4 Intersecting Domains That You Can Easily Confuse with Artificial Intelligence

Posted on by admin

← PART 1 | ARTIFICIAL INTELLIGENCE ESSENTIALS

Learn the Differences Between Artificial Intelligence, Machine Learning, Deep Learning, Data Science, and Big Data | AI Essentials

Figure 1. Photo by Michael Dziedzic on Unsplash

Once you start consuming machine learning content such as books, articles, video courses, and blog posts, you will often see the terms like artificial intelligence, machine learning, deep learning, big data, and data science being used interchangeably. These terms represent several closely related areas within the field of artificial intelligence. They are usually used interchangeably without adequate attention is paid to their scopes. It’s not entirely the authors’ fault since there is a slight ambiguity about these terms’ differences. With this post, we will put an end to this ambiguity and clarify their scopes.

We will cover five different adjacent fields:

  • Artificial Intelligence
  • Machine Learning
  • Deep Learning
  • Data Science
  • Big Data

and bonus part, I will share a visual in the end to clarify them even further.

Let’s start!

Figure 2. Photo by Markus Winkler on Unsplash

Artificial Intelligence (AI) is a broad umbrella term, and its definition varies across different textbooks. The term AI is often used to describe computers that simulate human intelligence and mimic “cognitive” abilities humans associate with the human mind. Problem-solving and learning are examples of these cognitive abilities. The field of AI contains machine learning (and, therefore, deep learning) studies since capability learning from experiences is a sign of intelligence. Generally speaking, machines with artificial intelligence is capable of:

  • Understanding and interpreting data;
  • Learning from data; and
  • Making ‘intelligent’ decisions based on insights and patterns extracted from data.

These terms are highly associated with machine learning. Thanks to machine learning, AI systems can learn and excel at their level of consciousness. Machine learning is used to train AI systems and make them smarter. I do not want to get into how the field of AI has developed over the years. But, to have a quick understanding, here is a summary timeline of the development of artificial intelligence:

Figure 3. A Brief Timeline of the Field of Artificial Intelligence (Figure by Author)

This timeline shows the fundamental studies that were done in artificial intelligence. While researchers tried to mimic the neurons between the 50s-70s, they focused on machine learning with expert systems. Beginning from the 2000s, the focus is currently on deep learning studies. As I mentioned, all of these domains are part of artificial intelligence.

Let’s talk more about these relevant domains…

Figure 4. Photo by Pietro Jeng on Unsplash

Machine learning is considered as a sub-discipline under the field of artificial intelligence. Machine Learning (ML) studies aim to automatically improve the computer algorithms’ performance designed for particular tasks with experience. In a machine learning study, the experience is derived from the training data, which may be defined as the sample data collected on previously recorded observations. Through this experience, machine learning algorithms can learn and build mathematical models to make predictions and decisions. The learning process starts with feeding training data (e.g., examples, direct experience, basic instructions), which contains implicit patterns, into the model. Since computers have more processing power than humans, they can find these valuable patterns in the data within a short amount of time. These patterns are -then- used to make predictions and decisions on relevant events. The learning may continue even after deployment if the developer builds a suitable machine learning system that allows continuous training.

Previously, we were able to use machine learning in a few sub-components of a system only. Now we actually use machine learning to replace entire sets of systems, rather than trying to make a better machine learning model for each of the pieces.

“There is an ever-increasing use of machine learning applications in different fields. These real-life applications vary to a great extent.” — Jeff Dean

Some machine learning use cases may be listed as follows:

  • Healthcare: Medical diagnosis, given the patient’s symptoms;
  • E-commerce: Predicting the expected demand;
  • Law: Reviewing legal documents and alerting lawyers about problematic provisions;
  • Social Network: Finding a good match given the user’s preferences on a dating app;
  • Finance: Predicting the future price of a stock given the historical data.

This is obviously a non-exhaustive list, and there are hundreds, if not thousands, of potential machine learning use cases. Depending on what your goal is, there are many different methods to create a machine learning model. These methods are usually grouped under four main approaches: (i) Supervised Learning, (ii) Semi-supervised Learning, (iii) Unsupervised Learning, and (iv) Reinforcement Learning. Each method contains distinct differences in its design, but they all follow the same underlying principles and conforms to the same theoretical background:

Figure 5. Basic Comparison of Four Different ML Approaches (Figure by Author)

After training a machine learning model, you can embed it in an artificial intelligence system. Now, let’s talk about deep learning.

Figure 6. Photo by Josh Riemer on Unsplash

Deep learning (DL) is a sub-field of machine learning that exclusively uses multiple layers of neurons to extract patterns and features from raw data. These multiple layers of interconnected neurons create artificial neural networks (ANNs). An ANN is a special machine learning algorithm designed to simulate the working mechanism of the human brain. There are many different types of artificial neural networks intended for several purposes. In summary, deep learning algorithms are a subset of machine learning algorithms.

Figure 7. An Example of an Artificial Neural Network (Figure by Author)

Just as in machine learning, all four approaches (supervised, semi-supervised, unsupervised, and reinforcement learning) can be utilized in deep learning. When data and computing power are abundant, deep learning almost always outperforms the other machine learning algorithms. Deep learning algorithms are instrumental in image processing, voice recognition, and machine translation. Convolutional Neural Networks, Recurrent Neural Networks, Autoencoders, Generative Adversarial Networks, Transformer Networks are some of the artificial neural network examples which makes deep learning possible.

Figure 8. Photo by Chris Liverani on Unsplash

Data science is an interdisciplinary field that sits at the intersection of artificial intelligence, a particular domain knowledge, information science, and statistics. Data scientists use various scientific methods, processes, and algorithms to obtain knowledge and draw insights from observed data.

In contrast with machine learning, a data science study’s goal does not have to be model training. Data science studies often aim to extract knowledge and insight to support the human decision-making process without creating an AI system. Therefore, although there is an intersection between data science and the other adjacent fields, data science differs from them since it does not have to deliver an intelligent system or a trained model.

Figure 9. Photo by imgix on Unsplash

Big data is a field that aims to efficiently analyze a large amount of data that cannot be processed with traditional data-processing methods and applications. Data with more observation usually brings more accuracy, while high complexity may increase false discovery rates. The field of big data studies how to efficiently capture, store, analyze, search, share, visualize, and update data when the size of a dataset is very large. Big data methods can be used in artificial intelligence (and its sub-domains) and data science. Big data sits at the intersection of all the other fields mentioned above since its methods are crucial for all of them.

Now that we briefly covered all these fields, let’s see the relationship between these domains in a taxonomy diagram.

Figure 10. The Taxonomy of Artificial Intelligence and Data Science (Figure by Author)

This taxonomy is almost clear evidence for the reasons behind the ambiguity. Whenever we are talking about deep learning, we are also talking about machine learning and artificial intelligence. Some might call it a data science project or a big data project when working on a deep learning project. These naming practices are not necessarily incorrect, but they are confusing. Therefore, it is vital to know the intersections and subtractions of these fields.

Now, you know their similarities and differences of these adjacent domains as well as their intersections. In this post, I tried to clarify the differences and I hope you can easily differentiate them

If you would like to have access to full codes of my tutorial posts on Google Colab, and have access to my latest content, consider subscribing to the mailing list:✉️

Subsribe Now

If you are interested in deep learning, also check out the guide to my content on artificial intelligence: