Unsupervised learning
Unsupervised learning is a machine-learning approach that discovers patterns, structure, or representations in data without labeled target outputs.
Data Without Labels
An unsupervised deep learning algorithm trains a multilayer neural model to discover structure or representations without target labels.
- How it is used
- Uses unlabeled input data to explore structure, segment observations, detect anomalies, and learn useful representations.
- How deep it goes
- Defines the label-free approach and develops clustering methods, while treating unsupervised neural training and autoencoders separately.
- Examples
- Illustrates grouping with shapes and data points, then uses customer segmentation, network anomalies, and cat-and-dog images.
- What you can do after
- Practice/perform customer segmentation and anomaly exploration by inspecting patterns found without target labels.
Read how Data Without Labels explains unsupervised learning →
Grokking Machine Learning
A machine-learning approach in which a model discovers structure or patterns from data without target outputs.
- How it is used
- Finds patterns in unlabeled data and can preprocess data before supervised learning.
- How deep it goes
- Defines the approach and outlines clustering, dimensionality reduction, and generative learning without detailing their algorithms.
- Examples
- Uses pet-image grouping, email clusters, housing-feature reduction, and generated faces, music, and poetry.
- What you can do after
- Understand how unlabeled data can reveal groups, simplify datasets, or support generation of similar data.
Read how Grokking Machine Learning explains unsupervised learning →
Financial AI in Practice
Unsupervised learning is a machine-learning approach that identifies structure or patterns in data without relying on labeled outcomes.
- How it is used
- Flags departures from predominantly legitimate financial activity to investigate potential novel fraud when labeled cases are scarce.
- How deep it goes
- Explains normal-behavior modeling and named anomaly methods, including reconstruction error, without specifying training settings or alert thresholds.
- Examples
- Illustrates suspicious transaction outliers, small transaction clusters, and anomalous inputs with high autoencoder reconstruction error.
- What you can do after
- Evaluate/understand it as an option for finding potential fraud patterns that labeled historical cases may not capture.
Read how Financial AI in Practice explains unsupervised learning →
Optimization Algorithms: AI techniques for design, planning, and control problems
Unsupervised learning is a machine-learning paradigm in which a model discovers structure or patterns in data without labeled target outputs.
- How it is used
- Uses unlabeled data to find structure and applies a self-organizing map as a TSP heuristic.
- How deep it goes
- Defines clustering, names several methods, and works through SOM training and tuning for TSP without deriving the other methods.
- Examples
- Illustrates unlabeled driving scenarios and compares a SOM-generated Qatar TSP route with a known optimum.
- What you can do after
- Apply it via a self-organizing map to generate and tune a TSP route, then compare its length with a known optimum.
Inside AI
Unsupervised learning is a machine-learning technique in which an algorithm analyzes unlabeled data to discover patterns or groupings.
- How it is used
- Finds patterns and groups in unlabeled data for decisions such as customer segmentation.
- How deep it goes
- Defines the approach and illustrates clustering through applications, without developing algorithms or theory.
- Examples
- Uses fruit grouping and retail customer segments, and identifies money laundering as an application.
- What you can do after
- Evaluate/understand when unlabeled data may reveal useful groups and why results still need quality checks.
Cyber Threat Hunting
Unsupervised machine learning is a machine-learning paradigm that discovers patterns and structures in data without predefined labels or specific instructions.
- How it is used
- Finds anomalous network behavior during threat hunts when no specific indicator, label, or fixed threshold is available.
- How deep it goes
- Defines unlabeled pattern discovery and walks through preprocessing, feature selection, k-means clustering, and cluster investigation while deferring deeper mathematics.
- Examples
- Uses network connection events and jittered command-and-control beaconing, including clusters containing suspicious internal-to-external connections.
- What you can do after
- Apply it via existing Python libraries to prepare connection data, cluster it, and investigate anomalous groups as threat-hunting leads.
Read how Cyber Threat Hunting explains unsupervised learning →
Statistics Slam Dunk
Unsupervised learning is a machine-learning approach that discovers structure or patterns in data without labeled target values.
- How it is used
- Finds previously unknown patterns, structures, or relationships when analysis has input variables but no target outcome.
- How deep it goes
- Defines the approach and illustrates clustering, but defers detailed algorithm implementation to K-means and hierarchical-clustering analyses.
- Examples
- Uses NBA draft segmentation, tanking analysis, K-means customer segmentation, and hierarchical clustering of draft-pick career expectations.
- What you can do after
- Apply it via R to construct clusters, compare candidate cluster counts, visualize results, and interpret actionable groupings.
