🔸Machine Learning can be divided into three main categories:
1. Supervised Machine Learning
2. Unsupervised Machine Learning
3. Reinforcement Learning
In this article, we will use examples of distinguishing between mice and cats to explain the three types of machine learning modes.
🔸Supervised Machine Learning: Supervised Machine Learning use labeled datasets to train models. First, we need to label the data with our target prediction(mice/cats). Machines are trained on labeled datasets and are able to predict the outcomes based on the training data. (Due to the model learning from labeled data, it can classify data or predict outcomes accurately.)
【Challenges】: Spending a lot of time on labeling data manually.
【Application】: Give the model a large amount of labeled data about 'mice and cats', the model will learn their characteristics and distinguish them.
【Other Applications】: Stock forecast, NLP.
🔸Unsupervised Machine Learning: The machine is trained by using unlabeled datasets, and it will sort the data into groups based on their similarities.
【Challenges】: Humans have no idea what the classification of Machine Learning is, so it's all interpreted by humans. Additionally, the results may not meet our expectations.
【Application】: Give the model the unlabeled data about ‘mice and cats’ and the model will sort them according to their characteristics.
【Other Applications】: Data classification.
🔸Reinforcement Machine Learning: Reinforcement learning is learning based on positive and negative feedback given by the user.
【Challenges】: The information provided by humans will have a major impact on model identification.
【Application】: The user gives information about ‘mice or cats’ while the model gives the answer; the user gives positive and negative feedback again, and the model will be adjusted for feedback.
【Other Applications】: Parking system in automatic driving.
🔸Tagtoo's products also incorporate a Machine Learning approach!
- Advertising products: Label Target Audiences Solution
- Machine Learning methods: Supervised Machine Learning product
- Functions: We hope to predict buying behavior by using the behavior data. Enables advertisers to use those audiences in advertising.
- Operation mode: We found that there is a common pattern in the consumer buying process. We therefore compiled those characteristics of purchasing behavior to predict the future behavior of consumers. Tagtoo consistently optimizes advertising effectiveness through data. We provide the best online ad experience for both advertisers and audiences.
If you have any advertising requirements, we are glad to have you contact us!
Tagtoo’s official website: Tagtoo Technology - Cross-platform Online Advertising
Email📩: tagtoo@tagtoo.org
References:
三大類機器學習：監督式、強化式、非監督式 – 工程師。日常 (wordpress.com)
技術文章-強化學習(Reinforcement Learning)：入門指南--MATLAB, Simulink, Rational 專業技術服務盡在鈦思科技 (terasoft.com.tw)