Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree

Benchmark Model Rank Results
image-classification-on-cifar-10Tree+Max-Avg pooling#159Percentage correct: 94.0
image-classification-on-cifar-100Tree+Max-Avg pooling#168Percentage correct: 67.6
image-classification-on-mnistTree+Max-Avg pooling#18Percentage error: 0.3
image-classification-on-svhnTree+Max-Avg pooling#17Percentage error: 1.7