Unsupervised Machine Learning

Unsupervised Machine Learning is a sophisticated approach where the system learns from a dataset without predefined labels.

This method involves feeding a dataset into the machine. The unsupervised ML algorithm scrutinizes the data, seeking patterns or criteria for classification, effectively identifying the labels (classes) for each example on its own.

Operational Mechanics

A dataset, known as the training set, is introduced into the system. Each example within this set is characterized by key features, yet remains uncategorized by its creator.

In unsupervised ML, the system operates without any guiding examples ( golden data ), and it is unaware of the specific classes or categories for classifying the inputs.

Example: Consider a training set containing data on four flower species, such as length and petal width. The catch is, the dataset does not reveal the species of each flower.

The algorithm independently formulates a classification model, deriving it from the shared characteristics, similarities, and discernible patterns found within the training set data.

Key categories of unsupervised machine learning encompass:

• Clustering
  • DBSCAN
  • K-means
  • Hidden Markov Model
  • Agglomerative
  • Fuzzy C-Means
  • Mean shift
• Anomaly Detection
  • k-Nearest Neighbours
  • Bayesian Belief Network
  • Decision Tree
  • Support Vector Machine
• Association
  • Bayesian Belief Network
  • Decision Tree
  • Neural Networks

Additionally, the field includes various pattern search algorithms (Apriori, Eclat, FP-Growth) and a range of general algorithms (T-Sne, PCA, LSA, SVD, LDA).