About
Research & Development
Accelerating Robot Actions: NinA’s Normalizing Flow Architecture for VLA Models
TLDR: NinA (Normalizing Flows in Action) is a new method for Vision-Language-Action (VLA) models that replaces traditional diffusion-based action decoders with Normalizing Flows. This...
Research & Development
Precision Retinal Image Registration: The GPO Framework for Enhanced Alignment
TLDR: Gaussian Primitive Optimization (GPO) is a novel iterative framework for deformable retinal image registration. It addresses challenges like large homogeneous regions and sparse...
Research & Development
TextOnly: A Unified Text Box for Smarter Smartphone Interactions
TLDR: TextOnly is a new smartphone interface that allows users to access various text-related functions across different applications by simply typing into one unified...
Research & Development
Optimizing Data Updates for Time-Sensitive Federated Learning
TLDR: DUFL is a new framework for federated learning that addresses data staleness in time-sensitive tasks. It uses an incentive mechanism with three controls...
Research & Development
Enhancing LLM Reasoning with Rubric-Guided Reinforcement Learning
TLDR: Researchers introduce RuscaRL, a new reinforcement learning framework for large language models (LLMs) that tackles the "exploration bottleneck." It uses checklist-style rubrics both...
Research & Development
Beyond the Text: Probing Implicit Knowledge in Black-box Language Models
TLDR: KnowProb is a novel method that uses FrameNet to construct a knowledge graph and then probes black-box language models (PLMs) to assess their...