Why I’m Excited about Artificial Neural Networks

Why I’m Excited about Artificial Neural Networks

I read a series of articles 8 months ago on a popular website called Wait But Why. The articles explain the types of AI, the force behind them, and how close we are to achieving true AI. Feel free to read the articles, they are easy to read and understand.

These articles made me think for days and weeks. They made me more aware and I started to focus more on the smart systems we interact with everyday, even if they are not visible. I though about the narrow AI behind product recommendations from Amazon, traffic notification alerts, Amazon Alexa understanding my commands, and many more. I became fascinated and I wanted to learn more about these smart systems around me. So that is what I did, I started reading basic articles on AI and watching a lot of videos on Youtube, like we all do (I hope I'm not the only one).

After a few days of research I realized two things. The first thing is that there are breakthroughs occurring almost everyday. DeepMind would be on the news one day for something cool they did and OpenAI the next day. The second thing is that Artificial Neural Networks (ANN) are pretty darn cool. Most, if not all, major breakthroughs in AI involve Artificial Neural Networks.

What are Artificial Neural Networks?

Artificial Neural Networks

Here are a few definitions:

An artificial neuron network (ANN) is a computational model based on the structure and functions of biological neural networks. Information that flows through the network affects the structure of the ANN because a neural network changes - or learns, in a sense - based on that input and output. - Techopedia

*** ...a computing system made up of a number of simple, highly interconnected processing elements, which process information by their dynamic state response to external inputs. - Maureen Caudill***

Artificial Neural Networks are programs modeled after a part of the brain called the neocortex. It is involved in higher functions such as sensory perception, generation of motor commands, spatial reasoning, conscious thought, and language. These programs, after being trained, are able to accomplish various tasks at human level ability. Some of these tasks include sentiment analysis, image classification, face recognition, language translation, natural language processing, and much more. These programs, just like our brains, typically need large datasets to train on.

Cat or Dog

We as humans learn to accomplish task like these through experience and as well as evolution. We learn what a cat and dog is by analyzing their features using the data passed by our senses. Our neurons get trained to distinguish a cat vs dog so that next time we hear a "bark" we know what it is. These programs work in a similar way, they consist of many neurons that are able to distinguish a cat vs dog when given enough data.

I encourage you to learn more about Artificial Neural Networks and perhaps study them. If you are scared of the math then you don't have to because there are high-level libraries that abstract that away. If you love math then you will love Artificial Neural Networks. Think of problems you can solve in your area of work or study and go solve them.

Below are some resources you might find useful:

Edgar Pino

About Edgar Pino

I’m a passionate software engineer working @ Pluralsight interested in distributed systems, machine learning, and the web. I am also a graduate student at the University of Illinois.