Kai Chen

STAIR enhances video question answering by breaking down queries into manageable tasks.

A study of public sentiment on Reddit following the October 2023 Hamas attack.

Research on superconducting diode effect promises to advance electronic device efficiency.

This article examines how language models can adopt ideological biases from training data.

A new method enhances safety features in multimodal AI systems without extensive training.

Research aims to improve machine interpretation of 3D environments for safety.

A new method for creating textures using simple text instructions.

A new framework transforms image interpretation through open-vocabulary scene graphs.

Exploring topological phases, surface states, and their implications in modern physics.

Combining LiDAR and camera data to improve self-driving technology efficiency.

A new method for improving model structures more effectively and efficiently.

Teams worldwide enhance self-driving car technology in challenging conditions.

MathBench assesses LLMs' math capabilities across various educational stages.

A new toolbox improves LiDAR segmentation for safer self-driving cars.

YOLOv10 improves speed and accuracy in object detection for diverse applications.

Introducing RoboBEV to test BEV algorithms under real-world conditions.

PatchScaler improves image resolution efficiently while maintaining quality.

A new model that enhances code generation using multi-source data.

A new dataset analyzes misleading information in LLM responses.

NutNet enhances object detection systems by effectively identifying adversarial patches.

Language agents are becoming more adaptable, improving their communication and problem-solving skills.

InternLM-Law enhances responses to diverse Chinese legal questions with advanced training.

A new framework for creating synchronized sound effects in videos.

Introducing MotionBooth, a new way to create customized animated videos.

ReGround3D improves understanding of human instructions in 3D environments.

Emilia provides a diverse dataset for improving speech generation models.

SuperFlow enhances 3D perception models using LiDAR and camera data for autonomous driving.

AI technology improves live video generation for smoother, consistent output.

A framework to assess LLMs' abilities in data-related tasks with code interpreters.

MindSearch improves online information seeking with a structured approach.

Improved depth estimation from endoscopic images enhances surgical precision.

A new method to speed up BERT-like models for online applications.

A new framework improves instruction data quality for language models.

Recent studies uncover unique wave properties in non-Hermitian systems, revealing practical applications.

A new method to safeguard individual rights from image misuse in animations.

EMOVA enhances human-computer interaction through emotional expression.

FedCoLLM connects large and small language models while ensuring privacy and efficiency.

Innovative methods improve video quality for autonomous vehicle training.

New methods improve flaw detection in industrial products using advanced models.

Assessing machine understanding of African languages with the Uhura Benchmark.

LLMs offer insights into social media during disasters, but challenges remain.