AI Helps Scientists Reduce the Genetic Code from 20 Amino Acids to 19
Researchers leveraged AI tools to redesign part of the ribosome's structure, successfully reducing protein synthesis dep…
📂 Research
Latest articles in Research
DeepSeek Gave AI a Cyber Finger — And Now It Can Truly See
While OpenAI, Google, and Anthropic compete on visual resolution, DeepSeek has taken a different path. Through visual gr…
A Single MRI Scan to Preview a Decade of Disease: AI Breakthrough in Predicting Alzheimer's Progression
Researchers have developed a deep learning-based brain age prediction model that can project the degenerative course of …
AI Agents Automatically Translate GPU Kernels: A New Breakthrough in cuTile Cross-Language Migration
NVIDIA's cuTile programming model now has a cross-language automated translation solution. Researchers have leveraged AI…
Microsoft Research: What Happens When AI Agents Interact at Scale
Microsoft Research publishes new findings revealing that individually safe AI agents don't guarantee the overall safety …
Airbnb Releases 2025 Annual AI Research Roundup
Airbnb reviews its 2025 academic research progress in AI, data science, and machine learning, significantly expanding it…
AI Reveals Major Discovery: DNA Is Not Locked Inside Cells
With the help of artificial intelligence, scientists have discovered that DNA is not entirely confined within cells. Lar…
QDTraj: Diverse Trajectory Primitives Empower Robots to Manipulate Articulated Objects
A latest arXiv paper proposes QDTraj, a method that automatically generates diverse low-level trajectory primitives, ena…
ATRS: A New Adaptive Trajectory Optimization Framework Driven by Neural Policy
Researchers propose ATRS, a method that achieves adaptive trajectory re-splitting through a shared neural policy, solvin…
GazeVLA: Driving Robotic Manipulation with Human Gaze Intent
A research team has proposed the GazeVLA framework, which extracts manipulation intent from human gaze data to effective…
RedVLA: The First Physical Safety Red-Teaming Framework for VLA Models
Researchers propose the RedVLA framework, the first systematic physical safety red-teaming approach for Vision-Language-…
CodeGraphVLP: Breaking Through Long-Horizon Robot Manipulation Bottlenecks with Code Planning and Semantic Graphs
A research team proposes the CodeGraphVLP framework, combining a Code-as-Planner module with semantic graph state repres…