Artificial Intelligence is all the rage! Suddenly, everyone, whether they understood it or not, was talking about it. The trend toward artificial intelligence seems unstoppable, but it really comes down to two very popular concepts: machine learning and deep learning. But recently, deep learning has become increasingly popular as it reigns supreme in terms of accuracy when trained with large amounts of data.

To show you the attention deep learning is getting, here are the keyword trends on Google:

Google’s “deep learning” trends

Now in the software industry We are moving towards machine intelligence. Machine learning has become an essential part of various industries as a way of making machine intelligence. In a simpler way, machine learning is a set of algorithms that can parse data, learn from them, and then apply what they learned to do Make informed decisions.

Examples of machine learning are everywhere. This is how Netflix knows which show to watch next or how Facebook recognizes your friends’ faces in digital photos. Alternatively, a customer service representative will know whether you'll be happy with their support before you take a CSAT survey.

The thing about traditional machine learning algorithms is that they look complex, but they still behave like machines. They require a great deal of domain expertise, and human intervention can only meet their needs, no more and no less. And for AI designers and the rest of the world, this is where deep learning becomes more promising.

What is deep learning?

In fact, deep learning is a subset of machine learning that achieves great power and flexibility by learning to represent the world as a nested hierarchy of concepts, each of which is related to simpler concepts. concepts are associated, while more abstract representations are computed with less abstract concepts.

To put it more precisely, deep learning technology gradually learns categories, such as letters, through a hidden layer structure, and then defines higher-level categories (such as words) and higher-level categories (such as sentences). In the case of image recognition, it means identifying light and dark areas before classifying lines, and then identifying shapes to allow recognition of faces. Each neuron or node in the network represents an aspect of the whole, and together they provide a complete representation of the image. Each node or hidden layer has a weight that represents the strength of its relationship to the output, and the weights are adjusted as the model evolves.

Deep Learning Architecture

Distinguishing Features of Deep Learning

One of the great advantages of deep learning, and a key part of understanding why it has become popular, is that it is Massive data driven. The "big data era" of technology will provide huge opportunities for new innovations in deep learning. As Ng puts it, "AI is similar to building a rocket ship. You need a huge engine and a lot of fuel. If you have a big engine and very little fuel, the rocket ship can't get into the right orbit. Or, if You have a small engine and a ton of fuel, and you can't even get a rocket ship to take off. To build a rocket, you need a huge engine and a lot of fuel.

To make an analogy with deep learning, In other words, the rocket engine is the deep learning model, and the fuel is the large amount of data we let these algorithms learn. ”

Deep learning requires high-end machines as opposed to traditional machine learning algorithms. GPUs have now become integral to executing any deep learning algorithm.

In traditional machine learning techniques, most application features need to be identified by domain experts in order to reduce the complexity of the data and make the pattern easier for learning algorithms. The biggest advantage of the deep learning algorithms discussed earlier is that they attempt to learn high-level features from data in an incremental manner. This eliminates the need for domain expertise and hard core feature extraction.

Another difference between deep learning and machine learning techniques is the approach to problem solving. Deep learning techniques tend to solve problems from start to finish, while machine learning techniques require breaking down the problem statement into different parts. Solve the problem first and then merge its results in the final stage.

For example, for multi-object detection problems, deep learning techniques like Yolo networks take images as input and provide the location and name of the object at the output. But in usual machine learning algorithms like SVM, a bounding box object detection algorithm is first required to identify all possible objects to take HOG as input to the learning algorithm to identify relevant objects.

Usually, deep learning algorithms take a long time to train due to the large number of parameters. The most popular ResNet algorithm takes about two weeks to train completely from scratch. While traditional machine learning algorithms take seconds to hours to train, this scenario is completely reversed during the testing phase. When tested, deep learning algorithms took much less time. However, if you compare it with the nearest neighbor algorithm, a machine learning algorithm, the test time increases as the data size increases. While this does not apply to all machine learning algorithms, some of them also have very short test times.

Explainability is a major issue why many industries use other machine learning techniques over deep learning. Let’s take an example and say we use deep learning to calculate the relevance score of a document.

The performance it provides is very good and close to human performance. But there's a problem, it doesn't reveal why it gave this score. In fact, you can mathematically find out which nodes of a deep neural network are activated, but we don't know what the neurons are supposed to model and what these layers of neurons do together. So we cannot interpret the results. This is not machine learning algorithms like decision trees, logistic regression, etc.

When to use deep learning or not?

1. Deep learning performs other techniques if the amount of data is large. But due to the smaller amount of data, traditional machine learning algorithms are more preferable.

2. Deep learning technology requires high-end infrastructure to conduct training within a reasonable time.

3. When there is a lack of domain understanding for feature introspection, deep learning technology surpasses other fields because you don’t have to worry about feature engineering.

4. Deep learning really excels in complex problems such as image classification, natural language processing, and speech recognition.