-

A Modest Proposal: Use Deep Learning for the Transformation Problem in Marxist Economics

组织者

高瑞泽 , 韩立岩 , 李振 , 龙飞 , 汤珂

演讲者

William Vickery

时间

2025年06月27日 15:00 至 16:00

地点

A3-2a-302

线上

Zoom 637 734 0280 (BIMSA)

摘要

The “AI for Science” push has produced remarkable results. Alphafold, which solved millions of protein structures in one fell swoop, is the most well known. DeePKS, DeePMD, Trajectorynet, DeePN2 are merely a few representative samples of a litany of new applications of deep learning tools to solving high dimensional problems from chemistry and biology resistant to classical algorithms. Economic data sets, such as input-output tables or tables of prices, are similarly high dimensional. Adam Smith acknowledged and Marx posited as foundational the idea that socially necessary labor time can be used as a unit of measure for the value for commodities. Both suggested that the value of a commodity should act as a constraint on its price. The transformation problem posits that prices should be proportional to the socially necessary labor time involved in the entire process of production. Samuelson wrote a seminal paper claiming to prove the transformation problem is impossible to solve. However, Ian Wright, Emmanuel D. Farjoun, Moshé Machover, David Zachariah, and other Marxist economists have posited alternative solutions. Is it possible to resolve this issue using big data and deep learning? We introduce the background of the topic, and we make some suggestions in this direction.

演讲者介绍

I graduated from University of Illinois at Chicago with a PhD in pure math. The topic of my dissertation was the stochastic heat equation on the torus. Past publications focus on Rough Path and Stochastic Analysis. My research interests include interdisciplinary topics such as AI.