|
gptkbp:instance_of
|
gptkb:television_channel
|
|
gptkbp:bfsLayer
|
7
|
|
gptkbp:bfsParent
|
gptkb:Cresnet_2.0
|
|
gptkbp:applies_to
|
Image Classification
|
|
gptkbp:architectural_style
|
gptkb:Deep_Learning
|
|
gptkbp:coat_of_arms
|
50
|
|
gptkbp:connects
|
gptkb:theorem
|
|
gptkbp:contribution
|
Improved training of deep networks
|
|
gptkbp:developed_by
|
gptkb:Kaiming_He
|
|
gptkbp:difficulty
|
high
|
|
gptkbp:established
|
Re LU
|
|
gptkbp:first_performance
|
0.1
|
|
gptkbp:has_variants
|
gptkb:Cresnet_152
gptkb:Cresnet_110
gptkb:Cresnet_164
gptkb:Cresnet_20
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
Cresnet 50
|
|
gptkbp:influenced_by
|
gptkb:Google_Net
VGG Net
|
|
gptkbp:input_output
|
224x224
|
|
gptkbp:inspired_by
|
Highway Networks
|
|
gptkbp:is_a_framework_for
|
gptkb:Graphics_Processing_Unit
gptkb:Py_Torch
|
|
gptkbp:is_available_in
|
Model Zoos
|
|
gptkbp:is_connected_to
|
gptkb:theorem
|
|
gptkbp:is_designed_for
|
Image Recognition
|
|
gptkbp:is_optimized_for
|
gptkb:theorem
Stochastic Gradient Descent
|
|
gptkbp:is_popular_in
|
gptkb:viewpoint
|
|
gptkbp:is_used_in
|
Kaggle Competitions
|
|
gptkbp:localization
|
gptkb:theorem
|
|
gptkbp:losses
|
Cross-Entropy Loss
|
|
gptkbp:notable_feature
|
Skip Connections
|
|
gptkbp:performance
|
Top-1 accuracy
|
|
gptkbp:predecessor
|
gptkb:Cresnet_34
|
|
gptkbp:provides_information_on
|
gptkb:Image_Net
gptkb:theorem
|
|
gptkbp:ram
|
high
|
|
gptkbp:rank
|
1st place in ILSVRC 2015
|
|
gptkbp:reduces
|
gptkb:theorem
|
|
gptkbp:related_to
|
gptkb:streaming_service
Fine-tuning
|
|
gptkbp:resolution
|
1000 classes
|
|
gptkbp:resulted_in
|
Top-5 accuracy of 76.0% on Image Net
|
|
gptkbp:roll_rate
|
gptkb:theorem
|
|
gptkbp:successor
|
gptkb:Cresnet_101
|
|
gptkbp:supports
|
Multi-GPU training
|
|
gptkbp:training
|
varies by hardware
|
|
gptkbp:uses
|
gptkb:theorem
Global Average Pooling
Residual Learning
|
|
gptkbp:year_created
|
gptkb:2015
|