Blog

As Machine Learning and its Disciplines is a wide topic. Machine learning is a powerful and popular tool for solving various tasks in computer science. Its ability to automatically learn from data makes it an attractive choice for many applications. The field of machine learning is divided into several subfields, including supervised and unsupervised learning, reinforcement learning, and deep learning. Deep learning is a subfield of machine learning that uses artificial neural networks to learn. These networks are composed of layers of interconnected neurons, and they can learn to recognize patterns in data by adjusting the strengths of the connections between the neurons.

As machine learning and deep learning are growing day by day, I would like to convey to you an overview as:

What is exactly machine learning is?

On providing data as input and output, machine learns the rule to process the data. After getting trained , it is ready to work. Now for getting results or outputs, machine learning model just needs an input. It is working normally as we run a program by defining our own rules. Machine learning model could have a little difference in result for which we need more AI ready data and more training.

Types of Machine Learning

There are three main types of machine learning: supervised learning, unsupervised learning, and reinforced learning.

1. Supervised Learning

In supervised learning, the computer is given a set of training data, and it is then tasked with learning how to correctly predict the values of certain variables based on the data.

Supervised learning is where a computer system is “trained” with a set of labelled data. The system is then able to learn and recognize patterns in new data. For example, if you wanted to create a computer system that could recognize different types of flowers, you would first give it a set of labelled data (i.e. images of different flowers with their names written underneath). The system would then learn to recognize different patterns in new data (i.e. images of flowers without any labels).

This set of AI called supervised learning, just learns input to output, or A to B mappings. On one hand, input to output, A to B it seems quite limiting. But when you find a right application scenario, this can be incredibly valuable.

  A to B mappings

  Input to Output mappings 

Let us see some examples:

Speech Recognition

If the input is an audio clip, and the AI’s job is to output the text transcript, then this is speech recognition.

Machine Translation

If you want to input English and have it output a different language, Chinese, Spanish, something else, then this is machine translation.

Machine Predictions

All the large online ad platforms have a piece of AI that inputs some information about an ad, and some information about you, and tries to predict, will you click on this ad or not?

Self Driving Cars

If you want to build a self-driving car, one of the key pieces of AI is the AI that takes as input an image, and some information from radar, or from other sensors, and outputs the position of other cars, so your self-driving car can avoid the other cars.

Visual Inspection

In Manufacturing, we take as input a picture of something you’ve just manufactured, such as a picture of a cell phone coming off the assembly line, and you want to output: Is there a scratch, or is there a dent, or some other defects on this thing you’ve just manufactured?

This is visual inspection which is helping manufacturers to reduce or prevent defects in the things that they’re making

2. Unsupervised Learning

In unsupervised learning, the computer is given data but not told what to do with it. It must learn how to organize and group the data on its own.

Unsupervised learning is where a computer system is “trained” with a set of data that has not been labelled. The system is then able to learn and recognize patterns in new data. For example, if you wanted to create a computer system that could recognize different types of fruits, you would first give it a set of data that has not been labelled (i.e. images of different fruits). The system would then learn to recognize different patterns in new data (i.e. images of fruits without any labels).

In this type of machine learning, it learns patterns from without tagged data(without labels). The most common method for such kind of learning is cluster analysis to find hidden patterns or groupings in data. That means exploratory data analysis.

Example:

Genetics DNA clustering patterns to analysis evolutionary biology.

Customer segmentation to build market strategies or other business.

3. Reinforcement Learning

Reinforcement learning is a type of machine learning that involves learning how to achieve a certain goal by performing actions and being rewarded or punished accordingly.

Reinforced learning is a combination of supervised and unsupervised learning. A computer system is “trained” with a set of labelled data, and then it is able to learn and recognize patterns in new data. However, the computer system is also able to “test” its own ideas by trying them out on new data. If the computer system is successful at recognizing patterns in new data, it is rewarded with a “reinforcement signal” (i.e. a positive feedback). If the computer system is unsuccessful at recognizing patterns in new data, it is punished with a “negative reinforcement signal” (i.e. a negative feedback).

It is alongside of Supervised and unsupervised learning concerned with how to suggest intelligent agents to take actions accordingly in an environment to make a sequence of decisions.

Example:

Automatic parking policies for self driving cars. Detection of moving objects around self driving cars.

Machine learning is used in a wide range of applications, include:

• Automatic email filtering
• Speech recognition
• Fraud detection
• Text categorization
• Predicting consumer behaviour
• Targeted advertising
• Predicting stock prices
• Detecting diseases