Projection as a way of thinking to find factors in factor space

Abstract

With the development of artificial intelligence, statistical learning methods are widely used. Among all artificial intelligence problems, obtaining excellent training models through massive data mining and huge computing power is currently a hot topic. Although such statistical learning methods have shown great potential in academic and industrial applications, the causal logic of machine learning still has poor interpretability. In the factor space theory proposed in 1982, Professor Peizhuang Wang first proposed causal reasoning to find the relationship between causation and effect and created factor space to illustrate causal relationships. Factors play a vital role in the factor space, which can be applied in various scenarios. The projection method provides a way of thinking and perspective for finding factors by projecting the vectors in the n-dimensional real vector space Rn into subspace, and this method has great value in knowledge representation and causal reasoning.