Statements (49)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
abbreviation
|
| gptkbp:application |
autonomous vehicles
machine translation diagnosis speech synthesis image classification object detection |
| gptkbp:canBe |
gptkb:convolutional_neural_network
gptkb:feedforward_neural_network recurrent neural network |
| gptkbp:developedBy |
1980s
|
| gptkbp:hasComponent |
hidden layer
input layer output layer |
| gptkbp:hasProperty |
multiple layers
nonlinear activation functions |
| https://www.w3.org/2000/01/rdf-schema#label |
DNN
|
| gptkbp:implementedIn |
gptkb:TensorFlow
gptkb:Keras gptkb:MXNet gptkb:Caffe gptkb:PyTorch |
| gptkbp:improves |
batch normalization
data augmentation dropout transfer learning residual connections attention mechanisms |
| gptkbp:limitation |
overfitting
computationally expensive vanishing gradient problem requires large labeled data |
| gptkbp:popularizedBy |
gptkb:Geoffrey_Hinton
gptkb:Yann_LeCun gptkb:Yoshua_Bengio |
| gptkbp:relatedTo |
gptkb:convolutional_neural_network
deep learning |
| gptkbp:requires |
large datasets
high computational power |
| gptkbp:standsFor |
gptkb:Deep_Neural_Network
|
| gptkbp:trainer |
backpropagation
gradient descent |
| gptkbp:usedIn |
gptkb:artificial_intelligence
gptkb:machine_learning computer vision natural language processing speech recognition |
| gptkbp:bfsParent |
gptkb:大冶北站
|
| gptkbp:bfsLayer |
8
|