Project Name
Advanced Technologies for Designing Trustable AI Services
Project Goal
- Develop advanced trustworthy AI technologies for smart services, such as multimodal conversational AI service
- Trustable technologies in every stage of AI : data collection, unbalance distribution, small data, annotation, model training, model deployment, attack/defense, incremental consistency.
Project Description
1. Reliable Conversational AI Services
＊Anti-spoofing
＊User verification
＊Anti-attack robustness
2. Trustworthy multimedia processing
＊Multi-modal, multi-task, multi-instance
＊Unified representation learning
＊Few-Shot Learning
＊Noisy label correction
3. Human-Centered Language Processing
＊Emotion perception
＊Multi-model intelligence
＊Discourse-aware mining
4. Large-scale Scene Text Recognition
＊Continuous learning
＊Auto data labelling
＊Essential technology for robot, automobile, and traveler
