Please submit via: https://www.zmeeting.org/submission/icdlt2026 (Select Special Session 1) 投稿链接

Autonomous Machine Intelligence represents a fundamental shift in artificial intelligence, moving beyond narrow task-specific systems toward machines capable of understanding, learning, and acting in complex real-world environments with minimal human supervision. This special session focuses on the theoretical foundations and practical applications of autonomous intelligent systems, with particular emphasis on three tightly interconnected paradigms: world models, energy-based models (EBMs), and joint embedding predictive architectures (JEPA).

Recent advances in autonomous AI architecture — particularly in perception-planning-action cycles, multimodal learning, and hierarchical predictive models — have opened new possibilities for building systems that can truly understand and interact with the physical world. EBMs provide a principled probabilistic framework for capturing dependencies through a learnable energy function, supporting contrastive and self-supervised learning. JEPA (I-JEPA, V-JEPA, V-JEPA 2) learns by predicting abstract representations in latent space, unifying EBM objectives with practical self-supervised learning. World models enable agents to simulate hypothetical scenarios and plan optimal actions via compressed latent representations of environment dynamics.

This session aims to provide a forum for researchers and practitioners to present and discuss the latest advances in autonomous machine intelligence, covering theoretical foundations, novel architectures, learning paradigms, and real-world applications. Relevant topics include, but are not limited to:

• Self-supervised learning for autonomous systems
• Energy-based models and latent variable models
• Joint Embedding Predictive Architectures (JEPA): I-JEPA, V-JEPA, V-JEPA 2
• World models and predictive learning for planning and control
• Perception-planning-action architectures
• Hierarchical and multi-scale representations
• Multimodal learning and fusion
• Intrinsic motivation and cost functions
• Model-predictive control for autonomous systems
• Video understanding and prediction
• Autonomous robotics and embodied AI
• Industrial automation, smart manufacturing, healthcare, and aquaculture applications
• Benchmarks and evaluation metrics for autonomous intelligent systems