A different convolutional neural network design was proposed in 1988 by Daniel Graupe, Ruey Wen Liu and George S Moschytz.[15] for applications to decomposition of one-dimensional EMG signals, This design was further modified in 1989 to other convolution-based designs by Daniel Graupe, Boris Vern, Greg Gruener, Aaron S. Field and Qiu Huang in [16] and by Qiu Huang, Daniel Graupe and Yih Fang Huang in.[17]
With the rise of efficient GPU computing, it has become possible to train larger networks. In 2006 several publications described more efficient ways to train convolutional neural networks with more layers.[18][19][20] In 2011, they were refined by Dan Ciresan et al. and were implemented on a GPU with impressive performance results.[5] In 2012, Dan Ciresan et al. significantly improved upon the best performance in the literature for multiple image databases, including the MNIST database, the NORB database, the HWDB1.0 dataset (Chinese characters), the CIFAR10 dataset (dataset of 60000 32x32 labeled RGB images),[7] and the ImageNet dataset.[21]