Statements (57)
| Predicate | Object |
|---|---|
| gptkbp:instance_of |
gptkb:Artificial_Intelligence
|
| gptkbp:can_be_combined_with |
RMSProp
momentum optimization |
| gptkbp:can_handle |
sparse gradients
|
| gptkbp:developed_by |
gptkb:2014
gptkb:D._P._Kingma |
| gptkbp:has_function |
beta1
beta2 epsilon learning rate |
| https://www.w3.org/2000/01/rdf-schema#label |
Adam optimization algorithm
|
| gptkbp:improves |
gptkb:Adagrad
SGD RMSProp |
| gptkbp:is_based_on |
first moment estimate
second moment estimate |
| gptkbp:is_compared_to |
other optimization algorithms
|
| gptkbp:is_compatible_with |
online learning
transfer learning mini-batch training |
| gptkbp:is_considered_as |
state-of-the-art
|
| gptkbp:is_described_as |
gptkb:Documentation
research papers online tutorials |
| gptkbp:is_evaluated_by |
accuracy
cross-validation grid search random search validation loss training loss |
| gptkbp:is_implemented_in |
gptkb:Tensor_Flow
gptkb:Keras gptkb:Py_Torch |
| gptkbp:is_less_sensitive_to |
initialization
|
| gptkbp:is_often_used_in |
computer vision
natural language processing |
| gptkbp:is_part_of |
model optimization
training process |
| gptkbp:is_popular_in |
gptkb:neural_networks
|
| gptkbp:is_recommended_by |
research settings
production settings |
| gptkbp:is_robust_to |
noisy gradients
|
| gptkbp:is_used_for |
backpropagation
gradient descent |
| gptkbp:is_used_in |
gptkb:machine_learning
deep learning |
| gptkbp:provides |
faster convergence
|
| gptkbp:requires |
hyperparameter tuning
|
| gptkbp:suitable_for |
large datasets
non-stationary objectives very small datasets highly oscillatory functions very deep networks ill-conditioned problems |
| gptkbp:uses |
adaptive learning rates
|
| gptkbp:bfsParent |
gptkb:Ada_Max
|
| gptkbp:bfsLayer |
6
|