KAN一作刘子鸣回国任教，清华官网盖章认证了

Core Insights - Liu Ziming, a prominent researcher in neural network architecture, will join Tsinghua University's School of Artificial Intelligence as an assistant professor in September 2024 [1] - The KAN (Kolmogorov-Arnold Networks) architecture, co-developed by Liu and Max Tegmark, has gained significant attention for its superior accuracy and interpretability compared to traditional multilayer perceptrons (MLPs) [1][5] - KAN's initial paper was published in April 2024 and has already garnered over 3,000 citations, indicating its impact in the academic community [8][10] Group 1: KAN Architecture - KAN serves as a powerful alternative to MLPs, providing new opportunities for improving deep learning models that heavily rely on MLPs [1] - The design of KAN allows for the observation of the influence paths of variables within the network, offering interpretability and interactivity that MLPs cannot provide [6][8] - KAN eliminates the need for linear weights by replacing them with single-variable functions, transforming the learning of complex high-dimensional functions into a series of simpler tasks [6] Group 2: Academic Background and Career - Liu Ziming, originally from Wuhan, has a strong academic background, having been a top student in physics competitions and later pursuing a PhD at MIT under Max Tegmark [5] - His research interests lie at the intersection of physics and machine learning, focusing on enhancing the interpretability of neural networks [5][13] - Liu's teaching position at Tsinghua was confirmed earlier, and he has already completed the recruitment of his first batch of PhD students [4] Group 3: Research Philosophy - Liu Ziming emphasizes a research style driven by curiosity and long-term impact, aiming for scientific inspiration and practical influence [15] - His approach combines theoretical rigor with experimental insights, seeking to bridge the gap between pure theory and practical applications [15] - Liu maintains a blog titled "Physics of AI," where he explores AI through the lens of physics, aiming to uncover insights that could significantly impact the field [16][18]