Aritra Das

Repeated paraphrasing can significantly enhance the indistinguishability of AI-generated text from human writing, challenging existing detection methods.

Maximizing $L_p$-norms over zonotopes is W[1]-hard for all fixed rational $p \in (1, \infty)$, revealing deep computational challenges in neural network sensitivity analysis.

A novel curation pipeline can optimize regression evaluation sets by maximizing capability coverage while adhering to strict query limits.

Robust generalization isn't as hard as you think: it only tweaks, rather than revolutionizes, the Lipschitz constant needed for smooth interpolation.