Home Experience Publications Projects



Projects
nthu Linear Bandits with Rotting Rewards

Jan'23-May'23

Advisor: Prof. Arpan Chattopadhyay
Formulated the problem of rotting multi-armed bandits in linear reward setting. Developed an index algorithm exploiting the problem structure and demonstrated better performance empirically.

nthu Explorations in Machine Learning (Independent Study)

Aug'21-Nov'21

Extensively studied Tensor Factorization and its use in data compression and applied it in Minimal Complexity Machine to generate a faster and more interpretable pipeline. Also developed a Separability based Classification Loss function for Neural Networks and demonstrated results on MNIST for classification task with better separation.

Code

nthu LightGBM based Botnet Detection Tool (3rd Prize among 1200+ Participants in IIT Kanpur Hackathon)

Aug'20

Developed a Botnet Detection Tool on network packet capture data using Gradient Boosted Decision Tree based LightGBM implementing feature extraction, feature design and compared it with AdaBoost and Random Forest.