PhD student, Illinois Institute of Technology
1 paper at NeurIPS 2025
We introduce ACIA, a measure-theoretic framework for anti-causal representation learning through two-level abstraction, supporting both perfect and imperfect interventions with theoretical guarantees.
