Demystifying AI, Machine Learning, and Deep Learning

Artificial Intelligence (AI) is the big overall bubble with many different subsets. One subset of AI is Machine Learning (ML), which deals with decision-making.

Within ML, another category or type of learning is Artificial Neural Networks (ANN). Various other categories/algorithms under ML, such as decision trees, support vector machines, etc., depend on the problem and the task.

Within ANN, another group or subset is focused on Deep Learning (DL). There are also other kinds of ANN that don’t use DL.

Some of the examples are:

Natural Language Processing. It sits in AI as a type of AI.; It contains some elements of Machine Learning, Artificial Neural networks and Deep Learning. It uses all of these and is not restricted to one.

Chat GPT is an example of Deep Learning.

Another example is Computer Vision, which can contain elements from ML, ANN, and DL.

“No free lunch” 
No one algorithm is better than the rest on all possible problems.        

Machine Learning:

Next-level Analytics. It's data-driven and, at its core, is about detecting patterns in a large amount of data.

Features of ML

• Self-directed computer system
• Algorithms that improve with experience
• Finds patterns and nuances in datasets
• Three types of Machine Learning
• Supervised learning - Give a list of no/yes examples, and the system learns from it. It's a bigger kind of learning.
• Unsupervised learning - Clustering. No known relation, but the system identifies the relationships from clusters of data. It’s a smaller category of learning.
• Reinforcement learning - It’s the smallest category of learning. Rewarded for good actions and penalised for bad actions. Like in Robotics.

Artificial Neural Networks (ANN):

Brain-inspired systems intended to replicate the way that humans learn.

Features of ANN

• Consist of input and output layers, as well as a hidden layer (as many as you want) consisting of units that transform the input into something that the output layer can use
• Highly adapted to finding patterns which are far too complex or numerous for a human programmer to extract and teach the machine to recognise

Nature & Capabilities of ANN -

Self-Learning.

Doesn't require structured data.

Capable of working on very large data.

Convoluted workings.

Deep Learning (DL):

The “Deep” part refers to the number of hidden layers in the neural network. Its a subset of Machine Learning.

Features of DL

• Deep learning is at the cutting edge of AI
• Allows for processing huge amounts of data to find relationships and patterns that humans are often unable to detect
• Progress in Deep Learning has been driven by huge growth in data, such as from the Internet, and processing power
• Deep learning has driven impressive progress of machine learning, with many different models accomplishing extremely difficult tasks such as:
• Image and speech classification
• Image and video manipulation/generation
• Digital assistants
• Autonomous driving

Blog Post Summary: Demystifying AI, Machine Learning, and Deep Learning

This blog post dives into the world of Artificial Intelligence (AI) and its various components, breaking down complex concepts into simpler terms.

Key Learnings:

• AI is an umbrella term encompassing various subfields, including Machine Learning (ML).
• ML deals with decision-making based on data analysis, with different algorithms like decision trees and neural networks used depending on the task.
• Artificial Neural Networks (ANNs) are inspired by the human brain and can learn complex patterns without explicit programming.
• Deep Learning (DL) is a subset of ML using ANNs with many hidden layers, allowing for processing massive data and uncovering intricate relationships.
• Examples of DL applications include image/speech recognition, digital assistants, and autonomous driving.

In essence, the blog post helps us understand how AI, ML, and DL work together to achieve seemingly intelligent tasks, highlighting their capabilities and potential.