The State of the Theory of Deep Neural Networks
My final-year research project: A summary of the theory behind deep neural networks.
Overview
For my final-year research project in university, I wrote about the mathematical foundations of deep neural networks. In the document, I cover the mathematical basics behind neural networks, explore some more interesting architectures such as CNNs, Reinforcement Learning and Transformers, then wrap up with some experiments on the effects of varying the depth of these models.
This project involved
- ›Combed through dozens of research papers over the course of the project.
- ›Distilled the key information in my own words to solidify my understanding.
- ›Greatly improved my Python programming proficiency.
- ›Coded many types of neural network using TensorFlow and from scratch.
- ›Collated all the relevant information into an organised, easy-to-follow narrative.
- ›Presented it clearly and concisely, so that readers unfamiliar with the field can follow along.
If you're interested, you can read my full paper below:
This was my first ever academic research paper. I learned and honed so many new skills throughout this project, and I'm thrilled with the result. The sense of satisfaction from finishing it was immense and I'd be very interested in potentially pursuing further research in this fascinating field!