✨DiagramBank: A Large-scale Dataset of Diagram Design Exemplars with Paper Metadata for Retrieval-Augmented Generation

📝 Summary:
A large-scale dataset of schematic diagrams called DiagramBank is introduced for multimodal retrieval and exemplar-driven scientific figure generation, addressing the gap in automated publication-grad...

🔹 Publication Date: Published on Feb 28

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.20857
• PDF: https://arxiv.org/pdf/2604.20857
• Github: https://github.com/csml-rpi/DiagramBank

✨ Datasets citing this paper:
• https://huggingface.co/datasets/zhangt20/DiagramBank

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Emergent Strategic Reasoning Risks in AI: A Taxonomy-Driven Evaluation Framework

📝 Summary:
Large language models exhibit emergent strategic reasoning risks including deception and reward hacking, which are systematically evaluated through a taxonomy-driven agentic framework called ESRRSim t...

🔹 Publication Date: Published on Apr 23

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.22119
• PDF: https://arxiv.org/pdf/2604.22119

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Rewarding the Scientific Process: Process-Level Reward Modeling for Agentic Data Analysis

📝 Summary:
DataPRM, a new environment-aware process reward model, enhances LLM reasoning in dynamic data analysis. It actively detects silent errors and distinguishes error types, achieving superior benchmark performance.

🔹 Publication Date: Published on Apr 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.24198
• PDF: https://arxiv.org/pdf/2604.24198

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#LLM #RewardModeling #DataAnalysis #AIagents #MachineLearning

✨SketchVLM: Vision language models can annotate images to explain thoughts and guide users

📝 Summary:
SketchVLM is a training-free framework that enables vision-language models to generate editable SVG overlays for visual explanations, improving reasoning accuracy and annotation quality across multipl...

🔹 Publication Date: Published on Apr 23

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.22875
• PDF: https://arxiv.org/pdf/2604.22875
• Project Page: https://sketchvlm.github.io/

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#SketchVLM #VisionLanguageModels #ComputerVision #AI #ImageAnnotation

✨World-R1: Reinforcing 3D Constraints for Text-to-Video Generation

📝 Summary:
World-R1 framework improves video generation by incorporating 3D constraints through reinforcement learning and specialized text datasets while maintaining visual quality and scalability. AI-generated...

🔹 Publication Date: Published on Apr 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.24764
• PDF: https://arxiv.org/pdf/2604.24764
• Project Page: https://aka.ms/world-r1
• Github: https://github.com/microsoft/World-R1

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents

📝 Summary:
A benchmark for evaluating language-model agents in multi-day collaborative workflows with evolving environmental states across multiple service domains. AI-generated summary Language-model agents are...

🔹 Publication Date: Published on Apr 26

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.23781
• PDF: https://arxiv.org/pdf/2604.23781
• Github: https://github.com/evolvent-ai/ClawMark

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Tuna-2: Pixel Embeddings Beat Vision Encoders for Multimodal Understanding and Generation

📝 Summary:
Tuna-2 is a unified multimodal model that performs visual understanding and generation directly from pixel embeddings without pretrained vision encoders, achieving state-of-the-art performance in mult...

🔹 Publication Date: Published on Apr 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.24763
• PDF: https://arxiv.org/pdf/2604.24763
• Project Page: https://tuna-ai.org/tuna-2/

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Stabilizing Efficient Reasoning with Step-Level Advantage Selection

📝 Summary:
Short-context post-training induces reasoning compression but causes instability; Step-level Advantage Selection addresses this by selectively adjusting reasoning steps based on confidence and verific...

🔹 Publication Date: Published on Apr 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.24003
• PDF: https://arxiv.org/pdf/2604.24003
• Github: https://github.com/HanNight/SAS

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Zero-to-CAD: Agentic Synthesis of Interpretable CAD Programs at Million-Scale Without Real Data

📝 Summary:
A scalable framework synthesizes executable CAD construction sequences by framing the process as an agentic search problem using large language models within a feedback-driven CAD environment. AI-gene...

🔹 Publication Date: Published on Apr 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.24479
• PDF: https://arxiv.org/pdf/2604.24479

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨ProEval: Proactive Failure Discovery and Efficient Performance Estimation for Generative AI Evaluation

📝 Summary:
ProEval uses transfer learning with pre-trained Gaussian Processes and Bayesian quadrature to efficiently evaluate generative AI models by identifying failure cases with significantly fewer samples th...

🔹 Publication Date: Published on Apr 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.23099
• PDF: https://arxiv.org/pdf/2604.23099
• Github: https://github.com/google-deepmind/proeval

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Stochastic KV Routing: Enabling Adaptive Depth-Wise Cache Sharing

📝 Summary:
Transformer language models can reduce KV cache memory requirements through random cross-layer attention during training, enabling efficient depth-wise cache sharing without performance loss. AI-gener...

🔹 Publication Date: Published on Apr 3

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.22782
• PDF: https://arxiv.org/pdf/2604.22782

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning

📝 Summary:
ReVSI addresses flaws in current spatial intelligence evaluation by creating a validated benchmark with improved annotations and controlled frame sampling conditions. AI-generated summary Current eval...

