Pith. sign in

INDEX

Every paper Pith has read and judged, in one searchable index.

95,987 reviewed papers in cs · newest first · page 13

Pith rank means recent reader upvotes with each upvote's ranking weight halved after 24 hours. The archive itself is sorted by arXiv submission date.

  1. cs.CL arXiv submitted 2026-08-03
    Two senses of one word diverge in predictions

    K. Jack Scott +2 · “Divergent large language model predictions from convergent representations in ambiguous word pairs”

    2608.01816
  2. cs.CL arXiv submitted 2026-08-03
    Few synthetic probes reveal hidden coupling in LLM judge rubrics

    Divyansh Singh +2 · “RADAR: Rubric-Aware Dependency and Redundancy Analysis for LLM-as-Judge Evaluation”

    2608.01810
  3. cs.CV arXiv submitted 2026-08-03
    Second-opinion gate checks anatomy net on just 9% of X-rays

    Siam Tahsin Bhuiyan +6 · “SecondOpinion: Anatomy-Aware Gated Reasoning for Efficient Medical Image Analysis”

    2608.01808
  4. cs.CV arXiv submitted 2026-08-03
    Per-target fusion and calibration top four RGBT benchmarks

    Zhaoding Ding +4 · “Parameter-Dynamic Adaptive Fusion and Calibration Network for RGBT Tracking”

    2608.01807
  5. cs.RO arXiv submitted 2026-08-03
    Flying robot runs impedance and admittance control at once

    Zicheng Luo +5 · “Hybrid Impedance-Admittance Control with Multi-Link Aerial Robot for Contact-Rich Surface Sliding Task”

    2608.01800
  6. cs.SD arXiv submitted 2026-08-03
    Four frozen speech models catch every unseen deepfake

    Seunghyun Kim +2 · “Multi-Backbone Self-Supervised Ensembles for Audio Deepfake Detection and a Cross-Track Analysis of Generation-Detection Asymmetry”

    2608.01796
  7. cs.CV arXiv submitted 2026-08-03
    Identity-aware training lifts multimodal embeddings by ~25 points

    Jiawei Cao +7 · “Illuminating Visual Identity in Universal Multimodal Embeddings”

    2608.01794
  8. cs.AI arXiv submitted 2026-08-03
    Benchmark reveals which VLMs overstate extraction confidence

    Priyashree Roy +8 · “Can You Trust the Confidence? ConfBench for Vision-Language Models on Document Extraction”

    2608.01792
  9. cs.LG arXiv submitted 2026-08-03
    Reshape token geometry first: energy models reach 98.6% AUROC

    Camile Lendering +3 · “ReFP-AD: Rectified Flow Preconditioning for Energy-Based Anomaly Detection”

    2608.01793
  10. cs.ET arXiv submitted 2026-08-03
    LLM agent writes all 48 photonic-design scripts

    Xiaohan Jiang +3 · “PICopilot: An LLM-based Agentic Framework for Assisting Photonic Integrated Circuit Design via Script Generation”

    2608.01791
  11. math.NA arXiv submitted 2026-08-03
    Gradient penalties force a log term no power series can match

    Marin Mi\v{s}ur · “Logarithmic Resonance at Mixed Boundary Junctions in Gradient-Dependent Semilinear Equationsv”

    2608.01790
  12. cs.NE arXiv submitted 2026-08-03
    Alpha mining's bottleneck: noisy fitness

    Xinwei Yu +5 · “Towards Autonomous Formulaic Alpha Discovery: An Evolutionary Computation Perspective”

    2608.01789
  13. eess.SP arXiv submitted 2026-08-03
    Radar sensing in CBRS favors hybrid classical-ML designs

    Madan Baduwal +1 · “Radar Detection in the CBRS Band: Techniques, Challenges, and Future Directions”

    2608.01786
  14. cs.DC arXiv submitted 2026-08-03
    Single-GPU omni serving: SLO attainment 4.9x higher

    Yuning Zhang +1 · “HorizonServe: Coordinating Request Scheduling with GPU Sharing for Omni-Model Serving”

    2608.01785
  15. cs.AI arXiv submitted 2026-08-03
    Mixture-of-experts cuts expert compute 15% in diffusion models

    Xiang Xia +5 · “REFLEX: Rethinking MoE Inference as Refinement-Aware Compute Allocation in Diffusion Language Models”

    2608.01784
  16. cs.SD arXiv submitted 2026-08-03
    Few-shot chain-of-thought matches teachers on music-essay scoring

    Baicheng Lin +2 · “Comparative Validation of GPT-4o-mini and Teacher Mean Scores for Automated Scoring of Music Analysis Responses: Single-Pass Deployment, Repeatability, and Strategy-Specific Bias”

