M37.1 CONNECT THE MECHANISM
Update connections from local activity
A single neuron, no labels, no teacher, just coincidences. Give it one small fix and it rediscovers PCA on its own. Without the fix, its weights explode.
LESSON OVERVIEW14 min lesson
Lesson overview
A single neuron, no labels, no teacher, just coincidences. Give it one small fix and it rediscovers PCA on its own. Without the fix, its weights explode.
What you’ll explore
- Hebbian rules change connections using pre- and postsynaptic activity; normalization, competition, and additional signals are needed to manage stability and connect local learning to useful tasks.
GO TO THE SOURCE
Original explanations, connected to the research.
A simplified neuron model as a principal component analyzer (Oja, 1982)Neuronal Dynamics, chapter 19: synaptic plasticity and learning (Gerstner, Kistler, Naud & Paninski, 2014)Random synaptic feedback weights support error backpropagation for deep learning (Lillicrap et al., 2016)Suggest a correction
A precise note can make an explanation better.
Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.