Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL) are buzz-words today. Many people use them interchangeably. However, AI, ML, and DL are NOT the same.
According to John McCarthy, who coined the term “Artificial Intelligence” in 1956, AI can be defined as “the science and engineering of making intelligent machines.” Simply put, AI is any technique or algorithm that enables humans to mimic, develop and demonstrate human intelligence. There are several techniques to achieve AI. These techniques include ML and DL.
Machine Learning is a subset of AI techniques which use statistical methods to enable machines to improve with experience.
DL is a particular type of ML. It is inspired by design and working of our brain cells. The human brain contains a network of neurons. In DL, a network of artificial neurons is used (called artificial neural network or ANN). ANN contains several layers of processing units. The output from one layer is used as input for the next layer. The word “deep” is attribute to the multiple layers of processing units.
Just like humans learn from experiences, systems can be made to learn by providing them with data-sets. By providing more and varied data-sets, systems can become more “intelligent” over time.
Machine Learning enables devices to perform tasks without being explicitly programmed. It is based on the models and methods borrowed from statistics and probability theory.
DL is a subset of ML. The focus of DL is on identifying data patterns rather than solving any particular problem.
Real Life examples of ML
- Image Recognition
- Speech Recognition
- Medical Diagnosis
- Financial Services
Some Companies using ML
- Google – Gmail, Google Photos, Google Assistant, Google Camera,
- Snapchat – uses augmented reality and ML for generating customized selfies.
- Netflix – identifies user’s behavior and provides customized content.
Image Recognition Using ML
Consider that we have been assigned the task of developing a system that can identify objects in images. Here, ML can come to our rescue.
E.g. if an image contains a car, the system should be able to identify it.
In order to solve this problem, let us understand how we, as humans, identify cars. Our mind has a set of features that identify a car, like 2 head lights, 4 wheels etc. When we see any object that has these features, we identify it as a car.
The same approach goes with ML. First, we have to provide a set of features to the system. When the system is provided with an image containing a car, it will try to compare the image with a set of features that it has stored.
The accuracy and scope of an ML system depend on the training data.
Real Life examples of DL
- Self-Driving Cars
- Voice Search
- Image Recognition
- Predicting natural disasters like Earthquakes and Floods
- House-keeping robots
Companies using DL
- Tesla – for self-driving cars
- Apple – Siri
- Google – Google Duplex uses DL for natural language and audio processing
- Amazon – Alexa
Automatically coloring images using DL
When a black-and-white image is passed through a DL-enabled system, processing takes place at multiple layers of ANN. In the starting layers, basic objects are identified. In the later layers, colors are applied based on the colors of matching objects (that the system has been trained on).