Intelligent Perception and Computation
智能感知与计算
| Chair: |
Co-chair: |
 |
 |
| Weisheng Dong |
Mingtao Feng |
| Xidian University, China |
Xidian University, China |
| |
|
| Summary: |
|
- Intelligent perception and computation are central to advancing artificial intelligence from data-driven paradigms toward cognition-driven intelligence. Despite significant progress enabled by deep learning and foundation models, current approaches still face critical challenges in open-world environments, including limited cross-domain generalization, insufficient robustness to uncertainty and degraded observations, and the difficulty of balancing multimodal integration with computational efficiency. These challenges are particularly evident in real-world applications such as autonomous driving, robotics, and complex system perception.
This forum on Intelligent Perception and Computation aims to provide a high-level platform for interdisciplinary exchange across computer vision, machine learning, and robotics. It will focus on emerging directions including multimodal perception and cross-modal learning, geometric and cross-view understanding, uncertainty-aware reasoning, efficient and scalable computation, and embodied perception systems. By bringing together academic and industry experts, the forum seeks to foster discussions on fundamental challenges and promote the integration of diverse research paradigms. Ultimately, the goal is to advance robust, generalizable, and deployable intelligent systems for complex real-world environments.
|
| |
|
- 智能感知与计算是推动人工智能从“数据驱动”走向“认知驱动”的关键基础。尽管近年来深度学习与大规模模型显著提升了感知能力，但在复杂开放环境中，现有方法仍面临跨场景泛化能力不足、对不确定性与观测退化的鲁棒性有限，以及多模态信息协同建模与高效计算之间难以兼顾等核心挑战。这些问题在自动驾驶、智能机器人与复杂系统感知等实际应用中尤为突出，亟需从方法体系与系统范式层面进行深入探讨。
论坛将聚焦若干前沿方向，包括多模态感知与跨模态学习、复杂场景下的几何与跨视角理解、不确定性建模与可信推理、高效与可扩展计算方法，以及具身智能与交互式感知系统等。通过邀请相关领域的代表性学者与产业专家，围绕关键问题展开深入讨论，促进不同研究范式之间的融合与创新。本论坛致力于凝聚学术界与工业界共识，推动构建兼具鲁棒性、泛化性与可部署性的下一代智能感知与计算方法体系，为人工智能技术在复杂真实环境中的落地应用提供理论支撑与技术路径。
|