✨World in a Frame: Understanding Culture Mixing as a New Challenge for Vision-Language Models

📝 Summary:
LVLMs struggle to preserve cultural identities in mixed visual scenes. Researchers created CultureMix, a VQA benchmark, finding consistent failures and background reliance. Supervised fine-tuning with diverse culture mixing data significantly improves model consistency and reduces background sens...

🔹 Publication Date: Published on Nov 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.22787
• PDF: https://arxiv.org/pdf/2511.22787

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#VisionLanguageModels #CulturalAI #ComputerVision #AIML #AIResearch

✨RefineBench: Evaluating Refinement Capability of Language Models via Checklists

📝 Summary:
RefineBench evaluates language models' self-refinement and guided refinement capabilities using 1,000 problems and a checklist. It finds that LMs perform poorly at self-refinement, often failing to improve without guidance, but excel at guided refinement with targeted feedback.

🔹 Publication Date: Published on Nov 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.22173
• PDF: https://arxiv.org/pdf/2511.22173

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

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#LLM #AI #NLP #ModelEvaluation #Refinement

✨From Pixels to Feelings: Aligning MLLMs with Human Cognitive Perception of Images

📝 Summary:
MLLMs struggle with human cognitive perception of images like memorability or aesthetics. CogIP-Bench evaluates this gap, showing post-training significantly improves alignment. This enhances human-like perception and improves creative AI tasks.

🔹 Publication Date: Published on Nov 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.22805
• PDF: https://arxiv.org/pdf/2511.22805
• Project Page: https://follen-cry.github.io/MLLM-Cognition-project-page/

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#MLLM #CognitiveAI #ImagePerception #AIAlignment #AIResearch

✨Adversarial Flow Models

📝 Summary:
Adversarial flow models unify adversarial and flow-based generative models for stable training and efficient one-step generation. They learn a deterministic noise-to-data mapping, achieving record FIDs of 1.94 on ImageNet-256px with a single pass, outperforming consistency models.

🔹 Publication Date: Published on Nov 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.22475
• PDF: https://arxiv.org/pdf/2511.22475

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#GenerativeAI #DeepLearning #AdversarialModels #FlowModels #ImageSynthesis

✨Find the Leak, Fix the Split: Cluster-Based Method to Prevent Leakage in Video-Derived Datasets

📝 Summary:
This paper introduces a cluster-based frame selection strategy for video datasets. It groups similar frames to prevent information leakage and create more balanced and reliable dataset partitions for training, validation, and testing.

🔹 Publication Date: Published on Nov 17

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.13944
• PDF: https://arxiv.org/pdf/2511.13944

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#VideoDatasets #DataLeakage #MachineLearning #Clustering #DatasetSplitting

✨YOLO Meets Mixture-of-Experts: Adaptive Expert Routing for Robust Object Detection

📝 Summary:
A new Mixture-of-Experts framework uses adaptive routing among multiple YOLOv9-T experts. This improves object detection performance, achieving higher mAP and AR.

🔹 Publication Date: Published on Nov 17

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.13344
• PDF: https://arxiv.org/pdf/2511.13344

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#ObjectDetection #YOLO #MixtureOfExperts #DeepLearning #ComputerVision

✨Recognition of Abnormal Events in Surveillance Videos using Weakly Supervised Dual-Encoder Models

📝 Summary:
This paper introduces a dual-backbone framework combining convolutional and transformer representations with top-k pooling to detect abnormal events in surveillance videos. The weakly supervised model achieved 90.7% AUC on the UCF-Crime dataset.

🔹 Publication Date: Published on Nov 17

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.13276
• PDF: https://arxiv.org/pdf/2511.13276

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#ComputerVision #DeepLearning #Surveillance #AnomalyDetection #WeaklySupervisedLearning

✨CaptionQA: Is Your Caption as Useful as the Image Itself?

📝 Summary:
CaptionQA assesses if AI captions adequately substitute images for downstream tasks. This benchmark uses over 33000 visual questions across 4 domains. It shows large utility gaps as MLLMs perform up to 32% worse with captions than with images.

