Are markets a type of neural network and what risks flow from this?
Автор: The University of Melbourne
Загружено: 2022-09-06
Просмотров: 2853
Описание: We argue that there is a deep analogy between market supply chains and neural networks. Not only do markets exhibit network structures, but they adjust resource allocations using a variant of gradient ascent by backpropagation, where the loss function is replaced by a profit function. On short timescales, this process adjusts prices and quantities of goods manufactured; on longer timescales, it also allocates resources for R&D and the development of new products. This talk explores both the mathematical structure of this claim, and its implications. Understood this way, we see both an algorithmic explanation for efficiency and equilibrium claims in economics, and the potential for new computational insights about the dynamics and failure modes of market economies (and of attempts to regulate them) In the second part of the talk we will explore security risks based on this hypothesis.
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
