ICML releases the list! Tsinghua Alibaba won the highest award, and Google DeepMind became a classic again

Zhidongxi Author | Cheng Qian Editor | Li Shuiqing Zhidongxi reported on July 6 that today, ICML, one of the three top conferences in the field of machine learning in the world, announced the list of winners. Two papers related to diffusion models from Tsinghua University and MIT won the Outstanding Paper Award. This award represents the highest level among the award-winning papers this year. ▲ICML officially announced the list of winners, one of which is a paper by Huang Gao’s team from Tsinghua University. In addition to researchers from Tsinghua University, researchers from Alibaba Group also appear in its author list. His paper proposed the minimalist algorithm JustGRPO, which has an accuracy of 89.1% on the GSM8K mathematical reasoning data set, while fully retaining the advantages of parallel decoding of the diffusion model. The paper that won the Outstanding Position Paper Award analyzed various current alignment technologies and found that malicious actors can easily use them to censor information and manipulate public opinion. The authors are Sarah Ball, a doctoral student at the University of Munich, and Phil Hackemann, an independent AI governance researcher. The paper that won the Test of Time Award came from a paper published by Google DeepMind in 2016, which introduced a reinforcement learning framework that uses multiple CPU cores to accelerate training on a single machine. It has been cited more than 15,000 times on arXiv. David Silver, chief researcher of Google DeepMind AlphaGo and AlphaZero, and Koray Kavukcuoglu, Google DeepMind CTO and Google chief AI architect, are among the authors. This ICML awarded a total of 2 Outstanding Paper Awards, 1 Outstanding Position Paper Award, and 1 Time Test Award. ICML, NeurIPS and ICLR are also known as the three top machine learning conferences. Founded in 1980, it is the world's first