Brief

This paper investigates the computational complexity of verification problems for Binarized Neural Networks (BNNs). Researchers demonstrated that BNN satisfiability is NP-complete by reducing it from the Boolean satisfiability problem (SAT). Additionally, they found that uniform image occlusion results in a piecewise-constant network output, allowing for a polynomial-time algorithm to check robustness. AI

Researchers have developed a method to analyze the internal trajectories of reasoning-trained language models, distinguishing between simply taking more steps and following different computational paths. By adjusting for generation length, they found that model difficulty correlates with corrected trajectory geometry, particularly in coding tasks where harder problems show more direct paths in reasoning models compared to standard instruction-tuned models. This distinction was also observed, though less pronounced, in mathematics and Boolean satisfiability problems, suggesting reasoning training can indeed alter a model's internal processing distinct from mere length. AI