    2608.01783
  17. cs.CV arXiv submitted 2026-08-03
    Comics surface far more social bias than single photos

    Junyeong Park +7 · “Investigating Social Bias in Narrative Image Generation”

    2608.01780
  18. cs.NE arXiv submitted 2026-08-03
    Pareto-optimal surrogate ensemble wins 251 of 300 matchups

    Kei Nishihara +2 · “An Evolutionary Algorithm Assisted by an Ensemble of Pareto-Optimal Surrogate Models”

    2608.01777
  19. cs.LG arXiv submitted 2026-08-03
    Bounded corrections sharpen flood-plateau timing

    Liangjun You +3 · “Multi-Source Dynamic Graph Learning for Compound-Flood Forecasting in Managed Coastal Systems”

    2608.01775
  20. cs.AI arXiv submitted 2026-08-03
    Skill-evolution loop lifts audit pass rate 9 points without retraining

    Xuhui Wang +6 · “FRAMES: Guarded and Dual-Objective Skill Evolution for Agents in Policy-Governed Enterprise Workflows”

    2608.01772
  21. cs.CV arXiv submitted 2026-08-03
    First real-time diffusion video relighting hits 15.78 fps

    Yue Ma +10 · “LiveLight: Real-time Streaming Video Relighting with Interactive Control”

    2608.01771
  22. cs.AI arXiv submitted 2026-08-03
    AI-guided food inspections beat manual plan: 11% vs 9%

    Dongqi Wang +3 · “Leveraging AI for fine-grained food safety risk forecasting in sparse data conditions”

    2608.01767
  23. cs.CR arXiv submitted 2026-08-03
    Logic check lifts verified LLM paths from 78% to 98%

    Kaustuv Mukherji +6 · “EntailLLM: Verifying LLM-Generated Vulnerability Discovery Paths with Domain Knowledge via Logic Programming”

    2608.01763
  24. cs.CV arXiv submitted 2026-08-03
    Decoupled 3DGS renders interactive driving sims at 45 FPS

    Siying Li +4 · “DecoupleGS: Interactive 3D Gaussian Splatting for End-to-End Autonomous Driving Testing”

    2608.01761
  25. cs.CV arXiv submitted 2026-08-03
    Superpixel-guided scanning restores weather-degraded images to 34.15 dB

    Dayu Li +5 · “Pixel Ignores, Superpixel Sees: Adverse Weather Image Restoration via Semantic-Center SSM”

    2608.01760
  26. cs.CR arXiv submitted 2026-08-03
    Harmless tasks add up to an 86% attack success on self-evolving agents

    Bingyu Yan +5 · “Benign Alone, Harmful Together: Exploiting Experience Composition in Self-Evolving LLM Agents”

    2608.01759
  27. cs.CV arXiv submitted 2026-08-03
    Stacking three AI vision models lifts blight detection over any single model

    Xiaohao Yang +4 · “Can Urban Blight Be Accessed with Vision-language Models: A Case Study in Detroit”

    2608.01753
  28. cs.CV arXiv submitted 2026-08-03
    All-band router lifts satellite segmentation accuracy 2.9 points

    Xingyan Li +4 · “SPECTRA: Band-Routed Embedding and Stage-Wise LoRA for Cross-Sensor Fine-Tuning of Geospatial Foundation Models”

    2608.01751
  29. cs.GR arXiv submitted 2026-08-03
    Ptychography can drop strict linearity and gain resolution

    Yun Xie +4 · “High-fidelity tabletop nanoscopy enabled by non-linear spectral preconditioning”

    2608.01746
  30. cs.LG arXiv submitted 2026-08-03
    Virtual queues pace network budgets inside every episode

    Yeonseo Jeong +3 · “Heterogeneous Multi-Agent Reinforcement Learning for Radio Resource Management under Coupled Finite-Horizon Constraints”

    2608.01745
  31. cs.AI arXiv submitted 2026-08-03
    RL finds interlocking assemblies where heuristic search stalls

    Xuyang Ma +5 · “RL-Lock: Reinforcement Learning for Generating Interlocking Assemblies”

    2608.01744
  32. cs.LG arXiv submitted 2026-08-03
    KL regularizer that ignores wrong answers stops LLM forgetting

    Li Wang +6 · “Toward Plasticity-Preserving KL Regularization for Capability Retention in LLM Reinforcement Learning”

    2608.01743
  33. cs.AI arXiv submitted 2026-08-03
    Dual-track memory lifts LLM agent QA accuracy by up to 8.8%

    Yufei Luo +2 · “MemSIF: From Structured Interactions to Dual-Track Fact Memory for LLM Agents”

