About
Research & Development
Self-Evolving AI Enhances GUI Understanding for Precise Interactions
TLDR: LASER is a new self-evolving framework that significantly improves how Vision Language Models (VLMs) understand and interact with Graphical User Interfaces (GUIs). It...
Research & Development
Unveiling Python’s Security Gaps: A New Benchmark for Vulnerability Detection
TLDR: A new research paper introduces PyVul, the first high-quality, large-scale benchmark for Python package vulnerabilities. Using an LLM-assisted data cleansing method, PyVul achieves...
Research & Development
Predicting 3D Geometry Using Advanced Image Editing AI
TLDR: FE2E is a new AI model that excels at predicting 3D geometry (like depth and surface normals) from a single image. It achieves...
Research & Development
Tackling Evolving Audio Deepfakes with the AUDETER Dataset
TLDR: AUDETER is a new, large-scale dataset (over 4,500 hours, 3 million clips) designed to improve deepfake audio detection in real-world, "open-world" scenarios. It...
Research & Development
PARCO: Enhancing Speech Recognition for Names and Homophones with Phoneme-Aware Context
TLDR: PARCO is a new ASR framework that improves the recognition of named entities, especially homophones, by using phoneme-aware encoding, contrastive entity disambiguation, entity-level...
Research & Development
Next-Gen Parking Prediction: Leveraging AI and Multi-Source City Data
TLDR: A new research paper introduces the Self-supervised learning enhanced Spatio-Temporal Inverted Transformer (SST-iTransformer), a novel AI model for highly accurate parking availability prediction...