Statements (23)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:physicist
|
| gptkbp:appearsIn |
deep learning
|
| gptkbp:benefits |
sparse data
|
| gptkbp:characteristics |
adaptive learning rate
|
| gptkbp:developedBy |
gptkb:D.P._Kingma
|
| gptkbp:hasDepartment |
accumulation of past gradients
|
| https://www.w3.org/2000/01/rdf-schema#label |
Adagrad
|
| gptkbp:improves |
gptkb:AdaDelta
|
| gptkbp:introduced |
2011
|
| gptkbp:isUpdatedBy |
based on historical gradients
|
| gptkbp:notableFeature |
computer vision
natural language processing recommender systems |
| gptkbp:performance |
better for infrequent features
|
| gptkbp:relatedTo |
gradient descent
|
| gptkbp:requires |
initial learning rate
|
| gptkbp:theory |
O(n)
|
| gptkbp:usedFor |
training machine learning models
|
| gptkbp:variant |
gptkb:Adam
gptkb:RMSprop AdaMax Nadam |
| gptkbp:worksWith |
stochastic gradient descent
|