Machine learning is the process of having a machine learn from data so that it can deal with various problems. It is attracting attention as one of the technologies responsible for the “learning” of AI.
Machine learning methods can be broadly classified into three categories: supervised learning, unsupervised learning, and reinforcement learning.
This article explains the meaning of machine learning, the types of learning methods, and differences from deep learning in an easy-to-understand manner.
What is machine learning?
First, let’s check what machine learning means, what it can do, and how it differs from AI and deep learning.
Meaning of machine learning
Machine learning is “learning” in AI. It has the meaning of “the machine itself learns” as a human learns.
In other words, the goal of machine learning is to enable a trained machine to do more than what the programmer programmed it to do.
Machine learning is a technology that supports AI, and deep learning is one of the machine learning methods.
Why machine learning is gaining attention
Machine learning is also closely related to fields such as AI and big data, which are attracting attention in the promotion of DX.
Machine learning is characterized by processing vast amounts of information and discovering features and rules from the data . You will be able to make predictions and judgments based on the derived characteristics and regularities.
In other words, machine learning is used to give AI the ability to learn and to process and analyze big data with large amounts of complex data .
Explain three learning methods of machine learning
① Supervised learning
“Supervised learning” to learn by giving a teacher signal
Supervised learning is to make artificial intelligence learn in one direction by giving a set of example questions and model answers (teacher signal) . Generally, a large amount of data is required, and the neural network itself judges whether the output result is correct or not based on the given data.
Although it is possible to make judgments and take actions based on examples even for cases that have not been learned, it has the disadvantage that it cannot respond to unknown events that humans cannot give knowledge in advance. In addition, there is an ability limit that “it will not be smarter than the person who gave the model answer”.
It is used for applications such as sales prediction using “regression” to derive trends (functions) based on past data and predicting future values, and image classification using “classification” to automatically classify unknown data. increase.
② Unsupervised learning
“Unsupervised learning” in which AI (artificial intelligence) learns by itself based on its own activities
Unsupervised learning does not require model answers, and AI accumulates data based on its own activities and learns by itself.
A consistent environment must be assumed, and events that cannot be simulated cannot be learned.
It is used for applications such as recommendation and customer segmentation that utilizes “clustering” that analyzes accumulated data and extracts and groups similar data.
③ Reinforcement learning
Reinforcement learning, in which AI finds the optimal action through trial and error
Reinforcement learning is a learning method in which AI repeats trial and error in its own environment to find the optimal behavior and value . In terms of AI recognizing and analyzing the results of its own actions, it can be regarded as unsupervised learning.
An important factor in reinforcement learning is to make AI firmly aware of its own actions and situations. Then, the evaluation value for the result under the environment is used as a “reward” and used as a clue for learning.
For example, let’s say you gave AI an environment to play a game. Due to the lack of a teacher, the AI does not show its strength at first, but the AI itself considers, “What can I do to get more rewards?” As you battle, you will accumulate data and become stronger.
In this way, reinforcement learning has a wide range of applications, and is very effective when the target to be learned cannot be modeled.
Difference between machine learning and deep learning
Deep learning is one of the methods of machine learning . As long as there is sufficient training data, it is possible to automatically extract data features using a neural network.
Deep learning has made it possible to learn from unstructured data (images, natural language, sounds) that were previously difficult to digitize.
In addition, the increased variety of digitalization has made it possible to generate natural language and detect anomalies, improving the accuracy of optimization and recommendations.
Python, a programming language used in AI (artificial intelligence) development
Python is the standard programming language used in AI development.
Among the many programming languages, the main reason is that the code is easy to handle, and it is suitable for processing the big data required for machine learning. In addition, it is useful because it is originally easy to execute scientific calculations and has a library for machine learning.
Compared to other languages, it is easier to learn even for programming beginners, and it is attracting attention along with the AI boom.
Example of work efficiency improvement using AI image recognition (Unimate)
An app that automatically measures uniforms using image recognition
Unimate Co., Ltd. is a company that develops rental uniform business.
The company’s problem was the frequent occurrence of erroneous orders due to measurement errors. In response to this issue, Monster Lab has developed an automatic measurement application “AI x R Tailor” that utilizes AI image recognition .
During the development, we successfully developed an original AI engine by deriving a method of “creating a 3D model from an image and predicting the actual size from it” through technical research. We repeated verification using the measurement data provided by Unimate to improve the accuracy of AI image recognition.
By using AI image recognition in the measuring process, we succeeded in reducing costs caused by human errors such as incorrect measurements .
Summary: Machine learning is one of the technologies included in AI
Machine learning is to let a machine learn and find features and rules from a huge amount of data. It is one of the technologies included in AI and is used in various services and products.
AI uses machine learning to learn from huge amounts of data and find rules, which is effective in improving the efficiency of simple tasks and reducing human error .
