🏢 School of Data Science, Fudan University
SlowFocus: Enhancing Fine-grained Temporal Understanding in Video LLM
·4826 words·23 mins·
loading
·
loading
AI Generated
Natural Language Processing
Vision-Language Models
🏢 School of Data Science, Fudan University
SlowFocus significantly improves fine-grained temporal understanding in video LLMs by using mixed-frequency sampling and a novel multi-frequency attention mechanism.
Near-Optimal Distributed Minimax Optimization under the Second-Order Similarity
·1858 words·9 mins·
loading
·
loading
AI Generated
Machine Learning
Optimization
🏢 School of Data Science, Fudan University
SVOGS: Near-optimal distributed minimax optimization is achieved under second-order similarity, balancing communication, computation, and achieving near-optimal complexities.
DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States
·2266 words·11 mins·
loading
·
loading
AI Applications
Autonomous Vehicles
🏢 School of Data Science, Fudan University
DeMo: Decoupling motion forecasting into directional intentions and dynamic states for improved autonomous driving.