Stacked What-Where Auto-encoders

8 Jun 2015Junbo ZhaoMichael MathieuRoss GoroshinYann LeCun

We present a novel architecture, the "stacked what-where auto-encoders" (SWWAE), which integrates discriminative and generative pathways and provides a unified approach to supervised, semi-supervised and unsupervised learning without relying on sampling during training. An instantiation of SWWAE uses a convolutional net (Convnet) (LeCun et al. (1998)) to encode the input, and employs a deconvolutional net (Deconvnet) (Zeiler et al. (2010)) to produce the reconstruction... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT LEADERBOARD
Image Classification CIFAR-10 SWWAE Percentage correct 92.2 # 56
Image Classification CIFAR-100 SWWAE Percentage correct 69.1 # 50
Image Classification MNIST Zhao et al. (2015) (auto-encoder) Percentage error 4.76 # 26
Image Classification STL-10 SWWAE Percentage correct 74.3 # 14
Semi-Supervised Image Classification STL-10, 1000 Labels SWWAE Accuracy 74.30 # 6