🔹 Publication Date: Published on Nov 26

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.21025
• PDF: https://arxiv.org/pdf/2511.21025
• Github: https://github.com/bronyayang/CaptionQA

✨ Datasets citing this paper:
• https://huggingface.co/datasets/Borise/CaptionQA

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#AICaptions #MultimodalAI #ComputerVision #AIevaluation #NLP

✨Fast3Dcache: Training-free 3D Geometry Synthesis Acceleration

📝 Summary:
Fast3Dcache accelerates 3D diffusion model inference using a training-free geometry-aware caching framework. It uses dynamic cache quotas and spatiotemporal stability criteria to reuse computations, achieving significant speed-up and FLOPs reduction with minimal geometric quality degradation.

🔹 Publication Date: Published on Nov 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.22533
• PDF: https://arxiv.org/pdf/2511.22533
• Project Page: https://fast3dcache-agi.github.io/
• Github: https://fast3dcache-agi.github.io/

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#3DGeometry #DiffusionModels #ComputerVision #DeepLearning #ComputationalEfficiency

✨OmniRefiner: Reinforcement-Guided Local Diffusion Refinement

📝 Summary:
OmniRefiner enhances reference-guided image generation by overcoming fine detail loss. It uses a two-stage framework: a fine-tuned diffusion editor for global coherence, then reinforcement learning for localized detail accuracy. This significantly improves detail preservation and consistency.

🔹 Publication Date: Published on Nov 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.19990
• PDF: https://arxiv.org/pdf/2511.19990
• Github: https://github.com/yaoliliu/OmniRefiner

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#DiffusionModels #ImageGeneration #ReinforcementLearning #GenerativeAI #ComputerVision

✨DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning

📝 Summary:
DeepSeekMath-V2 trains a self-verifying LLM for theorem proving. It uses a verifier as a reward model to incentivize rigorous, step-by-step derivations and issue resolution in proofs. This approach achieves gold-level scores in major math competitions.

🔹 Publication Date: Published on Nov 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.22570
• PDF: https://arxiv.org/pdf/2511.22570

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#DeepSeekMath #LLM #AI #MathematicalReasoning #TheoremProving

✨Layer-Aware Video Composition via Split-then-Merge

📝 Summary:
Split-then-Merge is a novel framework improving generative video composition. It learns dynamic foreground-background interactions by unsupervisedly splitting unlabeled videos into layers and then self-composing them. This approach achieves state-of-the-art performance and addresses data scarcity.

🔹 Publication Date: Published on Nov 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.20809
• PDF: https://arxiv.org/pdf/2511.20809
• Project Page: https://split-then-merge.github.io/

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#VideoComposition #GenerativeAI #ComputerVision #DeepLearning #AIResearch

✨MRI Super-Resolution with Deep Learning: A Comprehensive Survey

📝 Summary:
This survey comprehensively reviews deep learning methods for MRI super-resolution, enabling high-resolution imaging from low-resolution scans. It categorizes techniques, discusses challenges, and provides valuable resources for the community.

🔹 Publication Date: Published on Nov 20

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.16854
• PDF: https://arxiv.org/pdf/2511.16854
• Github: https://github.com/mkhateri/Awesome-MRI-Super-Resolution

🔹 Models citing this paper:
• https://huggingface.co/mkhateri/Awesome-MRI-Super-Resolution

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#DeepLearning #MRI #SuperResolution #MedicalImaging #AI

✨Nemotron-Flash: Towards Latency-Optimal Hybrid Small Language Models

📝 Summary:
This study optimizes small language models for real-device latency by identifying key architectural factors and efficient operators. It introduces Nemotron-Flash, a new family of hybrid SLMs that significantly improves accuracy, latency, and throughput compared to current models.

🔹 Publication Date: Published on Nov 24

🔹 Paper Links:
• arXiv Page: https://arxiv.org/pdf/2511.18890
• PDF: https://arxiv.org/pdf/2511.18890

🔹 Models citing this paper:
• https://huggingface.co/nvidia/Nemotron-Flash-3B-Instruct
• https://huggingface.co/nvidia/Nemotron-Flash-1B
• https://huggingface.co/nvidia/Nemotron-Flash-3B

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#SmallLanguageModels #LatencyOptimization #AI #DeepLearning #NLP

✨Xmodel-2.5: 1.3B Data-Efficient Reasoning SLM

📝 Summary:
Xmodel-2.5 is a 1.3B language model designed for efficient edge deployments. It uses maximal-update parameterization and a novel training curriculum that switches from AdamW to Muon, improving reasoning skills by 4.58% while maintaining efficiency.

