Statements (51)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:physicist
|
| gptkbp:adaptation |
learning rate
|
| gptkbp:appliesTo |
neural networks
|
| gptkbp:developedBy |
Geoff_Hinton
|
| gptkbp:hasFeature |
epsilon
decay rate |
| https://www.w3.org/2000/01/rdf-schema#label |
RMSprop
|
| gptkbp:improves |
stochastic gradient descent
|
| gptkbp:isBasedOn |
adaptive learning rates
vanishing gradients |
| gptkbp:isEvaluatedBy |
complex models
non-stationary objectives large neural networks |
| gptkbp:isInfluencedBy |
Adadelta
|
| gptkbp:isLocatedIn |
gptkb:PyTorch
gptkb:Keras TensorFlow Caffe MXNet Theano |
| gptkbp:isNotableFor |
sparse data
|
| gptkbp:isPartOf |
gptkb:Adam
SGD backpropagation gradient descent family Nesterov momentum |
| gptkbp:isPopularIn |
computer vision
natural language processing |
| gptkbp:isRecognizedFor |
small datasets
simple tasks convex optimization problems non-convex optimization problems shallow networks linear_models |
| gptkbp:isSimilarTo |
gptkb:Adagrad
|
| gptkbp:isSuitableFor |
large datasets
|
| gptkbp:isUsedFor |
training deep learning models
dropout |
| gptkbp:isUsedIn |
GANs
reinforcement learning speech recognition unsupervised learning image classification transfer learning time series forecasting semi-supervised learning |
| gptkbp:isVisitedBy |
momentum
|
| gptkbp:maintains |
a moving average
|
| gptkbp:requires |
hyperparameters
|
| gptkbp:usedIn |
machine learning
|
| gptkbp:wasAffecting |
2012
|