Agentic Knowledge-Graph Extraction of Bioinformatic Research Experiment Processes
A. Vasantlal, TBD ...
Pipeline for bioinformatics research reproduction with the assistance of AI.
in progress
research · publications & active work
Papers, labs, and
things being figured out.
A mix of published, in-flight, and currently-being-written work. The official PI for each thread is named in their tile.
publications
Benchmarking Autonomous Bug Report Testing for Android Applications
A. Vasantlal, Dingbang Wang, T. Yu
Eval framework for autonomous bug reproduction tools. First Author.
in progress
MORL-A2C: Multi-Objective Reinforcement Learning Reranker for Optimizing MOPI-HFRS
Aarya Vasantlal Josh Zolla, Chuxu Zhang
Pending peer review. First Author.
submitted
research threads · active
Where I’m thinking from.
AI for Software Engineering — UConn
Research Fellow · Dr. Tingting Yu · Dr. Jinze Liu
- Bioinformatic experiment process extraction via multi-agent LLM systems
- Benchmarking autonomous bug-report testing for Android using computer-use agents
EL GATO Lab — UConn
ML Research Fellow · Dr. Derek Aguiar
- Predicting motion outcomes for civil court cases (Beta-Bernoulli, TF-IDF + Word2Vec weighted embeddings)
- Decision Tree, XGBoost, SVM classifiers; up to 65% accuracy
- Anomaly detection with Isolation Forests and One-Class SVM
MOPI-HFRS · UConn Senior Design
Project Lead · Dr. Chuxu Zhang
- Extended a static GNN food recommendation system with A2C RL across 8,170 users × 6,769 items
- Fixed 3 compounding evaluation bugs misreporting accuracy by >50%
- Health alignment 0.39 → 0.73; A2C critic baseline + BC pretraining