🔹 Publication Date: Published on Nov 23

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.19496
• PDF: https://arxiv.org/pdf/2511.19496
• Github: https://github.com/XiaoduoAILab/Xmodel-2.5

🔹 Models citing this paper:
• https://huggingface.co/XiaoduoAILab/Xmodel-2.5

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#SLM #EdgeAI #LanguageModels #DeepLearning #ReasoningAI

✨Geometrically-Constrained Agent for Spatial Reasoning

📝 Summary:
Geometrically Constrained Agent GCA resolves the semantic to geometric gap in VLMs for spatial reasoning. It uses a formal task constraint to guide the VLM from semantic analysis to constrained tool execution, achieving SOTA performance.

🔹 Publication Date: Published on Nov 27

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.22659
• PDF: https://arxiv.org/pdf/2511.22659
• Project Page: https://gca-spatial-reasoning.github.io
• Github: https://github.com/gca-spatial-reasoning/gca

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#SpatialReasoning #VLMs #AI #Robotics #DeepLearning

✨LongVT: Incentivizing "Thinking with Long Videos" via Native Tool Calling

📝 Summary:
LongVT is an agentic framework that improves long video reasoning. It uses LMMs as tools for global-to-local video cropping and frame resampling to ground answers. This novel approach consistently outperforms existing baselines.

🔹 Publication Date: Published on Nov 25

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2511.20785
• PDF: https://arxiv.org/pdf/2511.20785
• Project Page: https://evolvinglmms-lab.github.io/LongVT/
• Github: https://github.com/EvolvingLMMs-Lab/LongVT

🔹 Models citing this paper:
• https://huggingface.co/longvideotool/LongVT-RFT
• https://huggingface.co/longvideotool/LongVT-SFT
• https://huggingface.co/longvideotool/LongVT-RL

✨ Datasets citing this paper:
• https://huggingface.co/datasets/longvideotool/LongVT-Source
• https://huggingface.co/datasets/longvideotool/LongVT-Parquet

✨ Spaces citing this paper:
• https://huggingface.co/spaces/longvideotool/LongVT-Demo

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#VideoAI #LMMs #AgenticAI #ComputerVision #AIResearch

✨GR-RL: Going Dexterous and Precise for Long-Horizon Robotic Manipulation

📝 Summary:
GR-RL improves VLA policies for dexterous long-horizon manipulation. It filters and augments demonstrations, then refines them with RL. This enables unprecedented complex tasks, notably autonomously lacing a shoe.

🔹 Publication Date: Published on Dec 1

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2512.01801
• PDF: https://arxiv.org/pdf/2512.01801

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#Robotics #ReinforcementLearning #DexterousManipulation #RoboticManipulation #AI

✨What about gravity in video generation? Post-Training Newton's Laws with Verifiable Rewards

📝 Summary:
NewtonRewards is a post-training framework that uses verifiable, physics-grounded rewards to improve physical realism and motion quality in AI-generated videos. It enforces Newtonian kinematics and mass conservation, significantly outperforming prior methods on various motion tasks. This offers a...

🔹 Publication Date: Published on Nov 29

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2512.00425
• PDF: https://arxiv.org/pdf/2512.00425
• Project Page: https://cvlab-stonybrook.github.io/NewtonRewards/
• Github: https://cvlab-stonybrook.github.io/NewtonRewards

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#AIVideoGeneration #PhysicsInAI #MachineLearning #GenerativeAI #ComputerVision

✨SpeContext: Enabling Efficient Long-context Reasoning with Speculative Context Sparsity in LLMs

📝 Summary:
SpeContext uses a distilled language model for efficient long-context LLM reasoning. This system co-design significantly reduces parameters and improves throughput by up to 24.89x in cloud and 10.06x in edge, with minimal accuracy loss.

🔹 Publication Date: Published on Nov 30

🔹 Paper Links:
• arXiv Page: https://arxiv.org/abs/2512.00722
• PDF: https://arxiv.org/pdf/2512.00722

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

For more data science resources:
✓ https://news.1rj.ru/str/DataScienceT

#LLM #AIResearch #DeepLearning #AIOptimization #ContextSparsity