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.
Provable Guarantees for Spectral Structured Prediction
Violet Zheng, Jean Honorio
First guarantees for spectral decoding on noisy signed graphs; the core theory is clean, but the irregular-graph corollary looks unsound as stated.
Predicting Human Disagreement for Calibrated Dynamic Facial Expression Recognition
Yiming Wang, Frederick W. B. Li, Jingyun Wang
DFER trained on DFEW 10-vote counts with a Dirichlet–Multinomial likelihood; controlled and unusually honest, but WAR trails SOTA and gains lack significance tests.
IMVS: Interactive Medical Volume Segmentation with Test-Time Adaptation - A New Method for Annotating Radiology Datasets
Abhilaksh Singh Reen, Kushal Borkar, Ritvik Mahapatra
Claims 1.9–14.4× faster medical-volume annotation via online-adapted 2D adapter + frozen tracker; promising, honestly scoped, but n=9 user study and stats are thin.
Oncohistone inhibition reshapes tumor-microenvironment communication in Diffuse Midline Glioma (DMG)
Khairkhah, N., Ibrahim, M. M. H., Galban, S. L., Faunce, M. et al.
Inducible/reversible H3K27M models reveal tumor-neuron reshaping in DMG; strong resource but n=2/group scRNA-seq, null survival, pending data temper causal claims.
Emotion Experience, Expression, and Perception: Emotion Analysis on Multimodal Social Media Posts
Christopher Bagdon, Carina Silberer, Roman Klinger
Introduces Mult2EMo, a rigorous 7,200-post author-vs-reader emotion dataset; findings credible but models show modality collapse and data is request-gated.
Functional organization and natural scene responses across mouse visual cortical areas revealed with encoding manifolds
Dyballa, L., Field, G. D., Stryker, M. P., Zucker, S. W.
This reanalysis of the Allen Visual Coding–Neuropixels dataset builds grating-derived 'encoding manifolds' over six mouse visual areas and reports a conserved U-shaped link between spatial-frequency tuning and natural-scene preference; methodologically sound for the grating results, but the headline natural-scene claim still needs receptive-field controls and stronger statistics.
Rank and computation of the pathlifting Jacobian of a DAG ReLU network
Manon Verbockhaven (OCKHAM)
Sound elementary re-proof of rank(∂θΦ)=d−h for DAG ReLU nets plus a backprop-free einsum computation; modest novelty, weak timing baseline.
StackTok: Accelerating VLMs Inference with Budget-Adaptive Visual Token Selection
Zhenbin Wang, Lei Zhang, Lituan Wang, Wei Huang et al.
Training-free VLM token selector that tops all tested training-free baselines; principled and well specified, but margins are thin and no latency is reported.
ViD: Vision-Dominant Gender Bias Mitigation for Large Vision-Language Models
Zhipeng Zhao, Zhaoqiang Wei, Peishun Liu, Youwei Zhao et al.
Training-free attention masking curbs LVLM gender bias with large gains, but inconsistent metrics, incoherent statistics, and weak baselines temper confidence.
SAVOR: Self-Aware Visual Grounding via Confidence-Calibrated Reinforcement Learning for Multimodal Hallucination Mitigation
Zixiu Ding, Zilin Zhao, Yingjie He, Xinlang Kang et al.
SAVOR reframes MLLM hallucination mitigation as confidence calibration inside GRPO and reports striking ECE reductions across two backbones, but underspecified baseline-confidence elicitation, a statistically fragile group-ECE estimator, and a suspicious citation block temper the claims.
Revisiting Distributed Sign-Based Variance Reduction
Wei Jiang, Zechao Li, Lijun Zhang
Theory paper: sign majority voting is biased; server-side compressed gradient tracking reaches centralized-optimal distributed rates; no experiments.
