New Benchmark BTF-2: Evaluating Strategic Reasoning Capabilities of AI Forecasting Agents
A new arXiv paper introduces "Bench to the Future 2," a benchmark that systematically evaluates reasoning strategy diffe…
📂 Research
Latest articles in Research
New Framework Extracts Multi-Level Trustworthy User Profiles from Behavioral Logs
A latest arXiv paper proposes a Hierarchical Multi-Persona Induction framework that aggregates user behavior logs into i…
Fine-Tuning Can Reawaken Copyright Memory in LLMs, Making Alignment Safety a Game of 'Whack-a-Mole'
A new study reveals that large language models can have their verbatim memorization of copyrighted books reactivated thr…
AdaAct: A Novel Weakly-Supervised Action Segmentation Method with Human-Object Interaction Awareness
Researchers propose AdaAct, an HOI-aware adaptive network that effectively addresses the ambiguity problem of similar ac…
EnerGS: A New Energy-Based Framework for Gaussian Splatting
A research team has proposed EnerGS, a method that introduces energy functions into the 3D Gaussian Splatting framework.…
MetaSR: A New Super-Resolution Paradigm Driven by Content-Adaptive Metadata Orchestration
A latest arXiv paper proposes the MetaSR framework, which breaks through the limitations of fixed conditioning designs i…
New Method SQI Stops Vision-Language Models from Being Fooled by Optical Illusions
Researchers propose the Structured Qualitative Inference (SQI) framework, a training-free approach that significantly en…
DepthPilot: Taking Colonoscopy Video Generation from Controllable to Interpretable
A research team has proposed the DepthPilot framework, achieving interpretability in colonoscopy video generation for th…
Camera-RFID Fusion: A New Breakthrough in Woodland Asset Tracking
A new study proposes a camera-RFID fusion approach that combines the low-cost advantages of passive RFID tags with the c…
ViBE: Achieving Visual-to-Brain Encoding Breakthroughs with Spatio-Temporal VAE
A research team has proposed the ViBE model, which leverages a spatio-temporal variational autoencoder and distribution-…
FASH-iCNN: Decoding Fashion Brand Aesthetic DNA with Multimodal CNNs
A research team has unveiled the FASH-iCNN system, which leverages multimodal learning on 87,547 Vogue runway images to …
Privacy-Preserving Clothing Classification: ViT Powers Intelligent Thermal Comfort Estimation
A new study proposes a privacy-preserving clothing classification approach based on Vision Transformer, enabling accurat…