Pattern recognition identifies recurring structure or regularity in data, helping machine learning interpret complex inputs such as images, text, audio, video, and sensor datasets. section 1.3
Without a target variable, unsupervised learning discovers patterns from input data alone, revealing structure for exploration, segmentation, and anomaly detection. section 1.5.2
Practical uses include:
- Grouping customers by attributes such as revenue, visit frequency, recency, age, and demographics to support targeted segmentation. section 1.5.2
- Finding unusual network activity by identifying outliers in historical data. section 1.5.2
- Examining medical image data for underlying factors that may support differentiated treatment. section 1.5.2
Figure 1.16 An unsupervised learning algorithm finds patterns in the data on the left and results in clusters on the right.
Compare raw data points with the clusters produced when an unsupervised algorithm discovers structure. section 1.5.2
