Document worth reading: “Why Machines Cannot Learn Mathematics, Yet”

Nowadays, Machine Learning (ML) is seen as a result of the frequent decision to boost the effectiveness of knowledge retrieval (IR) methods. However, whereas arithmetic is a precise and proper science, it is usually expressed by a lot much less appropriate and imprecise descriptions, contributing to the relative dearth of machine learning functions for IR on this space. Generally, mathematical paperwork discuss their knowledge with an ambiguous, context-dependent, and non-formal language. Given present advances in ML, it seems canonical to make use of ML strategies to represent and retrieve arithmetic semantically. In this work, we apply well-liked textual content material embedding strategies to the arXiv assortment of STEM paperwork and uncover how these are unable to accurately understand arithmetic from that corpus. In addition, we moreover study the missing sides that may allow arithmetic to be realized by pc programs. Why Machines Cannot Learn Mathematics, Yet