在计算机人工智能领域顶级会议ICML（International Conference on Machine Learning）上，ICML 2024的奖项信息在奥地利维也纳举行的大会上公布。作为AI领域的重要盛事，今年的ICML已经是第41届，吸引了全球范围内的科研人员和企业关注。本届大会共收到9473篇有效论文投稿，最终有2610篇论文被接收，其中包含144篇oral报告和191篇spotlight报告，录用率为27.5%。这些接收论文的主题关键词涵盖了当前AI领域最为热门的研究方向，如大语言模型、强化学习、深度学习、图神经网络、机器学习、联邦学习、扩散模型、Transformer、LLM、表示学习、生成模型等。
在ICML 2024的奖项中，特别值得关注的是贾扬清共一完成的论文DeCAF，因其在视觉领域的贡献获得了今年的时间检验奖。这一奖项认可了论文的长期影响力和对领域发展的贡献。贾扬清在社交媒体上表示，DeCAF的工作在视觉领域引入了通用的特征表示和深度嵌入的概念，为计算机视觉研究提供了基础框架，并促进了后续一系列关键技术和研究的发展，包括通用物体检测框架R-CNN、高性能异构计算框架Caffe、伯克利与NVidia合作的加速框架CuDNN、雅虎实验室的分布式训练框架CaffeOnSpark等。
同时，ICML 2024还公布了10篇最佳论文，涵盖了从强化学习到图像生成模型等AI研究的多个前沿方向。这些获奖论文代表了当前AI研究的顶尖水平，包括由Google DeepMind开发的用于构建复杂世界模型的Genie，以及在视频生成领域有突破的VideoPoet等。
这一系列奖项和数据不仅反映了ICML作为全球顶级AI会议的影响力，也展示了当前AI研究领域的热点和前沿趋势，对于推动全球AI技术的发展具有重要意义。
英语如下：
### Jia Yangqing’s Co-authored Paper Receives Time-Tested Award, ICML 2024 Best Paper Announced
In the top-tier conference of computer artificial intelligence, the ICML (International Conference on Machine Learning) awards for ICML 2024 were announced at the conference held in Vienna, Austria. As a significant event in the AI field, this year’s ICML marks its 41st edition, garnering attention from researchers and companies worldwide. Out of 9,473 valid paper submissions received, 2,610 papers were accepted, including 144 oral presentations and 191 spotlight presentations, with an acceptance rate of 27.5%. The themes of the accepted papers cover the most cutting-edge research directions in the AI field, such as large language models, reinforcement learning, deep learning, graph neural networks, machine learning, federated learning, diffusion models, Transformers, LLM, representation learning, and generative models.
Among the ICML 2024 awards, a particular highlight is the Time-Tested Award for Jia Yangqing’s co-authored paper, DeCAF, for its contributions to the field of vision. This award acknowledges the paper’s long-term impact and contribution to the development of the field. Jia Yangqing expressed on social media that the work of DeCAF introduced the concepts of universal feature representations and deep embeddings into the field of visual, providing a foundational framework for computer vision research and stimulating the development of subsequent key technologies and research, including the general object detection framework R-CNN, the high-performance heterogeneous computing framework Caffe, the accelerated framework CuDNN jointly developed by Berkeley and NVidia, and the distributed training framework CaffeOnSpark developed by Yahoo Labs.
Simultaneously, the ICML 2024 also unveiled the list of 10 best papers, covering various cutting-edge directions in AI research, from reinforcement learning to image generation models. These award-winning papers represent the pinnacle of current AI research, including Genie, developed by Google DeepMind for constructing complex world models, and VideoPoet, which made breakthroughs in video generation.
The series of awards and data not only reflect the influence of ICML as a global top AI conference but also showcase the current hotspots and frontiers in AI research, which are of great significance in driving the development of global AI technology.
【来源】https://www.jiqizhixin.com/articles/2024-07-24
Views: 3