过剩熵表明神经元的生理学为更高层次的计算做好了准备。
Excess Entropies Suggest the Physiology of Neurons to Be Primed for Higher-Level Computation.
机构信息
Institute of Neuroinformatics, University and ETH Zürich, Irchel Campus, Winterthurerstrasse 190, 8057 Zürich, Switzerland.
Department of Mathematics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139-4307, USA.
出版信息
Phys Rev Lett. 2021 Oct 1;127(14):148101. doi: 10.1103/PhysRevLett.127.148101.
Biological neuronal networks excel over artificial ones in many ways, but the origin of this is still unknown. Our symbolic dynamics-based tool of excess entropies suggests that neuronal cultures naturally implement data structures of a higher level than what we expect from artificial neural networks, or from close-to-biology neural networks. This points to a new pathway for improving artificial neural networks towards a level demonstrated by biology.
生物神经元网络在许多方面优于人工神经网络，但这一现象的起源尚不清楚。我们基于过剩熵的符号动力学工具表明，神经元培养物自然地实现了比我们从人工神经网络或接近生物学的神经网络中所预期的更高层次的数据结构。这为改善人工神经网络指明了一条新途径，使其达到生物学所展示的水平。
Zotero 插件