Skip to main content

Machine Learning: A Brief Introduction

Prachi is a Machine Learning Engineer, working in the respective field since 2015.


What is machine learning?

As the name suggests, machine learning means “the ability of a machine to learn by itself”. It is a part of artificial intelligence that helps the machine to gain intelligence through experience. They are not explicitly programmed to perform every action. The major focus is on the development of computer programs that machines use to gain wider knowledge in a specific field.

The process starts with data observation that includes instructions, examples and so on. The data is transformed into possible patterns to facilitate the decision-making behavior of the machines.

The process continues until the machine becomes efficient enough to automatically learn new concepts without any human assistance or intervention.

A programmer only needs to write a generic algorithm that can be used the machine to produce a feasible pattern or logic from the input data.

Related content: Machine Learning Algorithms

What are the methods of machine learning?

Depending on the type of raw data available, machine learning is widely studied under two categories:

Supervised Machine Learning

Supervised Machine Learning

1. Supervised Machine Learning

Supervised Machine Learning is used when the machine can successfully process new data all by itself with the help of past experiences. It already has a training dataset available to make predictions about the output of the new data. If the predicted result defers from the actual result, the machine can modify its model accordingly.

Unsupervised Machine Learning

Unsupervised Machine Learning

2. Unsupervised Machine Learning

Unsupervised Machine Learning studies the system when it has no previous classified data available to its rescue. Unlike supervised learning, a machine needs to extract the patterns out of unknown data. It is highly unlikely for the machine to deliver the correct results, rather it has to process the data and fetch the possible patterns.

Semi-Supervised Machine Learning

Semi-Supervised Machine Learning

3. Semi-Supervised Machine Learning

Semi-Supervised Machine Learning makes use of both classified and unclassified data. The amount of classified data is comparatively lesser than the unclassified data. This helps the machine to improve its accuracy and boosts its learning ability.

Scroll to Continue
Reinforcement learning

Reinforcement learning

4. Reinforcement Machine Learning

Reinforcement Machine Learning works on a trial and error basis. The machine learns through interactions and actions. It has to automatically figure out the ideal behavior to maximize its learning accuracy. On the best actions, it is rewarded with positive feedback, technically termed as reinforcement learning. And on the wrong actions, it is penalized.

Machine learning eases the evaluation of the massive amount of data with faster results and higher accuracy along with possible risks.

Examples of Machine Learning

Examples of Machine Learning

Examples of Machine Learning

Machine learning may seem like an advanced concept, but knowingly or unknowingly, each one of us deals with this application on a daily basis. For example,

  1. Online recommendation section on online shopping sites such as Amazon and streaming websites such as Netflix
  2. Call for insurance and other financial offers from banks and relevant companies
  3. Self-driving Google car
  4. Fraud detection
Machine Learning System

Machine Learning System

What are the prerequisites for creating a reliable machine learning system?

Along with the knowledge of certain algorithms and theorems such as Bayesian theorem, to build a reliable system, you must be well-prepared with the following prerequisites:

  1. Data creation and preparation capabilities
  2. Basic and advanced algorithms
  3. Automation and iterative processes
  4. Scalability
  5. Ensemble modeling

What types of industries are using machine learning?

Machine learning is on continuous growth and has already been adopted by several industries, including:

  1. Financial Services: Banks and other financial companies use machine learning to detect fraud and identify important insights in the company’s growth. Machine learning helps them to find the best investment opportunities, recognise clients with high risks, decide the right time to trade and so on.
  2. Government: Government agencies have a massive amount of people’s data, which is needed to be mined to identify certain people’s identify in case of theft and fraud.
  3. Health Care: Health care has been massively benefited by the advent of machine learning. It has become easier for physicians and doctors to assess the real time data of patients’ health and act accordingly. The technology also help them to identify the future health risks and prevention measures.
  4. Retail: Retail websites display recommend products and what others are buying section with the help of machine learning. Since these websites are used by millions. The data need to be analysed thoroughly to determine optimised price, effective marketing campaign, customer insights, merchandise planning and provide the user with the personalised shopping experience.
  5. Oil and Gas: Oil and gas industry is heavily based on exploration. Machine learning helps it to find the new energy and mineral sources. Besides detection, it also predicts refinery sensor failure.
  6. Transportation: Transportation industry has managed to find out highly reliable routes by preventing possible risks with the help of machine learning. It is now easier to predict the road damages, possible hazards and much more.

Though Machine learning has been existing for a century, its usability has become more widespread with the introduction of the Internet. Machine learning is already used by most of the organizations, it has immense potential, which still needs to be discovered.

© 2019 Prachi Sharma

Related Articles