Unsupervised learning
Unsupervised learning is a learning method that allows you to find patterns in the data on your own without a correct answer.
Unsupervised learning is a learning method that provides only data without correct labels and allows AI to discover hidden structures and patterns on its own. Just as thousands of fruits that we have never seen before can be grouped together even if they are lined up without name tags with similar colors and shapes, AI groups data into similar groups or finds unique ones.
This method has become important because most of the data in the world does not have correct labels, and labeling them costs a lot of money. It is used in customer type classification, abnormal transaction detection, recommendation systems, etc., and LLM pre-learning, which is learned from large amounts of text without labels, is also based on self-supervised learning, which is a similar type, so its significance has increased.
However, since there is no right answer, it is difficult to evaluate whether the results are correct, and there is a limitation that a human must ultimately interpret what the group found by AI means.
✅ Why it matters
- A large amount of data can be utilized without labeling costs
- Discovers patterns and groups that people were unaware of
- Wide range of practical applications, including customer analysis and anomaly detection
⚠️ Limits and debates
- There is no correct answer, so it is difficult to evaluate whether the results are right or wrong
- Ultimately, it is up to humans to interpret the meaning of discovered patterns
- In prediction problems with a clear purpose, accuracy is lower than supervised learning.