LeNet

E17289

LeNet is one of the earliest convolutional neural network architectures, pioneering modern deep learning approaches to image recognition and handwritten digit classification.

Observed surface forms (1)


Statements (46)

Predicate Object
instanceOf convolutional neural network architecture
deep learning model
image recognition model
activationFunction sigmoid function
tanh function
basedOn artificial neural networks
backpropagation
convolution operation
gradient descent
developedAt Bell Telephone Laboratories
surface form: AT&T Bell Labs
developer Léon Bottou
Patrick Haffner
Yann LeCun
Yoshua Bengio
field computer vision
deep learning
machine learning
pattern recognition
hasComponent convolutional layer
fully connected layer
output layer
pooling layer
subsampling layer
historicalSignificance one of the earliest successful CNNs
pioneered modern deep learning for vision
influenced AlexNet
ResNet
VGG
modern CNN architectures
inputDomain 28x28 pixel images
inputType grayscale images
learningParadigm end-to-end learning
lossFunction cross-entropy loss
notableDataset MNIST
notablePublication Gradient-based learning applied to document recognition
optimizationObjective classification accuracy
primaryTask bank check digit recognition
zip code recognition
publicationYear 1998
regularization weight decay
trainingMethod stochastic gradient descent
supervised learning
usedFor handwritten digit recognition
image classification
optical character recognition
yearProposed 1989

Referenced by (5)

Full triples — surface form annotated when it differs from this entity's canonical label.

Yann LeCun coined LeNet
Yann LeCun knownFor LeNet
Yann knownFor LeNet
subject surface form: Yann LeCun
Léon Bottou notableWork LeNet
this entity surface form: LeNet-5 convolutional neural network (with Yann LeCun and others)