Evolve Quick Reviews
Public Paper Reviews
Structured preliminary reviews of recent open-access papers, generated and continuously updated by the Evolve analysis engine. Each review condenses what the paper claims, how solid the evidence is, and the questions a rigorous peer reviewer would ask — English analysis with abstracts in your language. These are independent AI analyses, not peer review; the canonical record always lives at the publisher.
Certifying Lower Bounds for Risk-Sensitive Reinforcement Learning under Adversarial State Perturbations
Tong Li, Saunak Kumar Panda, Yisha Xiang
Extends RL robustness certification to risk-averse exponential utility via convex duality; initial-state-only threat model and figure-only evaluation; no code.
No-Box Vulnerability Analysis: Description-only Detection of Indirect Prompt Injection Vulnerabilities in MCP Servers
Zehua Zhang, Jie Hu, Pratham Hegde, Aditya Maheshbhai Gabani et al.
Metadata-only no-box IPI analysis: MCPSEC recovers 94/95 confirmed vulnerable tools (98.9% recall); author-annotated labels and 65.7% precision warrant care.
Scale-Aware 3D Deep Learning for Robust Brain Metastasis Detection in Multimodal MRI
Sylvain Jaume, Hongming Wang, Simon K. Warfield
Cross-FOV late fusion of 96³/64³ 3D U-Nets raises lesion F1 and cuts false positives on 97 patients; gains credible but dev-tuned, unconfirmed externally.
Are We Really Doing Few-Shot Learning? A Critical Examination of Pre-Training Assumptions
Alejandro Galan-Cuenca, Marcelo Saval-Calvo, Antonio Javier Gallego
Shows class-disjoint in-domain pre-training inflates few-shot results by 9.66pp vs OOD; systematic, credible, but limited to 2016-18 metric architectures.
Detectable Only Where It Is Confounded: What Verified Duplication Counts Say About Membership Evidence in Language Models
Arman Nik Khah
Claims loss-based membership signals in LLMs are faint below ~1,000 verified copies and confounded above; rigorous, but rests on tiny banks and six novels.
Explainable Prediction from Mobile Sensing Data through LLM-guided Concept Integration
Yuning Wang, Iman Azimi, Amir M. Rahmani, Pasi Liljeberg
Proposes LLM-supervised concept Transformer for mobile-sensing prediction; F1 gains are marginal and OCSVM ties on PHQ-9 — promising, modestly supported.
Story Imprinting: AI Assistants Absorb Traits from Human Characters They Resemble
Jorio Cocola, Lev McKinney, Harry Mayne, Jan Betley et al.
Claims AI assistants absorb traits from similar human characters in training stories; well-controlled evidence, though some effects are fragile.
Flow Duality and Source Geometry for Categorical Generation
Etrit Haxholli
Proves argmax-projection duality from continuous to discrete flows and derives source-dependent schedules; theory is sound, evidence is a single-run pilot.
When Validation Stops Learning: Auditing Update Admission for Continual Embodied Agents
Qinzhen Ma, Ruihai Wu
Careful paired-binomial admission audit shows range-based gates freeze continual learning; evidence is toy synthetic only, no robot/VLA validation.
Project Qualia: Recovering Experiential Music Structure from Session Co-occurrence Data
Nizam Mohammed, Abu B. S. Rahman, Dimuthu D. K. Arachchige
Argues Last.fm session co-occurrence carries artist-independent structure; plausible, but evidence is qualitative with no null model or released data.
DriftNet: A Dual-Head Trajectory Transformer for Detecting and Localizing Prompt Injection in LLM Agents
Asif Pinjari, Mithun Paul Saint-Germain
Claims first joint verdict+step-label injection detector; rigorous results (F1 0.983) but self-benchmarked on synthetic, single-generator AgentDrift data.
Learning Orthogonal Multi-Index Models Beyond Small Initialization: Incremental Learning, Competitive Dynamics and Symmetry
Mo Zhou, Weihang Xu, Simon S. Du, Maryam Fazel
Proves incremental learning persists under standard init via a novel symmetrized finite-width coupling; analyzes modified gradient flow in a narrow r regime.
