Artificial Intelligence vs Machine Learning vs Deep Learning (AI vs ML vs DL)

Alan Davis Babu
10 min readNov 5, 2019

--

Artificial Intelligence is the broader umbrella under which Machine Learning and Deep Learning come. And you can also see in the diagram that even deep learning is a subset of Machine Learning. So all three of them AI, machine learning and deep learning are just subsets of each other. So let us move on and understand exactly how they are different from each other.

The term artificial intelligence was first coined in the year 1956, but AI has become more popular these days. Why? Well, it’s because of the tremendous increase in data volumes, advanced algorithms, and improvements in computing power and storage.

The data we had was not enough to predict the accurate result. But now there is a tremendous increase in the amount of data. Statistics suggest that By 2020, the accumulated volume of big data will increase from 4.4 zettabytes to roughly 44 zettabytes or 44 trillion GBs of data.

What is Artificial Intelligence?

Artificial Intelligence is a technique which allows the machines to act like humans by replicating their behaviour and nature.

To understand Artificial Intelligence vs Machine Learning, let’s look at the categories of Artificial Intelligence.

There are two categories of Artificial Intelligence:

Weak Artificial Intelligence: In this, the machine programs act according to a well-defined response. They are confined to a set of rules that we provide, and they give response within the domain of those rules.

A common example can be a home appliance like an oven. We can set the timer and temperature in it according to the need. It performs the task according to the given instructions. It does not have the ability to make decisions and make any changes by itself. So, it comes under the category of Weak Artificial Intelligence.

Strong Artificial Intelligence: Machine Learning and Deep Learning comes under the category of Strong Artificial Intelligence. It involves designing algorithms for machines that try to learn by themselves using the input data and improve the accuracy in giving outputs.

Examples of Strong Artificial Intelligence are speech recognition, visual perception, and language translation. In language translation, a machine extracts the meaning of words and then the meaning of sentences. After that, it searches for similar-meaning words and sentences in another language and then translates them to that language. As it requires many tasks to accomplish translation and involves decision-making to put everything right, it comes under the category of Strong Artificial Intelligence.

By now, we have seen ‘What is Artificial Intelligence?’ After this, to make the picture more clear for Artificial Intelligence vs Machine Learning vs Deep Learning, we will now move on to Machine Learning.

Machine Learning came into existence in the late 80’s and early 90’s. But what were the issues with the people which made Machine Learning come into existence?

Statistics: How to efficiently train large complex models?

Computer Science & Artificial Intelligence: How to train more robust versions of the AI systems?

Neuroscience: How to design operational models of the brain?

Machine Learning is the subset of AI where Machine Learning algorithms are designed in such a way that the machine tries to learn by itself without being explicitly programmed on each and every instruction. So, as it will be exposed to more and more data, it tries to internally modify itself and adjust according to the data to which it is exposed so that it will not rely on human experts to program them.

What is Machine Learning?

“Machine Learning is a subset of artificial intelligence. It allows the machines to learn and make predictions based on its experience(data)“

Understanding Machine Learning with an Example

Let’s say you want to create a system which could predict the expected weight of a person based on its height. The first thing you do is collect the data. Let us say this is how your data looks like:

Each point on the graph represents one data point. To start with we can draw a simple line to predict the weight based on the height. For example, a simple line:

W = H — 100

Where W is weight in kg and H is height in cm

This line can help us to make predictions. Our main goal is to reduce the difference between the estimated value and actual value. So in order to achieve it, we try to draw a straight line that fits through all these different points and minimise the error and make them as small as possible. Decreasing the error or the difference between the actual value and the estimated value increases the performance.

Further, the more data points we collect, the better will our model become. We can also improve our model by adding more variables (e.g. Gender) and creating different prediction lines for them. Once the line is created, so in future, if a new data (for example height of a person) is fed to the model, it would easily predict the data for you and will tell his predicted weight.

There are basically three types of Machine Learning as shown below:

Supervised Learning

In supervised learning, the machine is provided with the labeled dataset. It already has input and output parameters. So, when the machine is given a new dataset, the supervised learning algorithm examines the data and produces the correct output according to the labeled data.

Use case: Detecting cancer patients

There would be some parameters and symptoms given for detecting cancer patients. The machine will try to classify the patients according to the symptoms to determine whether they are having cancer or not. So, supervised learning is best for classification and regression problems.

Unsupervised Learning

In unsupervised learning, the machine would not have any labelled dataset. The algorithm is designed in a way that it tries to learn by itself without any supervision of data. This involves clustering of data.

Example: Consider few objects such as pencil, eraser, and matchbox

Here, the machine does not even know what these objects are; rather, it makes clusters of similar objects, and when any input dataset is given, it gives the output by examining the data it has clustered.

