Meta-learning system

As explained in Deep Learning with R, Third Edition

A meta-learning system is a machine-learning system that learns strategies or reusable structures for solving multiple learning tasks.

A meta-learning system learns from many tasks so it can assemble a suitable model for a new task with very little data. section 19.4.7

Its long-term design combines reusable structures with search and abstraction:

• Stores recurring problem-solving patterns as reusable subroutines, functions, or classes in a global library. section 19.4.6 section 19.4.7
• Combines algorithmic blocks with geometric modules such as deep-learning components and pretrained representations. section 19.4.6 section 19.4.7
• Uses discrete program synthesis, guided by learned intuition, to search over combinations of reusable blocks. section 19.4.7

Reusable software components

An engineer who packages HTTP queries in R as a reusable library illustrates the intended pattern: solve a problem once, then adapt the stored solution to related tasks. section 19.4.6

Figure 19.16 A meta-learner capable of quickly developing task-specific models using reusable primitives (both algorithmic and geometric), thus achieving extreme generalization

Shows a meta-learner assembling task-specific models from reusable algorithmic and geometric primitives. section 19.4.6

How Deep Learning with R, Third Edition covers it

How deep it goes

Explains reusable libraries, component recombination, and program-synthesis search, but stops short of implementation details.

Examples

Uses HTTP queries in R, AlphaGo, self-driving cars, and a reusable-primitives architecture as illustrative examples.

What you can do after

Evaluate/understand how reusable components and learned search could support rapid adaptation across many tasks.

How it connects

Connects program synthesis, deep learning, pretrained representations, lifelong learning, and artificial general intelligence.

liveBookmanning.com

Read Deep Learning with R, Third Edition on liveBook Get the book at manning.com

Related concepts

• functionproduces · A function is a callable program unit that performs an operation, may accept inputs, and can produce a result or side effect.
• libraryuses · A software library is a reusable collection of code or components that other programs can incorporate or invoke to provide functionality.
• meta-learningimplements · Meta-learning is a machine-learning approach that uses experience across tasks to improve how systems learn or adapt to new tasks.

This explainer was generated by Manning's knowledge graph from the text of Deep Learning with R, Third Edition. The section links point to the parts of the book it draws on.