Training Neural Networks with Local Error Signals

Open paper
Benchmark Model Rank Results
image-classification-on-cifar-10VGG11B(2x) + LocalLearning + CO#112Percentage correct: 96.4
image-classification-on-cifar-100VGG11B(3x) + LocalLearning#128Percentage correct: 79.9
image-classification-on-fashion-mnistVGG8B(2x) + LocalLearning + CO#4Percentage error: 4.14
image-classification-on-kuzushiji-mnistVGG8B(2x) + LocalLearning + CO#4Accuracy: 99.01Error: 0.99
image-classification-on-mnistVGG8B + LocalLearning + CO#12Percentage error: 0.26
image-classification-on-stl-10VGG8B + LocalLearning + CO#43Percentage correct: 80.75
image-classification-on-svhnVGG8B + LocalLearning + CO#16Percentage error: 1.65

Rank counts only results with a code link.