Saying that computer science, or even artificial intelligence, is a broad field is quite an understatement. In my first week of research, I've barely scratched the surface of my topic, which is both an exciting and daunting prospect.
My first dive into artificial intelligence took me through a brief overview of computer vision and neural networks. In essence, computer vision is how a system can see the world. Although it sounds simple enough, creating a system for computer vision is very challenging. This is because even if the computer has a camera and can technically "see", it doesn't inherently know how to identify and distinguish objects.
Picture by rupixen from Unsplash
This is where neural networks come into play. Neural networks, a type of machine learning model, aim to imitate the neural network of the human brain.
They do this through layers of nodes, with an input and output layer as the outside layers. Based on the input and the supposed result, the neural network adjusts the weightage of the nodes in the layers in between the input and output, also called the hidden layers. Through high quantities of data, a neural network can use this method to create and train a very accurate model that is suited for a specific task (which, in the case of computer vision, could entail identifying and differentiating objects).
Even though I learned the basic premises of neural networks, my research has opened up many doors of questioning to me. What other types of neural networks are there, and why would a computer scientist choose one over the other for a given situation? Are there other ways to create a system for computer vision than neural networks? And zooming outwards in scope, what other subsets of artificial intelligence are there and how do they interact with each other?
This last question is one that I feel will be very pertinent to my research in the coming weeks. Before I jump into the specifics of one subtopic, knowing the foundational concepts of the field itself will help me have a more concrete understanding of the applications of my research and how it could interact with existing concepts and ideas.
As a result, I can start thinking ahead about how to apply my knowledge to an original work and a final product in order to create something that could help launch me towards my future.