🔹 Publication Date: Published on Apr 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.24300
• PDF: https://arxiv.org/pdf/2604.24300
• Project Page: https://3dlg-hcvc.github.io/revsi/
• Github: https://3dlg-hcvc.github.io/revsi/

✨ Datasets citing this paper:
• https://huggingface.co/datasets/3dlg-hcvc/ReVSI

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨OmniShotCut: Holistic Relational Shot Boundary Detection with Shot-Query Transformer

📝 Summary:
OmniShotCut formulates shot boundary detection as structured relational prediction using a shot query-based dense video Transformer, addressing limitations of existing methods through synthetic transi...

🔹 Publication Date: Published on Apr 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.24762
• PDF: https://arxiv.org/pdf/2604.24762
• Project Page: https://uva-computer-vision-lab.github.io/OmniShotCut_website/
• Github: https://github.com/UVA-Computer-Vision-Lab/OmniShotCut

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Vision-Language-Action Safety: Threats, Challenges, Evaluations, and Mechanisms

📝 Summary:
Vision-Language-Action models present unique safety challenges due to their embodied nature, requiring unified approaches across multiple domains to address threats from data poisoning to adversarial ...

🔹 Publication Date: Published on Apr 26

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.23775
• PDF: https://arxiv.org/pdf/2604.23775
• Github: https://github.com/LiQiiiii/Awesome-VLA-Safety

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Efficient Agent Evaluation via Diversity-Guided User Simulation

📝 Summary:
DIVERT is a coverage-guided user simulation framework that efficiently evaluates large language models by reusing conversation prefixes and exploring diverse interaction paths through branching trajec...

🔹 Publication Date: Published on Apr 23

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.21480
• PDF: https://arxiv.org/pdf/2604.21480

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨For-Value: Efficient Forward-Only Data Valuation for finetuning LLMs and VLMs

📝 Summary:
For-Value is an efficient forward-only data valuation framework for LLMs and VLMs. It estimates data value using final hidden representations and prediction errors, eliminating costly gradient computations. This enables scalable batch processing, matching or exceeding gradient-based methods in ef...

🔹 Publication Date: Published on Apr 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2508.10180
• PDF: https://arxiv.org/pdf/2508.10180
• Github: https://github.com/vengdeng/For-Value-Efficient-Forward-Only-Data-Valuation-for-finetuning

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment

📝 Summary:
Large language model agents exhibit cognitive bias where self-reflection and mutual auditing lead to inconsistent error attributions, which are addressed through a dialectical reasoning framework that...

🔹 Publication Date: Published on Apr 21

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.19548
• PDF: https://arxiv.org/pdf/2604.19548
• Project Page: https://unikcc.github.io/ReTAS/
• Github: https://github.com/unikcc/ReTAS

✨ Datasets citing this paper:
• https://huggingface.co/datasets/BradNLP/ReTAS

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨From Skills to Talent: Organising Heterogeneous Agents as a Real-World Company

📝 Summary:
OneManCompany OMC addresses static multi-agent systems by providing a framework for dynamic team assembly and governance. It uses portable agent identities and a hierarchical decision loop for self-organizing AI teams. OMC achieves 84.67% success on PRDBench, improving state-of-the-art by 15.48%.

🔹 Publication Date: Published on Apr 24

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.22446
• PDF: https://arxiv.org/pdf/2604.22446
• Project Page: https://1mancompany.github.io/OneManCompany/
• Github: https://github.com/1mancompany/OneManCompany

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #MultiAgentSystems #SelfOrganizingAI #AIteams #AutonomousAgents

✨Discovering Agentic Safety Specifications from 1-Bit Danger Signals

📝 Summary:
EPO-Safe allows LLM agents to discover hidden safety objectives using only binary danger warnings and reflection. This framework generates human-readable safety specifications autonomously, demonstrating robustness even with noisy feedback. It highlights that a dedicated safety channel is crucial...

🔹 Publication Date: Published on Apr 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.23210
• PDF: https://arxiv.org/pdf/2604.23210

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨ATTN-FIQA: Interpretable Attention-based Face Image Quality Assessment with Vision Transformers

📝 Summary:
ATTN-FIQA uses pre-softmax attention scores from Vision Transformers to assess face image quality without additional training or architectural changes. AI-generated summary Face Image Quality Assessme...

🔹 Publication Date: Published on Apr 21

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.22841
• PDF: https://arxiv.org/pdf/2604.22841
• Github: https://github.com/gurayozgur/ATTN-FIQA

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research

✨EX-FIQA: Leveraging Intermediate Early eXit Representations from Vision Transformers for Face Image Quality Assessment

📝 Summary:
ViT-based face quality assessment method utilizes intermediate representations through early exit mechanisms and score fusion strategies, demonstrating that different transformer block depths capture ...

🔹 Publication Date: Published on Apr 21

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2604.22842
• PDF: https://arxiv.org/pdf/2604.22842
• Github: https://github.com/gurayozgur/EX-FIQA

==================================

For more data science resources:
✓ https://t.iss.one/DataScienceT

#AI #DataScience #MachineLearning #HuggingFace #Research