牛津大学Rafal Bogacz、Thomas Lukasiewicz、Yuhang Song和河北科技大学Zhenghua Xu研究组合作取得一项新突破。他们的最新研究表明推断可塑性之前的神经活动为反向传播学习奠定基础。相关论文于2024年1月3日发表于国际学术期刊《自然-神经科学》杂志上。

研究人员提出了一种根本不同的信用分配原则，称为"前瞻性配置"。在前瞻性配置中，网络首先推断出学习应该产生的神经活动模式，然后修改突触权重以巩固神经活动的变化。研究人员证明了这种与反向传播不同的机制：（1）其是皮层电路模型家族中成熟学习的基础；（2）使生物体在面临复杂情况时的学习更有效率和效果；（3）再现了在人类和大鼠各种学习实验中观察到的神经活动和行为模式。

研究人员表示，无论对人类还是机器而言，学习的本质都是找出信息处理过程中哪些组件对输出错误负有责任，这一挑战被称为"学分分配"。长期以来，人们一直认为学分分配最好通过反向传播来解决，这也是现代机器学习的基础。

附：英文原文

Title: Inferring neural activity before plasticity as a foundation for learning beyond backpropagation

Author: Song, Yuhang, Millidge, Beren, Salvatori, Tommaso, Lukasiewicz, Thomas, Xu, Zhenghua, Bogacz, Rafal

Issue&Volume: 2024-01-03

Abstract: For both humans and machines, the essence of learning is to pinpoint which components in its information processing pipeline are responsible for an error in its output, a challenge that is known as ‘credit assignment’. It has long been assumed that credit assignment is best solved by backpropagation, which is also the foundation of modern machine learning. Here, we set out a fundamentally different principle on credit assignment called ‘prospective configuration’. In prospective configuration, the network first infers the pattern of neural activity that should result from learning, and then the synaptic weights are modified to consolidate the change in neural activity. We demonstrate that this distinct mechanism, in contrast to backpropagation, (1) underlies learning in a well-established family of models of cortical circuits, (2) enables learning that is more efficient and effective in many contexts faced by biological organisms and (3) reproduces surprising patterns of neural activity and behavior observed in diverse human and rat learning experiments.

DOI: 10.1038/s41593-023-01514-1

Source: https://www.nature.com/articles/s41593-023-01514-1