About
Research & Development
Unlocking Time Series Secrets: Introducing DPANet for Advanced Forecasting
TLDR: DPANet is a new deep learning model for long-term time series forecasting that uses a dual-pyramid architecture to simultaneously analyze temporal multi-scale dynamics...
Research & Development
Optimizing Smart City Ambulance Routes with AI for Better Battery Life and Service Quality
TLDR: This paper introduces an extension to a simulator called SimulatorOrchestrator, incorporating AI algorithms for dynamic multi-agent vehicular planning in 6G smart cities. The...
Research & Development
Specialized AI for Patents: A Deep Dive into ModernBERT Pretraining
TLDR: Researchers have developed ModernBERT-PT, a new language model specifically pretrained on over 60 million patent documents. This model uses architectural improvements like FlashAttention...
Research & Development
Enhancing Speech LLMs: A Dual-Channel Approach to Overcome Forgetting and Modality Gaps
TLDR: This research introduces a cross-modal knowledge distillation framework to address catastrophic forgetting and modality inequivalence in speech large language models (LLMs). By using...
Research & Development
M4Diffuser: Enhancing Robot Dexterity and Navigation with Multi-View Perception and Smart Control
TLDR: M4Diffuser is a new hybrid robotics framework that combines a Multi-View Diffusion Policy with a Reduced and Manipulability-aware QP (ReM-QP) controller to enable...
Research & Development
Navigating Dynamic Worlds: How DEER Enhances Reinforcement Learning’s Adaptability
TLDR: The research paper introduces Discrepancy of Environment Prioritized Experience Replay (DEER), a novel method designed to improve reinforcement learning (RL) in non-stationary environments...