Postdoc, Harvard University
1 paper at NeurIPS 2025
We demonstrate an efficient interactive verification protocol that allows resource-constrained parties to trust and reliably check the quality of expensive attribution models provided by powerful, untrusted parties.