Read how Statistics Slam Dunk explains unsupervised learning →
Investing for Programmers
Unsupervised learning is a machine-learning approach that identifies structure or patterns in data without predefined output labels.
- How it is used
- Uses unlabeled financial data to group assets, expose outliers, and support exploratory investment research.
- How deep it goes
- Defines structure discovery and demonstrates feature calculation, elbow selection, K-means clustering, plotting, and cautious interpretation.
- Examples
- Calculates stock returns and volatility, selects five clusters, plots S&P 500 assets, and examines Moderna, Palantir, and Super Micro Computer.
- What you can do after
- Apply K-means with pandas, yfinance, and interactive plots to categorize assets and identify candidates for further investigation.
Read how Investing for Programmers explains unsupervised learning →
Outlier Detection in Python
Machine learning that extracts structure, information, or patterns from data without labeled target values.
- How it is used
- Frames pattern extraction from unlabeled data, especially multivariate outlier detection across diverse data types.
- How deep it goes
- Defines the learning setup and surveys method families, but stops short of detailed algorithm derivations or implementation procedures.
- Examples
- Uses financial fraud, maintenance-related sensor readings, astronomical observations, images, text, video, audio, and network data.
- What you can do after
- Evaluate/understand unlabeled data through outlier-detection methods while interpreting flagged records cautiously.
Read how Outlier Detection in Python explains unsupervised learning →
The AI Pocket Book
A machine learning approach that seeks patterns in unlabeled data without a target output.
- How it is used
- Explores unlabeled data to discover patterns when no single correct output exists.
- How deep it goes
- Explains clustering and evaluation ambiguity, but stops short of specifying an algorithm or formal success measure.
- Examples
- Uses patient-record clusters and the curve’s knee or elbow as concrete illustrations.
- What you can do after
- Evaluate/understand pattern discovery and recognize why unsupervised models lack a definitive correctness test.
Read how The AI Pocket Book explains unsupervised learning →
Machine Learning Algorithms in Depth
Unsupervised learning is a machine learning paradigm that learns structure or representations from data without labeled target outputs.
- How it is used
- Uses unlabeled observations to discover patterns and learn representations, including clusters, density distributions, and lower-dimensional projections.
- How deep it goes
- Explains several algorithm families and Expectation-Maximization steps, then develops a VAE anomaly detector with an objective and implementation.
- Examples
- Uses customer segmentation, image features, projected clusters, and time-series anomaly detection with an LSTM-VAE.
- What you can do after
- Build it from scratch using Keras/TensorFlow to train an LSTM-VAE and flag likelihood drops in sensor data.
Read how Machine Learning Algorithms in Depth explains unsupervised learning →
Natural Language Processing in Action, Second Edition
Unsupervised learning is a machine-learning approach that derives patterns or representations directly from data without human-provided target labels.
- How it is used
- Uses raw, unlabeled text to learn word representations for natural-language applications.
- How deep it goes
- Explains dataset-generated labels, contextual prediction, vector representations, CBOW, and skip-gram training.
- Examples
- Uses corpora, nearby-word prediction, Marie Curie associations, and a learned nessvector as concrete examples.
- What you can do after
- Evaluate learned word representations and understand how raw text can support embedding training without manual semantic labels.
Read how Natural Language Processing in Action, Second Edition explains unsupervised learning →
Grokking AI Applications
Unsupervised learning is a machine learning approach that discovers patterns or structure in data without labeled examples.
- How it is used
- Explains how language models learn patterns before receiving conversational refinement.
- How deep it goes
- Defines pattern discovery and illustrates document reading and next-word prediction without detailing algorithm internals.
- Examples
- Uses grouped toy shapes, document collections, grammar learning, and next-word prediction as concrete examples.
- What you can do after
- Understand how GPT acquires language patterns before supervised and reinforcement learning improve conversation.
Read how Grokking AI Applications explains unsupervised learning →
Math and Architectures of Deep Learning
A learning setup that uses training inputs without manually provided target outputs.
- How it is used
- Uses unlabeled inputs to discover clusters, estimate distributions, generate samples, or learn representations.
- How deep it goes
- Explains the setup, GMM posterior assignment, density estimation, and autoencoder training, with PyTorch implementation details for reconstruction.
- Examples
- Uses GMM class assignment, cat-image density modeling, and convolutional autoencoders trained by reconstruction loss.
- What you can do after
- Build autoencoder pipelines in PyTorch and evaluate unlabeled inputs through mixture-model posteriors or learned densities.
Read how Math and Architectures of Deep Learning explains unsupervised learning →
Acing the CCNA Exam: Volume 2 Advanced Networking and Security
Unsupervised learning is a machine-learning approach that discovers patterns or structure in data without predefined labels.
