In recent years, AI (artificial intelligence) has been widely used in many fields, but machine learning is often talked about together with AI. However, many people may not be able to explain the relationship and differences between them.
In this column, we will introduce basic knowledge of machine learning, how it differs from AI and deep learning, and examples of how machine learning is used. If you are worried about “I want to know the basics of machine learning” or “I want to consider introducing it into my business, but honestly I don’t really understand it,” please read until the end.
INDEX
- What is machine learning?
- Three learning methods in machine learning
- Examples of machine learning usage around us
- Machine Learning FAQ
- summary
What is machine learning?
Machine learning is a data analysis technology that involves giving a computer sample data and learning iteratively to discover underlying patterns and rules. In recent years, it has been increasingly used in various fields and companies and is attracting attention.
Similar terms to machine learning include “deep learning” and “AI.” What is the difference between these and machine learning?
Difference between machine learning and deep learning
Deep learning is positioned as a type of machine learning.
The difference between machine learning and deep learning is the way computers learn. By incorporating a method called neural networks, deep learning allows for more accurate analysis than traditional machine learning.
*Note: A neural network is a mathematical model that mimics the network structure of nerve cells (neurons) in the human brain.
Machine learning is learned by humans instructing the purpose and content, but the difference with deep learning is that the computer itself determines what to learn and then learns.
Difference between machine learning and AI
Machine learning is a technology that allows AI to perform tasks.
Machine learning is an analysis technology included in AI, and AI and machine learning have a comprehensive relationship. Also, within machine learning, deep learning is a learning method that incorporates neural networks.
To summarize the relationship between each, it is AI > Machine Learning > Deep Learning.
Three learning methods in machine learning
The three learning methods in machine learning are as follows.
- supervised learning
- unsupervised learning
- reinforcement learning
Let’s take a closer look at each learning method.
supervised learning
Supervised learning is a method of making a computer learn using data that includes correct answers.
By giving the correct answers to a computer, it learns features and patterns, allowing it to classify and predict data. Here are some usage examples.
- translation
- Spam email classification
- Sales forecast
- image recognition
For example, until now, computer image recognition has been nothing more than a list of pixel values in image data. However, if you use machine learning’s supervised image recognition technology to learn a large number of photos and names of certain objects, you will be able to detect what the object is.
In this way, when the standards for data processing and what you want to detect are clear, supervised learning is suitable and has a wide range of uses.
unsupervised learning
Unsupervised learning is a learning method in which data is given without correct answers, and the computer itself finds features and regularities and groups them.
Supervised learning is a method of learning by giving the correct answer to a problem and deriving the regularity of the correct answer, but in unsupervised learning, similar data is classified based on several aspects from past data. In other words, in unsupervised learning, grouping itself is learning because classification is performed from a state where there are no correct answers.
Since it can find patterns and correlations from large amounts of data, it is used for clustering, anomaly detection, etc.
reinforcement learning
Reinforcement learning is a learning method that ultimately focuses on maximizing results, learns from data and experience, and continues to optimize subsequent actions. In reinforcement learning, the goal is to maximize the number of points that can be considered as a correct answer and learn while determining what actions should be taken to maximize the number of points.
Here are some usage examples.
- Simulation of games such as Go and Shogi
- automatic elevator control
- Product recommendations for Internet shopping
Examples of machine learning usage around us
Machine learning is used in everything around us in our daily lives. Let’s take a closer look at two use cases, including the one mentioned above.
Inquiry response using chatbot
A chatbot is a program or application that automatically responds to input text or voice without human intervention. The name is a combination of the words “chat,” which is used to exchange short messages in real-time, and “bot,” which refers to a robot that automates certain tasks.
Many of you may have seen a chat screen on a company’s contact page. A computer analyzes and categorizes the questions and answers them on behalf of a human.
Chatbots are popular in call centers and customer support. By introducing a chatbot, you can expect to reduce the number of man-hours required to respond to customer inquiries and improve customer satisfaction, and it will also lead to a reduction in personnel costs by eliminating the need for specialized operators.
Chatbots are divided into two types: “scenario type”, in which expected questions and answers are registered in advance, and “machine learning type (AI type)”.
A feature of machine learning chatbots is that they can accumulate data and provide optimal answers using machine learning. While it is possible to expect natural answers similar to those of a human conversation, there is a concern that the accuracy will be low during the initial stage of operation and when little data has been accumulated. You will need to prepare training data before starting the operation.
Forecasting demand for services and products
Demand forecasting is the short-term and long-term prediction of sales for your company’s products and services. There are various methods for predicting demand, but in general, past trends and market trends are analyzed to predict future needs.
Until now, demand forecasting has been done manually, but by using machine learning to learn from vast amounts of past data, highly accurate demand forecasting is now possible. In addition, by replacing human hands, it is expected that operational efficiency will be improved.
Demand prediction using machine learning is spreading not only in the manufacturing and distribution industries but also in various fields and fields.
Machine Learning FAQ
Python is a programming language suitable for machine learning
When researching machine learning, you often come across the word “Python.” What is Python?
Python is a programming language. Python is used for various development purposes, but it is said to be suitable for the development of AI and machine learning. For example, Python is also used in the product recommendation function for internet shopping mentioned earlier.
Python is suitable for the following reasons:
- Rich library
- Already used by many companies and services
- Code is simple and easy to understand
- The operation can be easily confirmed
Python is an open-source language that is characterized by its ability to be written with simple and short code, and by the fact that it has a wide variety of programs (libraries) that are useful during development.
Machine learning requires extensive knowledge
What kind of knowledge is required to master machine learning?
Mastering machine learning requires a variety of knowledge. For example, you should be able to read programming languages such as Python mentioned above to some extent, have at least knowledge of high school mathematics, and have knowledge of algorithms.
Also, if you want to convey the knowledge you have acquired as an objective fact, it is a good idea to obtain a qualification. Qualifications related to machine learning include G-certification (for generalists), E-certification (for engineers), and the Python Engineer Certification Data Analysis Exam.
However, acquiring knowledge takes time. If you want to incorporate machine learning and AI into your business and achieve results, one way is to borrow the help of external experts.
summary
Machine learning is included in AI and is one of the technologies that make AI function. Specifically, it is a technology that analyzes data by giving sample data to a computer and having it learn repeatedly to discover underlying patterns and rules.
There are three learning methods in machine learning: supervised learning, unsupervised learning, and reinforcement learning, and they are used in a variety of fields.
Machine learning has been widely used in business in recent years, but in order to actually incorporate it into business and achieve results, deep knowledge of machine learning itself is required. Why not consider an implementation method that suits your company, while seeking the help of experts as needed?
