MIT study shows neural networks can self-organize their own learning machinery
Overview
MIT researchers Qianli Liao, Liu Ziyin, Yulu Gan, Tomaso Poggio and Mark Harnett report in PNAS that artificial neural networks can build their own learning machinery from simple local rules.
Randomly connected units, following a "neurons that fire together wire together" rule for how connections change, self-organized into motifs that classified images. The team says this performed at a level similar to networks trained with back-propagation.
The authors argue that self-organization is a biologically plausible measure of a learning model, since the brain has no external designer.
Written by AI from the articles below · updated Oct 13, 5:14 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- MIT News · AIOfficialArtificial neural networks can organize their own learning machinery, MIT study finds
AIMIT researchers report in PNAS that artificial neural networks can build their own learning machinery when given simple rules for how connections change. Randomly connected units with a "neurons that fire together wire together" rule self-organized into motifs that classified images, performing at a level similar to networks trained with back-propagation. The authors argue that self-organization is a biologically plausible measure of a learning model, since the brain has no external designer.
Heat trend
Not enough continuous observations to show a trend yet.