429
1
0
2025-03-18 19:43:06
2025-03-24
Test-Time Training (TTT) has emerged as a paradigm to enhance model adaptability and robustness by allowing continuous learning directly from test data. Unlike traditional static training, TTT enables models to dynamically update their parameters or representations during inference, addressing challenges such as concept drift, domain shifts, and data scarcity. This approach leverages self-supervised learning, meta-learning, or online optimization to refine predictions on-the-fly, making it highly relevant for real-world applications like autonomous systems, healthcare monitoring, and dynamic environments.
Recent advances in TTT have shown promising results in improving generalization, reducing catastrophic forgetting, and enabling lifelong learning. However, key challenges remain, including theoretical guarantees for stability, efficient optimization algorithms for real-time adaptation, and scalable implementations across heterogeneous platforms. This session aims to foster discussions on cutting-edge TTT methodologies, their theoretical foundations, and interdisciplinary applications.
Theory of Test-Time Training in Dynamic Environments
Optimization Algorithms for Test-Time Training
Self-Supervised Learning for Test-Time Training
Transfer Learning for Test-Time Training
Meta-Learning Frameworks for Test-Time Training
Benchmarking and Evaluation Metrics for Test-Time Training
Lifelong Learning Systems for Test-Time Training
Healthcare Monitoring
Computer Vision
Drug Design
Financial Risk Detection
Paper submission: April 9, 2025
Paper Decision Notification: April 28, 2025
Conference: July 26-29, 2025, Ningbo, China
Papers submitted to this Special Session are reviewed according to the same rules as the submissions to the regular sessions of ICIC 2025. Authors who submit papers to this session are invited to mention it in the form during the submission. Submissions to regular and special sessions follow identical format, instructions, deadlines and procedures of the other papers.
Please, for further information and news refer to the ICIC website: http://www.ic-icc.cn/2025/index.php
Xueliang Li (Shenzhen University), National Engineering Laboratory for Big Data System Computing Technology, Email: lixueliang01@gmail.com
Lingjie Li (Shenzhen Technology University), College of Big Data and Internet, Email: lilingjie@sztu.edu.cn
Junkai Ji, (Shenzhen University), National Engineering Laboratory for Big Data System Computing Technology, Email: jijunkai@szu.edu.cn
