Recent Research Signals

Found 488 results

Filtering papers...

Paper	Possible Impact	Features	Possible Business Area
PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing CV and Pattern Recognition	Reads documents and recognizes seals better.	Baidu Code Data	Image Recognition (78%)	Score: 4 3 days ago
C2:Cross learning module enhanced decision transformer with Constraint-aware loss for auto-bidding Machine Learning (CS)	Helps online ads bid smarter and save money.	Kuaishou Code	A/B Testing (76%)	Score: 3 4 days ago
C2NP: A Benchmark for Learning Scale-Dependent Geometric Invariances in 3D Materials Generation Materials Science	Helps design tiny new materials for better batteries.	Code 3 Countries	Nanotechnology (78%)	Score: 3 6 days ago
Ostrakon-VL: Towards Domain-Expert MLLM for Food-Service and Retail Stores Artificial Intelligence	Helps stores understand what customers buy better.	Alibaba Code	Image Recognition (79%)	Score: 3 3 days ago
OffSeeker: Online Reinforcement Learning Is Not All You Need for Deep Research Agents Artificial Intelligence	Trains smart computer helpers without expensive online learning.	Tencent Code	Artificial Intelligence (78%)	Score: 3 6 days ago
C2:Cross learning module enhanced decision transformer with Constraint-aware loss for auto-bidding Machine Learning (CS)	Teaches computers to bid smarter in online ads.	Kuaishou Code	A/B Testing (76%)	Score: 3 4 days ago
M$^{\text{2}}$XFP: A Metadata-Augmented Microscaling Data Format for Efficient Low-bit Quantization Hardware Architecture	Makes computer brains work faster, use less power.	Huawei Code	DSP (76%)	Score: 3 5 days ago
SAM Audio Judge: A Unified Multimodal Framework for Perceptual Evaluation of Audio Separation Audio and Speech Processing	Tests sound quality automatically, like a human listener.	Johns Hopkins University Meta Code	Speech Recognition (77%)	Score: 3 5 days ago
Investigating the Development of Task-Oriented Communication in Vision-Language Models Artificial Intelligence	AI learns secret codes to work together.	University of California, Berkeley Code 3 Countries	Natural Language Processing (81%)	Score: 3 4 days ago
Enhancing Underwater Light Field Images via Global Geometry-aware Diffusion Process CV and Pattern Recognition	Clears up blurry underwater pictures.	Code 2 Countries	Diving (75%)	Score: 3 4 days ago
Meta Context Engineering via Agentic Skill Evolution Artificial Intelligence	Teaches computers to learn better and faster.	Code Data	Semantic Search (81%)	Score: 3 3 days ago
Late Breaking Results: Conversion of Neural Networks into Logic Flows for Edge Computing Machine Learning (CS)	Makes smart computer programs run much faster.	Code 2 Countries	Natural Language Processing (77%)	Score: 3 3 days ago
TopKGAT: A Top-K Objective-Driven Architecture for Recommendation Information Retrieval	Helps apps pick the best things for you.	Code 2 Countries	Semantic Search (75%)	Score: 3 6 days ago
VIBEVOICE-ASR Technical Report Sound	Listens to hours of talking, understands everyone.	Microsoft Code	Speech Recognition (82%)	Score: 3 6 days ago
Benchmarking Multimodal Large Language Models for Missing Modality Completion in Product Catalogues Multimedia	Computers create missing product pictures or words.	Code 3 Countries	Visual Search (76%)	Score: 3 5 days ago
Accelerating Generative Recommendation via Simple Categorical User Sequence Compression Information Retrieval	Makes online shopping suggestions faster and more accurate.	Huawei Code 2 Countries	Semantic Search (77%)	Score: 3 5 days ago
