Seyed-Mohsen Moosavi-Dezfooli

A General Framework for Black-Box Attacks Under Cost Asymmetry

The Potential of CoT for Reasoning: A Closer Look at Trace Dynamics

On the Anisotropy of Score-Based Generative Models

Information Gap in Chain-of-Thought Induces Implicit Thinking that Fails in Length Generalization

Thinking into the Future: Latent Lookahead Training for Language Models

Geometric Inductive Biases of Deep Networks: The Role of Data and Architecture

Tracing the Roots: Leveraging Temporal Dynamics in Diffusion Trajectories for Origin Attribution

LORE: Lagrangian-Optimized Robust Embeddings for Visual Encoders