2023-08-31 ノースカロライナ州立大学(NCState)
◆従来のAIは同一の人工ニューロンから成るネットワークを使用しますが、この研究ではAIに、種類と接続強度の異なる人工ニューロンを選択し、学習中にネットワーク内で異なるサブネットワークを作成する能力を与えました。この方法により、AIは複雑な問題を解決する際に最大10倍の精度向上を示し、特に混沌とした問題ほど劇的な改善が見られました。
＜関連情報＞
- https://news.ncsu.edu/2023/08/an-introspective-ai-finds-diversity-improves-performance/
- https://www.nature.com/articles/s41598-023-40766-6
ニューロンの多様性により、物理学以降の機械学習が改善される Neuronal diversity can improve machine learning for physics and beyond
Anshul Choudhary,Anil Radhakrishnan,John F. Lindner,Sudeshna Sinha & William L. Ditto
Scientific Reports Published:26 August 2023
DOI:https://doi.org/10.1038/s41598-023-40766-6
Abstract
Diversity conveys advantages in nature, yet homogeneous neurons typically comprise the layers of artificial neural networks. Here we construct neural networks from neurons that learn their own activation functions, quickly diversify, and subsequently outperform their homogeneous counterparts on image classification and nonlinear regression tasks. Sub-networks instantiate the neurons, which meta-learn especially efficient sets of nonlinear responses. Examples include conventional neural networks classifying digits and forecasting a van der Pol oscillator and physics-informed Hamiltonian neural networks learning Hénon–Heiles stellar orbits and the swing of a video recorded pendulum clock. Such learned diversity provides examples of dynamical systems selecting diversity over uniformity and elucidates the role of diversity in natural and artificial systems.
