NeurIPS 2026 reviews are up: reviews, scores, rebuttals and meta-reviews for 7,877 papers. See the highest rated
acceptodds: the papers of ICLR 2027
3,952 results for “keyword:"ad"” in ICLR 2027. 1,017 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]From Fooling CNNs to Fooling MLLMs: Generating Unrestricted Adversarial Examples via Diffusion-Based Style Injection▼5.8k2.33
[PDF]An Infinite Information Exponent that Gradient Descent Still Learns at the Optimal Rate: The Value Catalyst in Softmax Attention▼71.25
[PDF]Recurrent Refinement Moves Candidate-Quality Readability Earlier and Stabilizes Its Coordinates▼1.2k1.67
[PDF]Frames Don't Add Up: A Factorial Causal Audit of Contextual Evidence Utility in Video-Language Models▼1361.33
[PDF]HARPS: Mitigating Multimodal Hallucinations via Head-Selective and Adaptive Perceptual Sink Reweighting▼3.7k2.25
[PDF]Equilibrium of Thoughts: Reasoning as Iterative Stabilization Under Adversarial Probing▼4.3k2.00
[PDF]DO STATE SPACE MODELS NEED SEPARATE B AND C PROJECTIONS? REVISITING B-C TYING FROM S4D TO MAMBA▼1.2k1.67
[PDF]One Fuzzer, Many Reasoners: Adaptive Neuro-Symbolic Fuzzing with Reliability-Calibrated and Context-Gated Guidance▼1361.33
[PDF]Normative Rewriting on the Reading Axis: How to Evade Chinese AI-Text Detectors While Preserving Quality▼1361.33
Withdrawn or desk rejected (1,017)
[PDF]FairWatch: Interaction Topology as a First-Order Variable Shaping Safety and Fairness in Multi-Agent DeliberationDesk rejected2.00