|
gptkbp:instance_of
|
gptkb:neural_networks
|
|
gptkbp:application
|
image classification
|
|
gptkbp:architecture
|
deep learning
|
|
gptkbp:coat_of_arms
|
8
|
|
gptkbp:competitors
|
gptkb:Goog_Le_Net
|
|
gptkbp:contribution_to_ai
|
popularized deep learning techniques
|
|
gptkbp:convolutional_layers
|
gptkb:5
|
|
gptkbp:developed_by
|
gptkb:Alex_Krizhevsky
|
|
gptkbp:drops
|
0.5
|
|
gptkbp:evaluates
|
top-5 accuracy
|
|
gptkbp:field_of_study
|
computer vision
|
|
gptkbp:fully_connected_layers
|
gptkb:3
|
|
gptkbp:gpu
|
gptkb:NVIDIA_GTX_580
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
Alex Net
|
|
gptkbp:image_normalization
|
mean subtraction
|
|
gptkbp:impact
|
pioneered deep learning in computer vision
|
|
gptkbp:influenced_by
|
Hinton's work on neural networks
|
|
gptkbp:initialization_method
|
Xavier initialization
|
|
gptkbp:initiated_by
|
Re LU
|
|
gptkbp:input_output
|
227x227 pixels
|
|
gptkbp:is_a_framework_for
|
gptkb:Caffe
|
|
gptkbp:is_standardized_by
|
no
|
|
gptkbp:local_response_normalization
|
yes
|
|
gptkbp:losses
|
cross-entropy loss
|
|
gptkbp:notable_feature
|
use of Re LU activations
|
|
gptkbp:orbital_period
|
60 million
|
|
gptkbp:output_classes
|
1000
|
|
gptkbp:paper_title
|
Image Net Classification with Deep Convolutional Neural Networks
|
|
gptkbp:pooling_type
|
max pooling
|
|
gptkbp:predecessor
|
gptkb:Le_Net
|
|
gptkbp:provides_information_on
|
gptkb:Image_Net
yes
|
|
gptkbp:significance
|
won the Image Net Large Scale Visual Recognition Challenge
|
|
gptkbp:successor
|
gptkb:VGGNet
|
|
gptkbp:top5_accuracy
|
84.6%
|
|
gptkbp:training
|
stochastic gradient descent
5-6 days
|
|
gptkbp:year_established
|
gptkb:2012
|
|
gptkbp:bfsParent
|
gptkb:Image_Net
gptkb:Image_Net_competition
gptkb:Le_Net_architecture
gptkb:Alex_Krizhevsky
gptkb:neural_networks
|
|
gptkbp:bfsLayer
|
4
|