CV
General Information
| Full Name | Aditya Chattopadhyay |
| Date of Birth | 6th July 1994 |
| Languages | English, Hindi, Bengali |
Education
-
2024 (Expected) Ph.D.
Johns Hopkins University - Advisors - Prof. René Vidal and Prof. Donald Geman
- Thesis - An information-theoretic framework for designing interpretable predictors.
-
2018 M.S. in Computational Natural Sciences
International Insitute of Information Technology, Hyderabad - Advisor - Prof. U. Deva Priyakumar
- Thesis - A probabilistic framework for constructing temporal relations in Monte Carlo trajectories.
-
2018 B.Tech. in Computer Science
International Insitute of Information Technology (IIIT), Hyderabad
Areas of Interests
- Explainable AI, Computer Vision, Optimization Methods, Probabilistic Graphical Models, Information Theory, Generative Models
Work Experience
-
Jun 2021 - Nov 2021 Applied Science Intern
Amazon - Supervisor - Dr. Himanshu Arora
- Summer internship at Amazon Visual Search
- Worked on generative models for indoor scene synthesis.
- Work published as a conference paper at WACV 2023.
-
Aug 2018 - Jun 2018 Research Assistant
Indian Institute of Technology (IIT), Hyderabad - Supervisor - Prof. Vineeth N Balasubramanian
- Worked on developing algorithms for explainable AI.
- Developed Grad-CAM++ which is now a popular method for interpreting decisions made by a CNN, publised at WACV 2018.
- Developed Causal Attributions, a method for interpretating decisions made by a deep network build from first principles of causality. This work was published at ICML 2019.
-
Jan 2017 - July 2017 Research Assistant
International Institute of Information Technology (IIIT), Hyderabad - Supervisor - Prof. U Deva Priyakumar
- Worked on developing ML methods for understanding biomolecule unfolding.
-
Jun 2015 - Aug 2015 Research Intern
WestfalischeWilhelms-Universitat, Muenster - Supervisor - Dr. Mark P Waller
- Worked on using ML methods to predict molecular structure from their Infrared Spectra.
Projects
-
2023 IP-OMP
- A sparse-coding algorithm with applications to explainable-AI for image classification problems.
-
2022 Variational Information Pursuit
- A framework for making interpretable ML decisions for classification problems.
-
2018 Causal Attributions
- A post-hoc interpretbility method for understanding deep network predictions with causal guarantees.
-
2018 Grad-CAM++
- A post-hoc interpretbility method for understanding CNN predictions.
Achievements
-
2021 - Awarded the MINDS Summer & Spring Fellowships.
-
2018 - Awarded Gold Medal for outstanding academic achievement at IIIT, Hyderabad (given to students with the highest cumulative grade point average [CGPA] in graduating class).
- Consistently in Dean’s List for all semesters in IIIT, Hyderabad which is awarded to top 15% of students in each programme.