Harnessing Image Question Dependence for Better VLM Test-time Reinforcement Learning
Xinrui He, Ting-Wei Li, Junting Wang, Mengting Ai et al.
TTIQ uses image-question dependence as test-time RL reward for VLMs; results are strong and well ablated, but no significance tests and λ tuned on test set.
Relatively Smart II: Tractable or Semi-Supervised Instance-Optimal Learning
Shaddin Dughmi, Alireza F. Pour
Resolves open questions of DP26: ERM is relatively smart, unlabeled data removes the labeled-sample blowup, but tractable oracle learners cannot; rigorous.
ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLMs
Yizhan Li, Jianxin You, Mengyang Xiong, Yinhuan Chen et al.
Reproducible simulated benchmark scores MLLM reactive hazard decisions; strong diagnostic design, but findings rest on descriptive stats, unvalidated metrics.
Symmetry-aware super-resolution of crystal orientation maps via invariant latent-space learning
Umang Garg, Warren Zamudio, McLean P. Echlin, Samantha H. Daly et al.
Symmetry-aware latent SR lets a 27-49k-param model rival 15-16M EBSD baselines on tail/boundary metrics; credible and honest, though IN718 mean favors RCAN.
DCRA: Diffusion-Conditioned Representation Alignment for Robust Time-Series Learning
Wenrui Xu, Anas Enanaa, Keshab K. Parhi
DCRA uses diffusion forward noise as a corruption scheduler for EEG representation learning; gains are modest, single-dataset, lack significance tests.
Fixed State, Long Reach: What a Constant-Size Cache Buys Block Diffusion at Scale
Vaibhav Singh, Pierre-André Noël, Torsten Scholak, Eugene Belilovsky et al.
Claims constant-size Mamba cache gives flat latency/memory to 256k and 8–16× length reach vs attention; credible controlled 3B study, no external baselines.
Does Linguistic Structure Enrichment Enhance Coherence Assessment? Not With Current Architectures
Victor Mazzotti, Luiz Pereira, Marina Bitencourt dos Santos, Helena Maia et al.
Useful negative result: RST/POS input enrichment loses to plain text in coherence classification, but the incompatibility explanation exceeds the evidence.
The Platonic brain bridge hypothesis: human brain networks as an architectural prior for multimodal large language models
Pengfei Zhang, Biao Tian, Xiangang Li, Li Liu
Claims human cortical networks are a usable prior for omni LLMs; rigorous, but headline gains (+6.42 pp) conflate added capacity with brain-specific benefit.
ObstaDiff: Generalizable Diffusion Policy Learning via Obstacle-aware Representations
Jiawen Wang, Kevin Yao, Khalid Jawed
ObstaDiff's obstacle-aware representation boosts diffusion-policy success to 75.41% in greenhouse trials; credible ablations but no statistical testing.
SearchAtlas: Analyzing Agentic Search Strategies via Evidential Query Graphs
Jiacheng Sang, Mengyuan Li, Sanxing Chen, Yukun Huang et al.
Validated query-DAG framework exposes search-agent process failures and beats LLM judges, but gold set is small, author-built, and no significance tests.
Structurally Speaking: Motif-Oriented Graph Captioning through Bidirectional Graph-Text Translation
Hsiao-Ying Lu, Dongyu Liu, Kwan-Liu Ma
Shows structured prompting makes GPT-5.1 graph captions 9x shorter yet recoverable; diagnostic, but only 40 synthetic graphs, one model, no significance tests.
Can We Trust LLM Judges: A Study of Capability-Dependent Biases and Multi-Judge Ensemble for Bias Calibration
Gemma Zhang, Prachi Badarayani, Asmi Kumar, Sadid Hasan et al.
Finds capability-based leniency in LLM judges on absolute scoring; label-free WMV tracks oracle in simulation. Sound diagnosis, but ensemble evidence is thin.