Publications

A Non-intrusive Machine Learning Solution for Malware Detection and Data Theft Classification in Smartphones

Published in The 21st International Conference on Computational Science (ICCS 2021) Kraḱow, Poland, 2021

Lead a team towards condensing one of the largest datasets (10 billion+ records) for malware classification in smartphones to 0.1% of its feature set whilst retaining threshold accuracy reported by competing data intensive methods through efficient data preprocessing and online learning.

Code

Real-time Surveillance based Crime Detection for Edge Devices

Published in The 15th International Conference for Computer Vision Theory and Applications (VISAPP 2020), Valleta, Malta

I lead a team to implement efficient multiple instance learning on optical flow by introducing early stopping and intuitive undersampling to tackle otherwise computationally expensive crime detection for video. Deployed and tested on a resource constrained device: Raspberry Pi3 (ARM v8) performing 2.3x times faster then the benchmark servers (Intel Xeon) using just a third of the number of cores.

Code

A Two-Stage Machine Learning Approach to Forecast the Lifetime of Movies in a Multiplex

Published in The Future of Information and Communication Conference (FICC 2020), San Francisco, USA.

Developed a two-stage model to estimate the number of weeks a movie in a multiplex would successfully run based on past occupancy/crowd behaviour. The solution is currently in use by one of India’s leading cinema exhibitors.

Code