Projects

@ CMU

Activity Localization using Hard Attention (Spring 2017)

Advisor: Prof. Abhinav Gupta
Designed a neural network architecture to temporally localize actions using weak supervision. Slides

@ IIT-Bombay

Texture Classification (Autumn 2014)

Advisor: Prof. Suyash Awate
Designed and implemented image categorization and segmentation based on visual textures through a texton-based learning approach to model and represent visual textures. Supervised Learning performed using SVM and Decision Tress to create model for the database. Report

Virtual Memory for Experimental OS (Spring 2014)

Advisor : Prof. Dhananjay D Dhamdhere
Designed and implemented an effective data structure and algorithm to handle memory allocation, swap space management, on Pranali, a virtual OS built on top of Linux. Report

ANN for Twitter Sentiment Analysis (Spring 2014)

Advisor : Prof. Phuskpak Bhattacharya
Implemented a Artificial Neural Network to predict sentiments of tweets on twitter. Overall accuracy was 78%. We used bag of words as the feature vector to train a 3 layer neural network. code

Sequence Alignment on GPUs (Autumn 2013)

Advisor : Prof. Bernard Menezes
Implemented a Sequence Alignment problem on GPUs with parallel version of Needleman- Wunsch algorithm. Investigated Parallel Prefix and Diagonal based approach to solve the prob- lem. Achieved O(n) complexity as compared to qudaratic of the Serial Version. code