liveBook · Manning

Local learning rule

As explained in Machines that Think

A local learning rule adjusts a model component using information available at that component or its immediate connections rather than a globally propagated error.

A local learning rule adjusts a model component using information available at that component or its connections, rather than relying on an error signal propagated globally. This makes local rules relevant when feedback is sparse, delayed, or noisy. section 5.3.5

In neuromorphic computing, local learning rules are part of an approach to efficient, parallel, event-driven hardware. Limiting learning to locally available information can suit constrained systems, though it may also limit flexibility and prioritize stability or efficiency. section 5.3.4

How Machines that Think covers it

How deep it goes
Explains the motivation and constraints but stops short of specifying an update algorithm or mathematical mechanism.
What you can do after
Understand why local updates may suit sparse feedback and constrained hardware, while recognizing potential limits on flexibility.
How it connects
Connects local learning rules to neuromorphic computing, spiking neurons, asynchronous communication, continual learning, and credit assignment.

This explainer was generated by Manning's knowledge graph from the text of Machines that Think. The section links point to the parts of the book it draws on.