Handwritten zip code recognition with multilayer networks

By Yann LeCun, O. Matan, Bernhard E. Boser, John S. Denker, D. Henderson, Richard E. Howard, Wayne E. Hubbard, L. D. Jacket, Henry S. Baird. 1990

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Density Modeling of Images using a Generalized Normalization Transformation (2015, background): “…image into non-overlapping blocks (which produces artifacts at block boundaries) or simply increasing the size of the transformation, which would require a much larger number of parameters for the linear transform than a convolutional solution (which allows “weight sharing” (LeCun et al., 1990)).”
Rotation Equivariant Vector Field Networks (2016, background): “This fact is vital for the applicability of deep neural networks to images [18].”
A Survey of Unsupervised Deep Domain Adaptation (2018, background): “USPS [131] This is another handwritten digit dataset (digits 0–9).”
Adversarial Transfer Learning (2018, background): “CyCADA increases accuracy from 54% to 82% for a synthetic season adaptation dataset [53], from 90.1% to 96.5% on USPS→MNIST, and from 85.0% to 90.4% on SVHN→MNIST (Table 3) by incorporating pixel-level adaptation using a GAN to perform image-to-image translation in addition to feature-level adaptati”
The Invariance Hypothesis Implies Domain-Specific Regions in Visual Cortex (2014, unknown): “, [53, 58–60].”
Image and Video Compression With Neural Networks: A Review (2019, methodology): “With the introduction of parameter-sharing for MLP 1990 [33], a more light-weighted version of neural network called convolutional neural network was proposed and applied in the documents recognition, which makes the large scale neural network training possible.”
Structural Deep Clustering Network (2020, background): “• USPS[13]: The USPS dataset contains 9298 gray-scale handwritten digit images with size of 16x16 pixels.”
Comparison of classifier methods: a case study in handwritten digit recognition (1994, background): “The deficiencies of the linear classifier are well documented [3]and it is included here simply to form a basis of comparison for more sophisticated classifiers.”
Global training of document processing systems using graph transformer networks (1997, background): “They have had numerous applications in handwriting recognition [10, 2, 9], and object location in images, particulaly faces [13].”
Multi-Digit Recognition Using a Space Displacement Neural Network (1991, background): “In previous work (Le Cun et al., 1990) we have demonstrated a feed-forward backpropagation network that recognizes isolated handwritten digits at state-of-the-art performance levels.”