Exploring the Subsets of Artificial Intelligence..
As promised, we’re slowly getting more technical! In this series, we’ll dive deep into the various concepts and terms associated with AI. While most people are familiar with Machine Learning—a well-known subset of AI—there are five more, equally important subsets that often receive less attention. So take a deep breath, and let’s venture deeper into the world of AI!
Deep Learning
Deep Learning is a subset of AI that uses Neural Networks to model complex patterns in large datasets. For those unfamiliar with neural networks, these are algorithms modeled after the neurons in our brain, designed to process data through multiple layers.
Does it work exactly like the human brain? No—but it takes inspiration from the brain's structure.
1-Layer Neural Network (Simplified Example)
There are three layers in a basic neural network:
Input Layer
Hidden Layer
Output Layer
For example, let’s predict whether a student will pass or fail based on their study hours and attendance.
The Input Layer will have 2 neurons (for study hours and attendance).
This input is connected to the hidden layer, with each connection assigned a weight, determining how important each input is for making the prediction.
The weighted sum of the inputs is passed to the Hidden Layer, where an activation function is applied to introduce non-linearity.
(Don’t worry, we’ll have a detailed article dedicated to activation functions soon!)
Image by - Buse Köseoğlu
Natural Language Processing (NLP)
Who hasn’t heard about ChatGPT? It is an NLP algorithm. This is the Subset of AI that deals with interaction between humans and machines, which enables machines to read, interpret and generate human language.
Text Analysis: Extracting meaning from text.
Sentiment Analysis: Detecting emotions or opinions in text.
Speech Recognition: Converting spoken language into text.
Computer Vision
This Subset of AI enables machines to process information from our world, such as photos and videos. How are Tesla cars able to detect humans and other cars? Computer vision. Here algorithms are trained to differentiate between different objects, this is done by labelling thousands on images and using either supervised or reinforcement learning algorithm.
Image Classification: Recognizing objects in images.
Object Detection: Identifying and locating objects in a visual scene.
Facial Recognition: Detecting and recognizing human faces.
Robotics
As the name suggests, Robotics involves training robots to perform tasks. It’s a combination of AI, mechanical engineering, and sensors, enabling machines to interact with their environment and perform tasks autonomously or semi-autonomously.
Sensing and Perception: Robots use sensors to gather data about their environment.
Motion Planning: Calculating how to move through an environment.
Actuation: Controlling the physical movements of the robot.
Expert Systems
AI systems use knowledge base of human experts to solve problems often mimicking decision making of the human experts. This subset I find particularly interesting as maybe overtime machines might be able to how humans make decisions!
Machine Learning
Last but the most famous subset is machine learning, where computers are trained using data and are allowed to learn from it to eventually be able to make their own decisions. There are 3 subsets in machine learning -
Supervised Learning: Learning with labeled data.
Unsupervised Learning: Learning patterns from unlabelled data.
Reinforcement Learning: Learning through interaction with the environment, receiving rewards and penalties.
Now that we are finally done with introductions I can move on to more focused and detailed articles where we will explore each concept in detail! I am excited for the upcoming articles, I hope you are too! Until next time.