MIRAGE: Enabling Real-Time Automotive Mediated Reality Human-Computer Interaction	Makes driving safer by showing helpful info.	Code 2 Countries	Augmented Reality (84%)	Score: 3 5 days ago
Modeling Cascaded Delay Feedback for Online Net Conversion Rate Prediction: Benchmark, Insights and Solutions Machine Learning (CS)	Helps online stores sell more by predicting returns.	Alibaba Code	Image Recognition (75%)	Score: 3 5 days ago
BLENDER: Blended Text Embeddings and Diffusion Residuals for Intra-Class Image Synthesis in Deep Metric Learning CV and Pattern Recognition	Makes computer vision	Meta Code	A/B Testing (75%)	Score: 3 4 days ago
DRAINCODE: Stealthy Energy Consumption Attacks on Retrieval-Augmented Code Generation via Context Poisoning Software Engineering	Slows down AI that writes computer code.	Huawei Code 2 Countries	Energy Storage (73%)	Score: 3 4 days ago
AgentLongBench: A Controllable Long Benchmark For Long-Contexts Agents via Environment Rollouts Computation and Language	Tests AI's ability to solve tricky puzzles.	Code Data	Natural Language Processing (78%)	Score: 3 4 days ago
Evolutionary Strategies lead to Catastrophic Forgetting in LLMs Machine Learning (CS)	AI learns new things without forgetting old ones.	University of California, Berkeley Code	Machine Learning (77%)	Score: 3 4 days ago
RTeAAL Sim: Using Tensor Algebra to Represent and Accelerate RTL Simulation (Extended Version) Hardware Architecture	Makes computer chips design much faster.	University of California, Berkeley Code	Simulation (77%)	Score: 3 6 days ago
When Life Gives You AI, Will You Turn It Into A Market for Lemons? Understanding How Information Asymmetries About AI System Capabilities Affect Mark… Human-Computer Interaction	Helps people trust and use smart computer programs.	Code Data 4 Countries	Artificial Intelligence (82%)	Score: 3 3 days ago
LANCER: LLM Reranking for Nugget Coverage Information Retrieval	Helps computers write longer, more complete reports.	Johns Hopkins University Code 3 Countries	Semantic Search (77%)	Score: 3 3 days ago
Learning Hamiltonian Flow Maps: Mean Flow Consistency for Large-Timestep Molecular Dynamics Machine Learning (CS)	Lets computer models run simulations much faster.	Microsoft Code 2 Countries	Simulation (74%)	Score: 3 3 days ago
LMK > CLS: Landmark Pooling for Dense Embeddings Computation and Language	Makes computer searches better for long texts.	IBM Code	Natural Language Processing (73%)	Score: 3 3 days ago
Towards Robust Dysarthric Speech Recognition: LLM-Agent Post-ASR Correction Beyond WER Audio and Speech Processing	Fixes computer speech errors for clearer understanding.	Code Data 2 Countries	Speech Recognition (78%)	Score: 3 3 days ago
SERA: Soft-Verified Efficient Repository Agents Computation and Language	Trains AI to understand your private code.	University of Washington Code	Semantic Web (78%)	Score: 3 4 days ago
AgentLongBench: A Controllable Long Benchmark For Long-Contexts Agents via Environment Rollouts Computation and Language	Tests AI's ability to solve puzzles with changing clues.	Code Data	Natural Language Processing (78%)	Score: 3 4 days ago
Are Video Generation Models Geographically Fair? An Attraction-Centric Evaluation of Global Visual Knowledge CV and Pattern Recognition	AI videos show world landmarks equally well.	Code Data	Image Recognition (79%)	Score: 3 6 days ago
Randomization Boosts KV Caching, Learning Balances Query Load: A Joint Perspective Machine Learning (CS)	Makes AI answer questions much faster.	Google Code	A/B Testing (77%)	Score: 3 6 days ago
SAGE: Steerable Agentic Data Generation for Deep Search with Execution Feedback Artificial Intelligence	Creates smart search tools that learn to answer hard questions.	Google Code Data	Semantic Search (81%)	Score: 3 6 days ago
