked recommendations.

Computer Vision

Brain Tumor Detection and Segmentation in MRI

Constructed a deep learning model that uses the VGG16 and Mask R-CNN architectures with transfer learning to detect the presence of tumors in MRI images and locates them through pixel-based segmentation.

Time Series Forecasting

Trend Detection in Stock Prices

A web application that scrapes real time stock market data, detects the current trend (Uptrend/Downtrend) of a given stock and attempts to predict the next day’s closing price.

Computer Architecture

Instruction Branch Prediction with Perceptron

Utilized machine learning techniques to improve instruction branch prediction accuracy against existing approaches - GShare and Tournament.

Computer Networks

Network Intrusion Detection

Explored machine learning techniques to detect malicious attacks in network traffic of a system and further classifiy the type of attack.

Moody Lyrics NEW