Skip to content

Latest commit

 

History

History
5 lines (5 loc) · 2.27 KB

Introduction.md

File metadata and controls

5 lines (5 loc) · 2.27 KB

Introduction

About a year ago, I began to learn about neural networks on my own time. Machine learning, specifically utilizing neural networks, is a field that every major technology company is performin research in today. Google translate is a neural network. Facebook uses neural networks. There are many more such examples. If one wants to get into the technology field, then studying neural networks is a pretty safe option that is unlikely to fade away soon. Over the past year, I started to learn about how they function and their uses. As time went on, I decided that I wanted to apply the knowledge that I had gained, and realised that senior project would be a perfect opportunity for me to do so. My actual knowledge prior to this project was limited; I had known a decent amount about neural networks from studying them over the summer. However, I only knew about the theory behind neural networks, not how to apply the theory; to give an analogy, if I were building a house, then I knew how to create a blueprint; I needed to learn how to physically build the house. I knew from the start of this project that, despite my ambitions, I would probably not create a network that would make large advancements in the field; I instead hoped that some of the things that I would learn while creating this network would help others in creating their own. My goal was to do something slightly different from what I had previously done, but still be something that I am familiar with. I decided to work with images, as I had worked with images before. However, instead of performing traditional classification, I decided to instead have my network look What I created can be looked at as a default scenario for someone who would attempt to create a neural network that has not been created before; there are publically available networks to base your work off of, but there is no existing database. This means that although it is not necessary to create a neural network from scratch, a database still needs to be made. In order to accomplish my goals, I needed to gain a large amount of base knowledge. While completing my project, I would learn low level TensorFlow, modeling with blender (both manually and through coding), python, the strange ways that neural networks function, and how to debug programs.