From Verifiable Dot to Reward Chain: Harnessing Verifiable Reference-based Rewards for Reinforcement Learning of Open-ended Generation Computation and Language	Makes AI write better by checking its work.	Huawei Code 2 Countries	Autonomous Vehicles (75%)	Score: 3 6 days ago
Agentic Very Long Video Understanding CV and Pattern Recognition	Lets AI remember weeks of what you see.	Meta Code	Image Recognition (81%)	Score: 3 6 days ago
M2XFP: A Metadata-Augmented Microscaling Data Format for Efficient Low-bit Quantization Hardware Architecture	Makes AI smarter with less computer power.	Huawei Code	DSP (76%)	Score: 3 5 days ago
Dynamic Multi-Expert Projectors with Stabilized Routing for Multilingual Speech Recognition Computation and Language	Helps computers understand many languages spoken.	IBM Code 2 Countries	Semantic Search (76%)	Score: 3 5 days ago
Dynamic Thinking-Token Selection for Efficient Reasoning in Large Reasoning Models Artificial Intelligence	Makes AI think faster by skipping extra steps.	Code Data	Natural Language Processing (78%)	Score: 3 6 days ago
Typhoon-S: Minimal Open Post-Training for Sovereign Large Language Models Computation and Language	Makes AI understand local languages and laws.	Code Data	Natural Language Processing (77%)	Score: 3 6 days ago
Benchmarking Multimodal Large Language Models for Missing Modality Completion in Product Catalogues Multimedia	Computers create missing product pictures and words.	Code 3 Countries	Visual Search (76%)	Score: 3 5 days ago
Calibrating Beyond English: Language Diversity for Better Quantized Multilingual LLM Computation and Language	Makes AI understand many languages better.	Code Data	Natural Language Processing (77%)	Score: 2 6 days ago
Contextual Range-View Projection for 3D LiDAR Point Clouds CV and Pattern Recognition	Makes self-driving cars see better in 3D.	Code	Image Recognition (77%)	Score: 2 6 days ago
InkIdeator: Supporting Chinese-Style Visual Design Ideation via AI-Infused Exploration of Chinese Paintings Human-Computer Interaction	Helps artists create Chinese art with AI.	Code 2 Countries	Visual Search (79%)	Score: 2 6 days ago
Temp-R1: A Unified Autonomous Agent for Complex Temporal KGQA via Reverse Curriculum Reinforcement Learning Computation and Language	Lets computers answer questions about changing facts.	Code	Natural Language Processing (76%)	Score: 2 6 days ago
MindCine: Multimodal EEG-to-Video Reconstruction with Large-Scale Pretrained Models CV and Pattern Recognition	Shows what you're seeing from brain waves.	Code	Motion Capture (79%)	Score: 2 6 days ago
FP8-RL: A Practical and Stable Low-Precision Stack for LLM Reinforcement Learning Machine Learning (CS)	Makes AI faster and use less memory.	NVIDIA Code	Natural Language Processing (74%)	Score: 2 6 days ago
POPE: Learning to Reason on Hard Problems via Privileged On-Policy Exploration Machine Learning (CS)	Helps AI learn hard problems with hints.	Code	Natural Language Processing (77%)	Score: 2 6 days ago
Robust Learning of a Group DRO Neuron Machine Learning (CS)	Teaches computers to learn from messy, changing information.	Microsoft Code	A/B Testing (74%)	Score: 2 6 days ago
MindCine: Multimodal EEG-to-Video Reconstruction with Large-Scale Pretrained Models CV and Pattern Recognition	Shows what you're seeing from brain waves.	Code	Motion Capture (79%)	Score: 2 6 days ago
Low Cost, High Efficiency: LiDAR Place Recognition in Vineyards with Matryoshka Representation Learning CV and Pattern Recognition	Helps farm robots know where they are.	Code	Image Recognition (80%)	Score: 2 6 days ago