- How it is used
- Uses unlabeled data to discover structure, patterns, relationships, and clusters for machine-learning and network-data analysis.
- How deep it goes
- Explains unlabeled training, pattern discovery, clustering, and human interpretation, but does not detail specific algorithms or implementation steps.
- Examples
- Groups unlabeled cat-and-dog images into clusters before a human assigns interpretations.
- What you can do after
- Evaluate how unlabeled data can reveal clusters, patterns, anomalies, and other network insights.
Introduction to Generative AI, Second Edition
Unsupervised learning is machine learning in which training data are provided without labels.
- How it is used
- Frames unsupervised learning as one of three broad strategies for training machine-learning systems.
- How deep it goes
- Defines unlabeled training and pattern discovery, but stops short of explaining algorithms or implementation.
- Examples
- Uses underlying data patterns and clusters as the principal examples of unsupervised learning.
- What you can do after
- Evaluate whether a learning setup uses unlabeled data to discover patterns rather than predict labeled outputs or optimize rewards.
Read how Introduction to Generative AI, Second Edition explains unsupervised learning →
Inside Deep Learning: Math, Algorithms, Models
A learning setting in which models are trained without human-provided labels, often by exploiting structure in the data.
- How it is used
- Frames label-free training as a way to learn from abundant data when manually labeling examples is expensive.
- How deep it goes
- Defines implicit labels and demonstrates reconstruction, but leaves the usefulness of learned representations difficult to evaluate.
- Examples
- Uses image inpainting, shuffled-image sorting, autoencoding, and reconstructed handwritten digits as concrete examples.
- What you can do after
- Build an autoencoder that reconstructs inputs using PyTorch modules and mean squared error.
Read how Inside Deep Learning: Math, Algorithms, Models explains unsupervised learning →
The Art of AI Product Development
Unsupervised learning is machine learning in which a model learns patterns from data without explicitly provided target labels.
- How it is used
- Use unsupervised learning to explore poorly understood data and uncover structures before making product decisions.
- How deep it goes
- Defines unlabeled data and clustering, explains exploratory use, and stops short of implementation details or algorithmic mechanics.
- Examples
- Uses clustering of behavioral and demographic data to uncover user segments for personalization, marketing, and feature decisions.
- What you can do after
- Apply it via clustering to structure an unfamiliar dataset and inform segmentation, personalization, marketing, or feature development.
Read how The Art of AI Product Development explains unsupervised learning →
Advanced Algorithms and Data Structures
Unsupervised learning is machine learning that analyzes unlabeled data to discover structure or patterns without predefined categories.
- How it is used
- Uses unlabeled data to discover structure for segmentation, noise detection, and preprocessing.
- How deep it goes
- Explains the objective through clustering, then compares algorithms, evaluation metrics, and hyper-parameter selection.
- Examples
- Uses customer segmentation, signal-noise detection, unusual-point discovery, and grouping people by demographic and financial attributes.
- What you can do after
- Apply it via clustering to group unlabeled points, inspect discovered categories, detect noise, and prepare data for later algorithms.
Read how Advanced Algorithms and Data Structures explains unsupervised learning →
Software Testing with Generative AI
A machine-learning approach that trains a model on unlabeled data to discover implicit patterns or structure.
- How it is used
- Describes how predictive language models learn from large collections of unlabeled data.
- How deep it goes
- Defines unlabeled-data training and implicit pattern learning, then sketches datasets, algorithms, probabilities, and model weights.
- Examples
- Uses keyboard next-word prediction and contrasts it with labeled medical data paired with illness outcomes.
- What you can do after
- Evaluate/understand how unlabeled training produces predictive probabilities and implicit patterns.
Read how Software Testing with Generative AI explains unsupervised learning →
Related concepts
- supervised learningcontrasts with · Supervised learning is a machine-learning approach that learns to map inputs to target outputs from examples paired with known labels or values.
- K-means clusteringincludes · K-means clustering partitions data into a specified number of groups by assigning observations to representative centers and iteratively updating those centers to reduce within-group distances.
- density estimatorincludes · A density estimator is a statistical method or model that estimates the probability density underlying observed data.
- hierarchical clusteringincludes · Hierarchical clustering is a clustering method that organizes observations into nested groups by successively merging or splitting clusters.
- machine learningkind of · Machine learning is a field of computational methods in which systems learn patterns or relationships from data to make predictions, decisions, or other outputs.
- clusteringincludes · Clustering is the technique of grouping data items according to similarity or shared characteristics.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.