Bio-Inspired Self-Supervised Learning for Wrist-worn IMU Signals
Tarale, P., Chu, K., Varghese, A., Liu, K., Xu, M. A., Iyyer, M., & Lee, S. I. (2026). "Bio-Inspired Self-Supervised Learning for Wrist-worn IMU Signals." arXiv:2603.10961.
CV
Education
- Ph.D., Computer Science, University of Massachusetts Amherst — Feb 2023 – Present
- Advisor: Dr. Ivan Lee, AHHA Lab
- M.S., Computer Science (Specialization: Data Science), University of Massachusetts Amherst, Feb 2021 – Dec 2022
- B.Tech., Computer Science and Engineering, Indian Institute of Technology (IIT) Goa, Aug 2016 – Mar 2020
Research Interests
- Machine learning for healthcare and clinical applications
- Neuromorphic computing and bio-inspired learning
- Self-supervised learning for wearable sensor signals
- Human activity recognition and digital health
Experience
- Data Science Researcher & Programmer, AHHA Lab, UMass Amherst — July 2025 – Present
- Working on applying machine learning methods to sensor data in the healthcare domain.
- PhD Research Intern, Epsilon, Boston MA — May 2024 – August 2024
- Created a new product recommendation system using Transformer-based item embeddings.
- Embeddings in vector space enabled recommendation of similar, complementary, and intermediate items.
- Gained experience with Databricks, Azure, and distributed systems; applied Transformers beyond NLP.
- Data Science Researcher & Programmer, LASER Lab, UMass Amherst — March 2022 – May 2025
- Worked on automated software verification and repairing biased models with fairness guarantees using machine learning and statistical tools.
Projects
- Activity Classification using Accelerometer Data — May 2025 – Present
- Recognizing activities from wrist-worn accelerometer data using Transformers, signal processing, and computer vision techniques.
- Software Verification using Reinforcement Learning — August 2022 – March 2024
- Used reinforcement learning to automatically generate proofs for Coq specifications.
- Achieved better problem-solving rate on the CoqGym benchmark than previous state of the art. Published at ICSE 2025.
- Biologically Inspired Vision + Auto-resizing Deep Networks — Oct 2021 – April 2022
- Applied locality and geometric constraints to generate CNN-like filters in fully connected networks, inspired by biological visual neurons.
- Achieved >60% neuron reduction on simple datasets with minimal accuracy loss.
- Progressive Functional Abstraction — March 2021 – May 2021
- Extended Progressive Functional Abstraction (PFA) to reinforcement learning agents playing games.
Publications
QEDCartographer: Automating Formal Verification Using Reward-Free Reinforcement Learning
Sanchez-Stern, A., Varghese, A., Kaufman, Z., Zhang, D., Ringer, T., & Brun, Y. (2025). "QEDCartographer: Automating Formal Verification Using Reward-Free Reinforcement Learning." Proceedings of ICSE 2025. Ottawa, Ontario, Canada.
Skills
- Programming: C++, Python, Haskell, R, HTML, CSS, JavaScript, PHP, LaTeX, MySQL, Qt, Bash, MATLAB, Scilab, GNU Octave
- Concepts: Data Structures & Algorithms, Machine Learning, Statistics, Simulation
- Management: Leadership, Relationship building, Community service
Volunteer & Leadership Experience
- Teaching Assistant, UMass Amherst & IIT Goa — Various courses, involving 500+ students in total
- Facilitator/Instructor, Google Explore ML — Taught ~100 students via the Explore ML track offered by Google
- Co-Founder, ML Circle, IIT Goa — Created a community/platform to connect researchers and students for learning and collaboration
- Election Officer, IIT Goa — Built software to conduct student council elections; oversaw the election process
- Food Distribution/Prep Volunteer — Amherst Survival Center and First Baptist Church, Amherst
Awards & Honors
- 2020 Winner, MIT Covid-19 Challenge Hackathon (MIT)
- 2019 Technical Excellence Award (IIT Goa)
- 2019 2nd Position, Smart India Hackathon (Govt. of India)