Applications of Functional Programming To Machine Learning

点击翻译按钮获取中文摘要

This project revolves around the area of functional programming and recursion schemes and exploring its relationship with deep learning; this intends to research how functional concepts can integrate with and benefit the understanding of neural networks. I intend to focus on analysing the functional abstractions that can exist in the structure and training of neural networks, and how recursion schemes can be found as an inherent characteristic. This will engage with existing flaws in current neural network development, and aims to show how better transparency, representativeness, and reliability can be achieved in unifying the fields of functional programming and deep learning. The project will involve implementing a recursion-scheme based neural network library whilst formalizing its functionality into a practical theoretical understanding.