Training-free token reduction methods for Vision Transformers (ToMe, ToFu, PiToMe, and MCTF) employ different scoring mechanisms, yet they share a closely matched cliff-like collapse at high compression. This paper explains \emph{why}. We develop a diagnostic framework with two tools, ranking consistency $ρ_s$ and off-diagonal correlation $ρ_\text{off}$, that decomposes the collapse into (1)a signal-agnostic error amplifier inherent to layer-wise reduction, predicting convex Pareto curves and $r_{\text{crit}} \propto 1/L$; and (2)shared reliance on \emph{pairwise} similarity signals whose ranking consistency degrades from $ρ_s{=}0.88$ to $0.27$ in deep layers. Pairwise rankings are inherently unstable ($O(N_p^2)$ joint perturbations) while unary signals enjoy greater stability ($O(N_p)$ perturbations, CLT). From three design principles derived from this diagnosis, we construct CATIS as a constructive validation: unary signals raise the trigger threshold, triage suppresses the gain. On ViT-Large at 63% FLOPs reduction, CATIS retains 96.9% of vanilla accuracy (81.0%) on ImageNet-1K where all baselines collapse to 43--65%.
AES Systems Are Both Overstable And Oversensitive: Explaining Why And Proposing Defenses
Adobe Media Data Science Research(Adobe媒体数据科学研究院);IIIT-Delhi(德里印度信息技术学院);SUNY at Buffalo(纽约州立大学布法罗分校);New York University(纽约大学);University of Texas at Austin(德克萨斯大学奥斯汀分校)
An application of incomplete pairwise comparison matrices for ranking top tennis players
Institute for Computer Science and Control, Hungarian Academy of Sciences (MTA SZTAKI)(匈牙利科学院计算机科学与控制研究所(MTA SZTAKI));Corvinus University of Budapest(布达佩斯考文纽斯大学)
Norms at a Price: Why RL-Based Alignment Can Promise Conditional Compliance at Best
German Research Center for Artificial Intelligence (DFKI)(德国人工智能研究中心(DFKI));Hamburg University of Technology (TUHH)(汉堡工业大学);University of Oxford(牛津大学);Oxford Internet Institute(牛津互联网研究所)
An Explainable Market Integrity Monitoring System with Multi-Source Attention Signals and Transparent Scoring
Department of Computer Science University of New York(纽约大学计算机科学系);DeepMind(深度Mind);Mila, Université de Montréal(蒙特利尔大学Mila);Google Research(谷歌研究);CIFAR Fellow