Active learning
Active learning is a machine-learning approach in which a learner selects informative data examples for labeling or feedback to improve a model.
Build AI Drug Discovery Pipelines
Active learning is a machine-learning approach that iteratively selects the most informative unlabeled examples for labeling to improve a model with limited labeling resources.
- How it is used
- Prioritizes expensive docking or free energy evaluations and feeds experimental results back into drug-discovery models.
- How deep it goes
- Builds a docking loop with initialization, acquisition, oracle calls, retraining, and metrics; extends the pattern to free energy calculations in pseudocode.
- Examples
- Demonstrates VEGFR-2 docking with a precomputed-score oracle and illustrates sample selection through an apple-pie analogy.
- What you can do after
- Apply it via PyTorch and RDKit to select and evaluate batches in a docking-screening workflow.
Read how Build AI Drug Discovery Pipelines explains active learning →
Human-in-the-Loop Machine Learning: Active learning and annotation for human-centered AI
Active learning is a machine learning approach in which a learning system selects data for annotation to improve its learning process.
- How it is used
- Targets limited human annotation toward data likely to improve a supervised model quickly or efficiently.
- How deep it goes
- Explains an iterative labeling and retraining workflow, compares sampling strategies, and discusses evaluation and iteration tradeoffs.
- Examples
- Uses disaster-related message classification and autonomous-driving video annotation to illustrate uncertainty and annotation scale.
- What you can do after
- Build it from scratch as a simple labeling and retraining cycle using confidence and outlier sampling.
Machine Learning Algorithms in Depth
Active learning is a supervised learning technique in which a learner selects unlabeled instances for labeling to improve performance with fewer labeled examples.
- How it is used
- Selects unlabeled training instances for oracle labeling when labels are costly and fewer labeled examples are desired.
- How deep it goes
- Develops pool-based selection, compares query criteria, and implements a ranking class; flags full-pool ranking as costly at scale.
- Examples
- Compares 30 random labels with 30 entropy-selected queries on synthetic binary data, yielding 90% and 92.5% held-out accuracy.
- What you can do after
- Build it from scratch as a candidate-ranking class and apply it to logistic regression queries.
Read how Machine Learning Algorithms in Depth explains active learning →
AI-Powered Search
Active learning is a machine learning strategy in which the learner selects informative examples for labeling or feedback.
- How it is used
- Improves learning-to-rank systems by expanding feedback beyond results already favored by the deployed model.
- How deep it goes
- Explains presentation bias, exploration versus exploitation, Gaussian-process guidance, retraining, and A/B evaluation.
- Examples
- Uses unexplored search results, the “dryer” query, and an updated “transformers dvd” ranking experiment.
- What you can do after
- Apply it via an automated LTR pipeline that explores candidates, gathers signals, rebuilds judgments, retrains models, and reruns tests.
Machine Learning System Design
Active learning is a machine-learning technique that selectively chooses informative or uncertain examples for labeling or inclusion in a dataset.
- How it is used
- Guides dataset extension and labeling toward informative, uncertain, or error-prone examples rather than simply increasing volume.
- How deep it goes
- Defines error- and confidence-based selection, but defers detailed comparison of the many active-learning methods.
- Examples
- Uses PhotoStock query–image relevance pairs, model confidence, and model errors as concrete selection signals.
- What you can do after
- Apply active learning to prioritize human labeling and extend datasets with more informative examples.
Read how Machine Learning System Design explains active learning →
LLM Evaluation and Alignment, The Foundational Ideas
Active querying is a data-collection technique that selects examples for labeling according to their expected information value.
- How it is used
- Uses ensemble disagreement to reduce labeling budgets during preference collection for reinforcement learning from human feedback.
- How deep it goes
- Explains ensemble-based selection and its rationale, while noting that modern pipelines often favor uniform sampling and throughput.
- Examples
- Uses simulated MuJoCo backflips, Atari games, trajectory pairs, and three disagreeing reward models as examples.
- What you can do after
- Apply it to prioritize informative preference queries when labeling budgets are small.
Read how LLM Evaluation and Alignment, The Foundational Ideas explains active learning →
Also covered in
- Bayesian Optimization in Action
- Grokking BayesIn early access (MEAP)
- Outlier Detection in Python
- Generative AI in Action
- AI for Everyday IT
What to know first
- decision boundaryA decision boundary is a boundary in input or feature space separating regions assigned different classes or decision outcomes by a predictive model.
- machine learningMachine learning is a field of computational methods in which systems learn patterns or relationships from data to make predictions, decisions, or other outputs.
- supervised learningSupervised learning is a machine-learning approach that learns predictive relationships from examples paired with known target outputs or labels.
- AI modelAn AI model is a computational system that learns or encodes patterns to map inputs to outputs such as predictions, classifications, or generated content.
Related concepts
- Intersection over Unionused by · Intersection over Union is the ratio of the intersection area or volume of two regions to the area or volume of their union.
- uncertainty samplingincludes · Uncertainty sampling is an active-learning strategy that selects unlabeled instances about which a model is most uncertain.
- acquisition functionuses · An acquisition function is a scoring rule that ranks candidate evaluations by their expected usefulness for selecting the next sample or experiment.
- labelerused by · A labeler is a person or process that assigns labels, ratings, judgments, or other structured annotations to data or model outputs.
- diversity-based samplingincludes · Diversity-based sampling selects examples to represent variation or cover underrepresented regions of a candidate set or feature space.
Questions about active learning
What is active learning?
Active learning is a machine-learning approach in which a learner selects informative data examples for labeling or feedback to improve a model.
Which Manning books explain active learning?
Build AI Drug Discovery Pipelines, Human-in-the-Loop Machine Learning: Active learning and annotation for human-centered AI, Machine Learning Algorithms in Depth, AI-Powered Search, Machine Learning System Design and LLM Evaluation and Alignment, The Foundational Ideas.
What should I know before learning active learning?
Manning books that teach active learning usually introduce decision boundary, machine learning, supervised learning and AI model first.
What kinds of active learning are there?
Kinds of active learning include uncertainty sampling and diversity-based sampling.
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.