Probabilistic Linear Explanations
Frederic Koriche, Jean-Marie Lagniez, Chi Tran
Probabilistic XAI via sparse anchored linear explanations: NPPP-hardness, relevance-fidelity bounds, MIP+IHT solvers; rigorous, honest, and empirically convincing.
The Automaton Underneath: The Additive Input Pathway Is a Parasitic Attractor for State Tracking in Householder Linear RNN
Gunner Levi Howe
Pre-registered ablation: deleting the additive input term flips shortcut collapse into exact state tracking in Householder RNNs; small-scale but disciplined.
A Convergence Framework for Deep $V$-Learning: Error Propagation and Sharp Action-Gap Bounds
Yury Kolomeytsev
Rigorous, sharp nonasymptotic convergence theory for deep V-learning; purely theoretical with no experiments and oracle-idealized consistency.
Label-free steering: Compressing test-time reinforcement learning into bias-only subspaces
Naveen Vakada, Mingyuan Li, Shaoxiong Ji
TTRL majority-vote rewards + ~100K bias-only steering reach near-full-FT MATH-500 scores; honest, but train-on-eval and MMAU confounds temper claims.
Accurate Trace Estimation with Fewer Random Bits via Recursive TensorSketch
Mohammad Azhar Khan, Rameshwar Pratap, Amit Sharma
Estimates implicit-matrix trace with O(p(d+m)log m) random bits via Recursive TensorSketch, polynomial-in-p variance; sound theory but no experiments.
Beyond Quadratic Loss: The Stability Phase Diagram of Adam
Gaoxiang Tang, Huanran Chen, Ziming Liu
Adam loss spikes follow a near-linear (β1,β2) stability boundary; systematic evidence, but the key left-boundary law is an empirical single-seed 1D fit.
On the Identifiability of Mixed Ordinal and Exponential Family Causal DAGs under Linear Parametric Models
Sambit Mishra, Urbashi Mitra
Proves every-parameter identifiability of mixed ordinal-exponential causal edges with sharp converses; rigorous theory, but synthetic-only validation.
Entropy in Conversational AI: Structured Unpredictability as Inferrable Interiority
Sebastian Cochinescu
Registered test of a stateful selection layer for 'structured unpredictability'; central criteria honestly fail, leaving a partly tautological lexical-novelty effect.
Dataset-Dependent Effects of Cross-Depth Aggregation and Soft-Routed Experts in EEG Foundation Model Fine-Tuning
Mingyang Jiang, Yamin Li, Daniel Moyer, Fan Ma et al.
Honest matched 3-seed study: AttnRes and soft-routed experts yield dataset-dependent, sometimes negative gains over full CBraMod fine-tuning at 2-3x overhead.
A Probe Shift Is Not a Fairness Fix: The Limits of Representation Steering in Speech Models
Nicolas Bourrel, Abderrahmane Issam, Gerasimos Spanakis
Tests whether readable speaker attributes in ASR encoders can steer away WER gaps; finds large probe shifts but <0.7pp effects, so no fairness fix.
Faking Good and Faking Bad in LLMs: Response Distortion Across Dark Triad Personality Traits
Victoria Popa, Guglielmo Cola, Caterina Senette, Maurizio Tesconi
LLMs shift Dark Triad scores under fake-good/bad framing; solid psychometrics and stats, but framing confounds untested and key numbers appear only in figures.
Function Lives Where Variance Doesn't: Task-Weighted Charts of a Language Model's Computation
Alexandre Quemy
认为下一个 Token 的函数需要 70-90% 的残差宽度，而方差不包含其中任何一部分；实验工具具有说服力，但核心数值取决于容忍度的选择。
Anchoring What Matters: A Dual-Level Learning Framework for Visually-Grounded Multimodal Reasoning
Xinxin Song, Siyuan Li, Tingxiong Xiao, Jinli Suo
PIVOT adds replay anchors and vision-weighted token advantages to multimodal RLVR with solid gains over DAPO, but single-seed evidence tempers confidence.