Unsupervised learning

As explained in 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.

Unsupervised learning trains models without human-provided labels by exploiting structure in the input data. Self-supervision creates implicit labels from the input, allowing familiar regression or classification losses to drive learning. section 7.0

Common self-supervised tasks include:

• Inpainting: reconstruct an obscured portion of an image. section 7.0
• Image sorting: restore the correct order of shuffled image pieces. section 7.0
• Autoencoding: reconstruct the original input from the input itself. section 7.0

Autoencoding digits

A larger PyTorch autoencoder expands the encoded representation, reconstructs 28×28 digit images, and trains with mean squared error. Better visual reconstruction does not necessarily prove that the representation is useful. section 7.3

Figure 7.1 Three different types of self-supervised problems: inpainting, image sorting, and autoencoding. In each case, we do not need to know what the image is of because the network will try to predict the original image no matter what. In the first case (red), we randomly block out a portion of the image and ask the network to fill in the missing piece. In the second case, we break the image into pieces, shuffle them, and ask the network to put them back in the correct order. In the last case, we simply ask the network to predict the original image from the original image.

Compare inpainting, image sorting, and autoencoding as three ways to generate targets directly from images. section 7.0

How Inside Deep Learning: Math, Algorithms, Models covers it

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 it assumes

Assumes familiarity with regression or classification losses while motivating label-free learning from zero.

What you can do after

Build an autoencoder that reconstructs inputs using PyTorch modules and mean squared error.

How it connects

Connects self-supervision with autoencoding, regression, classification, reconstruction, and learned representations.

liveBookmanning.com

Read Inside Deep Learning: Math, Algorithms, Models on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Data Without LabelsAn unsupervised deep learning algorithm trains a multilayer neural model to discover structure or representations without target labels.
• 

  liveBookmanning.com
  Grokking Machine LearningA machine-learning approach in which a model discovers structure or patterns from data without target outputs.
• 

  liveBookmanning.com
  Financial AI in PracticeUnsupervised learning is a machine-learning approach that identifies structure or patterns in data without relying on labeled outcomes.
• 

  liveBookmanning.com
  Optimization Algorithms: AI techniques for design, planning, and control problemsUnsupervised learning is a machine-learning paradigm in which a model discovers structure or patterns in data without labeled target outputs.
• 

  liveBookmanning.com
  Inside AIUnsupervised learning is a machine-learning technique in which an algorithm analyzes unlabeled data to discover patterns or groupings.
• 

  liveBookmanning.com
  Cyber Threat HuntingUnsupervised machine learning is a machine-learning paradigm that discovers patterns and structures in data without predefined labels or specific instructions.
• 

  liveBookmanning.com
  Statistics Slam DunkUnsupervised learning is a machine-learning approach that discovers structure or patterns in data without labeled target values.
• 

  liveBookmanning.com
  Investing for ProgrammersUnsupervised learning is a machine-learning approach that identifies structure or patterns in data without predefined output labels.
• 

  liveBookmanning.com
  Outlier Detection in PythonMachine learning that extracts structure, information, or patterns from data without labeled target values.
• 

  liveBookmanning.com
  The AI Pocket BookA machine learning approach that seeks patterns in unlabeled data without a target output.
• 

  liveBookmanning.com
  Machine Learning Algorithms in DepthUnsupervised learning is a machine learning paradigm that learns structure or representations from data without labeled target outputs.
• 

  liveBookmanning.com
  Natural Language Processing in Action, Second EditionUnsupervised learning is a machine-learning approach that derives patterns or representations directly from data without human-provided target labels.
• 

  liveBookmanning.com
  Grokking AI ApplicationsUnsupervised learning is a machine learning approach that discovers patterns or structure in data without labeled examples.
• 

  liveBookmanning.com
  Math and Architectures of Deep LearningA learning setup that uses training inputs without manually provided target outputs.
• 

  liveBookmanning.com
  Acing the CCNA Exam: Volume 2 Advanced Networking and SecurityUnsupervised learning is a machine-learning approach that discovers patterns or structure in data without predefined labels.
• 

  liveBookmanning.com
  Introduction to Generative AI, Second EditionUnsupervised learning is machine learning in which training data are provided without labels.
• 

  liveBookmanning.com
  The Art of AI Product DevelopmentUnsupervised learning is machine learning in which a model learns patterns from data without explicitly provided target labels.
• 

  liveBookmanning.com
  Advanced Algorithms and Data StructuresUnsupervised learning is machine learning that analyzes unlabeled data to discover structure or patterns without predefined categories.
• 

  liveBookmanning.com
  Software Testing with Generative AIA machine-learning approach that trains a model on unlabeled data to discover implicit patterns or structure.

Compare how 20 books explain 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.

This explainer was generated by Manning's knowledge graph from the text of Inside Deep Learning: Math, Algorithms, Models. The section links point to the parts of the book it draws on.