Handwritten digit recognition: applications of neural network chips and automatic learning

By Yann LeCun, L. Jackel, Bernhard E. Boser, John S. Denker, Hans Peter Graf, Isabelle M Guyon, D. Henderson, Richard E. Howard, Wayne E. Hubbard. 1989

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

AI2: Safety and Robustness Certification of Neural Networks with Abstract Interpretation (2018, methodology): “For the convolutional networks, we used the LeNet architecture [26], which consists of the following sequence of layers: 2 convolutional, 1 max pooling, 2 convolutional, 1 max pooling, and 3 fully connected layers.”
Disentangling Factors of Variation via Generative Entangling (2012, background): “gineering feature sets that are invariant to common sources of variation in data. This is the motivation behind both the inclusion of feature pooling stages in the convolutional network architecture (LeCun et al., 1989) and the recent trend toward representations based on large scale pooling of low-”
Efficient Processing of Deep Neural Networks: A Tutorial and Survey (2017, methodology): “ts the accuracy if only a single crop was used (i.e., the DNN is run only once), which is more consistent with what would likely be deployed in real-time and/or energy-constrained applications. LeNet [11] was one of the ﬁrst CNN approaches introduced in 1989. It was designed for the task of digit cl”
Radio Machine Learning Dataset Generation with GNU Radio (2016, methodology): “In more recent years we’ve seen the addition of convolutional layers (Le Cun et al., 1989) for visual invariance, the use of Dropout (Srivastava et al., 2014) as a powerful regularizer, the introduction of much faster gradient descent methods using RMSprop (Tieleman & Hinton, 2012) and Adam (Kingma…”
Privacy-preserving Machine Learning through Data Obfuscation (2018, background): “multilayer perceptrons [28], convolutional neural networks [19] and recurrent neural networks [29].”
Medical image classification with convolutional neural network (2014, background): “Artificial Neural Network (ANN) has been studied for many years to solve complex classification problems including image classification [7].”
Eyeriss: A Spatial Architecture for Energy-Efficient Dataflow for Convolutional Neural Networks (2016, background): “These state-of-the-art CNNs [2– 5] are orders of magnitude larger than those used in the 1990s [9], requiring up to hundreds of megabytes for filter weight storage and 30k-600k operations per input pixel.”
Model inversion attacks against collaborative inference (2019, background): “perceptrons [36], convolutional neural networks [28] and recurrent neural networks [38].”