In our ongoing effort to demystify AI for business owners and executives, our team at WeAreBrain has just released a book titled, Working Machines – An Executive’s Guide to AI and Intelligent Automation. In it, we cover the various uses of AI and automation, and provide a detailed explanation of how to develop your own AI strategy for your business. It is now available for purchase on Amazon, Google Books, Apple Books and Audible. Buy it today and outperform your competition tomorrow.
Artificial Intelligence (AI) and Machine Learning (ML) are terms thrown around a lot these days and they are often used interchangeably. Although they are connected in many ways, they are two different processes that create different results. And where does Deep Learning (DL) fit in? Already confused?
Don’t worry, we’re here to help demystify all the terms, ideas and jargon so you are able to understand these interesting technologies. Once you understand these processes better you will begin to realise just how much of your life revolves around these technologies.
Let’s get started…
What is Artificial Intelligence?
Before we dive into what machine learning and deep learning are and the functions they perform, we should clear up the confusion about Artificial Intelligence first. We’ve written a fair amount of articles about AI so go check them out if you require more detail. What we’d like to do here is just give you the overall context:
AI is the broader scope of the entire system. Simply put, AI is the process where we use computers to mimic the cognitive abilities and functions of humans in order for machines to carry out tasks in an intelligent way. We do this by creating sets of algorithms by which computers navigate and complete these processes. Got it? Good. Now let’s go into more detail.
What is Machine Learning?
Machine learning is a subset of AI which focuses primarily on the ability of machines to receive data and then learn for themselves. Computers can be trained to automate tasks that would be impossible and time-intensive for a human to complete.
Machine learning requires little human intervention after deployment as it uses data to feed an algorithm that can understand the relationship between input and output. From this, computers can form their own predictive capabilities based on the amount of data input they receive. They can also alter algorithms as they learn more about the information they are processing.
ML’s capabilities are best explained via the idea of popular music streaming applications which are able to suggest similar songs or artists based on the ones you listen to. It receives data, then uses algorithms to find patterns in other data to provide you with more recommendations (i.e. ‘if you like this artist, then you will likely enjoy these others’). The suggestions get better and better the more you listen because ML’s biggest pro is that it continues to learn and sharpen its understanding with the more data it processes.
Getting it so far? We thought so. Let’s continue…
How do machines think like humans?
We are able to train computers to think like we do partly through the use of neural networks. Neural networks are a series of algorithms that are modelled after the human brain. Sounds scary, right? Don’t be alarmed, the robots aren’t taking over just yet. Neural networks aim to mimic our brain’s ability to recognise patterns and categorise information.
Much like our own brain, neural networks try to make sense of the information it receives and processes, and to do this it needs to assign items to categories. When we discover something new, our brain quickly attempts to compare it to similar things in order for us to identify and make sense of it. For example, much like how we know an apple is an apple partly because it is not an orange (but still is a fruit), neural networks try to achieve the same categorisation disciplines based off these principles.
Neural networks are highly beneficial as they are capable of extracting meaning from complicated data and through this, they are able to detect trends and identify patterns often too complex for humans to decipher. They achieve this by learning by example and they do this really, really fast.
Still with us? Great! Now let’s go deeper…
What is Deep Learning?
As machine learning is a subset of artificial intelligence, so deep learning is, in turn, a subset of machine learning. Think of it as the last, tiny piece of the Russian doll of AI. Deep learning can sometimes be referred to as ‘deep neural networks’ as they have many layers involved. Where a ‘normal’ neural network has only one single layer of data to process, deep learning can have multiple layers, hence the name.
Because multiple layers of data are being processed, a hierarchy of related concepts and decision trees are created. This leads to numerous variations of one input being processed, as to answer one question, deep learning finds multiple leads/variations to source deeper related questions as a result.
For example, if you ask a computer to find denim jeans, AI will show results relating to shops in your broad area which sell all kinds of denim jeans. ML will narrow down the desired pair of jeans based on the data it receives and learns by way of your search trends. And deep learning will not only do the above but also investigate the type of denim, the thread count, the colour, what jeans are, when they were created, who wears them, etc. It’s like a rabbit hole — a digital matrix of a rabbit hole.
Deep learning networks rely on the amount of data they receive. The data needs to be good, ‘clean’ data in order to make effective predictions and yield better results. The difference between ML and DL is that deep learning doesn’t require its parameters to be defined and programmed by a human, the systems learn from exposure to big data. It trains itself.
Deep learning is therefore highly beneficial if you have a lot of data intended to find specific use cases for, in order to derive multiple interpretations. It is particularly useful as it solves problems too complex for machine learning capabilities. But this power comes with a cost — computing power and resource cost. In order to get the full benefit of deep learning, you will need to have incredibly powerful computing machines, and a lot of money to get your hands on them. This, coupled with a lot of time, as deep learning takes a few days to spew out (albeit impressive and highly complex) interpretations.
To wrap it up
There you have it — all the basic ideas regarding the most important and powerful technologies on the planet. This is of course just a broad overview of what these technologies are, as to go into them in detail will require a lot more information (and brain processing power). But for those of you who needed a straightforward understanding of what these processes are and what functions they perform, we hope this outline was beneficial.
Now when you’re out with friends and the topic of AI comes up, you can have the floor and impress all with your (broad, tip of the iceberg) knowledge. Go get ‘em!
Read the previous parts of our series on AI:
- Part 1: What does an AI engineer do all day?
- Part 2: Exploring the industries which benefit from artificial intelligence
- Part 3: Getting started with an AI strategy and roadmap
- Part 4: What is an AI Canvas and how do you use it?
- Part 5: Using a Machine Learning Canvas