Learning discriminative features within forward-Forward algorithm using convolutional prototype

doi:10.1016/j.patcog.2026.113139

Learning discriminative features within forward-Forward algorithm using convolutional prototype

2026-01-24

0

PRE

AI

DOI：10.1016/j.patcog.2026.113139

Original

Original request for help

Save

Abstract

En 中文

• We introduce prototype learning in forward-forward (FF) algorithm for the first time. • We propose PLFF to improve properties of features learned by FF algorithm. • We evaluate FF methods on long-tailed datasets for the first time. • Our PLFF achieves state-of-the-art performance across various public datasets.

Learning discriminative features within forward-Forward algorithm using convolutional prototype

Abstract

Journal

Organization