    2608.01742
  34. cs.LG arXiv submitted 2026-08-03
    Two forecasts' disagreement speeds up diffusion sampling

    Yanchao Li +8 · “Disagree to Accelerate: Closing the Loop on Diffusion Feature Forecasts”

    2608.01740
  35. cs.RO arXiv submitted 2026-08-03
    Drone that grips its target holds still within 3 mm

    Joshua Taylor +4 · “A Tilt-Rotor UAV with a Gripper for Stable Contact-Based Tasks via Environmental Anchoring”

    2608.01736
  36. cs.AI arXiv submitted 2026-08-03
    Cut privilege illusion

    Jianyu Wu +4 · “DAPD: Dual-Anchored Policy Distillation”

    2608.01735
  37. cs.LG arXiv submitted 2026-08-03
    LLM-chosen drug neighbors top complex baselines in cell-line tests

    Betty Xiong +4 · “LLM-Guided Retrieval for Prediction of Molecular Perturbation Responses”

    2608.01734
  38. cs.RO arXiv submitted 2026-08-03
    Robots learn to use arms and chest from a wearable twin

    Takahide Kitamura +3 · “TWINS: A Tactile Wearable Isomorphic Arm Networked System for Contact-Rich Manipulation Learning”

    2608.01733
  39. cs.IR arXiv submitted 2026-08-03
    Graph-grounded LLM explanations beat popularity ranking by 17.1%

    Meenakshi Rajpurohit +1 · “X-KGRank: A Knowledge Graph RAG Framework for Explainable Recommendations via Pattern Mining and LLM Re-Ranking”

    2608.01732
  40. cs.IR arXiv submitted 2026-08-03
    Repeated best responses make recommendation lists mutually optimal

    Yoji Tomita · “MODE: Mutual Optimality in Direct Effects of Reciprocal Recommendations in Matching Markets”

    2608.01731
  41. cs.CV arXiv submitted 2026-08-03
    Image-quality model tops benchmarks at every resolution

    Hakan Emre Gedik +2 · “Learning Where to Look and How to Judge: Resolution-agnostic Image Quality Assessment with Quality-aware Saliency”

    2608.01730
  42. cs.NI arXiv submitted 2026-08-03
    Ten-factor score reorders CVSS in a synthetic SD-WAN test

    Saeed Alam · “Predictive Exposure and Cryptographic Readiness: A Vendor-Neutral Framework, a, Bounded Multivocal Evidence Analysis, and Reproducible Synthetic Evaluation for SD-WAN Environments”

    2608.01728
  43. cs.CV arXiv submitted 2026-08-03
    Rig 3D Gaussians with no rigging data via diffusion priors

    Yuxin Yao +4 · “G-Skin: Learning to Bind 3D Gaussians with Generative Visual Priors”

    2608.01726
  44. cs.NI arXiv submitted 2026-08-03
    One telemetry model cuts HPC slowdown by up to 77%

    Zifan Zhang +3 · “CENTILE: A Telemetry Foundation Model Evaluated by the Decisions It Drives”

    2608.01725
  45. cs.CL arXiv submitted 2026-08-03
    Semester-long team recordings lift next-speaker prediction 13.8 points

    Heechan Lee +5 · “TIDES: A Longitudinal Bilingual Dataset for Modeling Multi-Party Social Dynamics”

    2608.01724
  46. cs.CV arXiv submitted 2026-08-03
    One-step latent flow beats dark RAW motion-blur rivals by 7.42 dB

    Zepu Wang +3 · “When Extreme Darkness Meets Motion Blur: MeanFlow for Unified RAW Restoration”

    2608.01720
  47. cs.CR arXiv submitted 2026-08-03
    Scoped disclosures cut agent leakage to 0.035

    Jinghan Xu +4 · “MNC: Scope-Bound Semantic Declassification for Private LLM-Agent Communication”

    2608.01719
  48. cs.AI arXiv submitted 2026-08-03
    Lookahead check stops LLM cache-poisoning attacks

    Jiacheng Liang +3 · “LaCache: Robust Semantic Caching for LLM Serving”

    2608.01718
  49. cs.LG arXiv submitted 2026-08-03
    External rollouts train diffusion language models to 98.4% on Sudoku

    Wonseok Lee +3 · “Beyond On-Policy Exploration: Integrating External Policy Rollouts for Reinforcement Learning in Diffusion Language Models”

    2608.01717
  50. cs.SE arXiv submitted 2026-08-03
    Agent-built tests expose 589 bugs official AtCoder suites miss

    Shuyang Xie +4 · “Coding Agents as Test-Suite Auditors: Finding What Official Suites Miss While Approaching What They Catch”

    2608.01715