Yuhang Song

Papers

Total Citations

H-Index

About

Yuhang Song is a computational neuroscience and machine learning researcher whose work sits at the fascinating intersection of biological intelligence and artificial neural networks. Song's primary research focus is **predictive coding**, a neuroscientifically-grounded framework for understanding and improving how artificial neural networks learn. His most notable contributions challenge the dominance of backpropagation — the algorithm that has powered the deep learning revolution — by proposing biologically plausible alternatives that better reflect how learning actually occurs in the brain. Song's landmark paper, "Predictive Coding: Towards a Future of Deep Learning beyond Backpropagation?" has garnered over 30 citations across its iterations, signaling growing interest in his vision for next-generation learning algorithms. His work addresses a critical limitation of standard deep learning: backpropagation's reliance on sequential, non-local computations that are difficult to parallelize and biologically unrealistic. By advancing predictive coding as a viable alternative, Song is helping to lay the theoretical groundwork for more scalable, efficient, and brain-like AI systems. His research appeals to both the neuroscience community seeking computational models of cognition and the machine learning community seeking more efficient training paradigms.