Netflix recommendation system works on the same technique as it saves the users’ watched history and recommends a similar content to the user.

Reinforcement Learning

In reinforcement learning, the algorithms are designed in such a way that the machine tries to find an optimal solution. It adopts the principle of reward and punishment, and by this approach it moves to the correct result.

Consider a scenario where a young cricketer tries to learn the technique of hitting a 6 for a particular shot. Whenever he tries to play the shot and misses it, he gets a score of −1.

The bowler bowls again, and the batsman tries to hit that shot by adjusting his position. After trying for five to six times, he finally hits a 6 with a suitable position and gets a score of +6.

This way he learns the technique for hitting a 6. Reinforcement algorithm works in a similar way.

Types of Machine Learning at a Glance

Now, while understanding Artificial Intelligence vs Machine Learning vs Deep Learning, here is the last topic down the hierarchy that is Deep Learning.

Deep Learning has evolved from Machine Learning. It works in a layered architecture and uses the artificial neural network, a concept inspired by the biological neural network. The human brain usually analyses and converts the information it receives and tries to identify it from the past information the brain has stored. The brain does this through labelling and assigning information into various groups in a fraction of a millisecond.

What is Deep Learning?

“Deep learning is a particular kind of machine learning that achieves great power and flexibility by learning to represent the world as nested hierarchy of concepts or abstraction”

The concept of deep learning is not new. But recently its hype has increased, and deep learning is getting more attention. This field is a special kind of machine learning which is inspired by the functionality of our brain cells called artificial neural network. It simply takes data connections between all artificial neurons and adjusts them according to the data pattern. More neurons are needed if the size of the data is large. It automatically features learning at multiple levels of abstraction thereby allowing a system to learn complex functions mapping without depending on any specific algorithm.

So, to understand Deep Learning vs Machine Learning, you must know the difference between Machine Learning and Deep Learning and the major difference is that we need to provide the features manually in Machine Learning. But in Deep Learning, it automatically extracts features for classification which in turn demands a huge amount of data for training DL algorithms. So, in Deep Learning, the accuracy of the output depends on the amount of data.

Now, let us look at artificial neural networks in Deep Learning.

Artificial Neural Network (ANN)
It is a concept inspired by the biological neural network. It consists of three layers:

Input Layer: The input layer is used for taking the input data from external sources and then passing it on to the hidden layers of the neural network. It does not perform any computation.
Hidden Layer: This layer consists of many hidden layers. All the computation is performed in this layer. After all the computation is done, it passes the output to the output layer.
Output Layer: This layer is used for computing and giving the output to the outside world.

These layers consist of nodes that are interconnected with each other. The nodes interact with each other with the help of links by which they are connected. This connection of nodes is designed in such a way that it produces output for a given input.

The links are associated with a real number that is called the weight of those links. These weights are initialized randomly, and hence there could be a large difference between the actual values and the predicted values. Due to this, it will not give the desired output in one iteration.

Even after the weight is assigned and the computation is done if it does not give the desired output, we go back and update the weight of the link with the current value to get closer to the desired output. We do this progressively until we get the best possible output.

Also, the weights assigned to the links decides how fast the triggering of the activation function will occur.

There is another term ‘bias,’ which is used to decide when to trigger the activation function.

Now, we will see the details of the activation function.

In the real world, the data is always 3-dimensional. For example, let’s say, an image of a car is given as an input to the neural network; we can plot the length and height of the car in a 2-D plane. However, there are a lot more attributes to be considered while computing to recognize it as a car. The computation of this data is a complex task. So, to reduce this difficulty, we use the activation function. It pumps the wide range of values assigned to the data into a specific domain so that computation becomes easy.

This is how the artificial neural network works and helps in achieving perfection in Artificial Intelligence.

Let’s take an example of a machine which recognises the animals. The task of the machine is to recognize whether the given image is of a cat or a dog.

What if we’re asked to resolve the same issue using the concepts of machine learning, what we would do? First, we would define the features such as check whether the animal has whiskers or not, or check if the animal has pointed ears or not or whether its tail is straight or curved.

In short, we will define the facial features and let the system identify which features are more important in classifying a particular animal.

Now when it comes to deep learning. It takes this to one step ahead. Deep Learning automatically finds out the features which are important for classification, comparing to Machine Learning where we had to manually give the features.

By now I guess my blog- AI vs Machine Learning vs Deep Learning has made you clear that AI is a bigger picture, and Machine Learning and Deep Learning are its subparts, so concluding it I would say the easiest way of understanding the difference between machine learning and deep learning is to know that deep learning is machine learning. More specifically, it’s the next evolution of machine learning.

--

--