Statements (53)
Predicate | Object |
---|---|
gptkbp:instanceOf |
gptkb:spacecraft
|
gptkbp:aimsTo |
maintain variance
|
gptkbp:appliesTo |
weights
|
gptkbp:designedBy |
gptkb:Xavier_Glorot
|
https://www.w3.org/2000/01/rdf-schema#label |
Xavier initialization
|
gptkbp:isAttendedBy |
researchers
practitioners |
gptkbp:isAvenueFor |
fully connected layers
convolutional layers |
gptkbp:isConsidered |
best practice
|
gptkbp:isDiscussedIn |
research papers
machine learning courses |
gptkbp:isEvaluatedBy |
performance metrics
|
gptkbp:isExaminedBy |
tutorials
documentation |
gptkbp:isInfluencedBy |
network architecture
activation functions layer depth |
gptkbp:isLocatedIn |
gptkb:PyTorch
TensorFlow |
gptkbp:isPartOf |
model training process
random initialization model optimization training strategies normal initialization uniform initialization |
gptkbp:isRecognizedFor |
ReLU activation function
sigmoid activation function tanh activation function |
gptkbp:isRelatedTo |
hyperparameter tuning
loss functions backpropagation gradient descent |
gptkbp:isSimilarTo |
He initialization
|
gptkbp:isUsedIn |
computer vision
deep learning natural language processing transformers reinforcement learning transfer learning graph neural networks generative models autoencoders adversarial networks reinforcement learning algorithms sequence models |
gptkbp:isUtilizedIn |
experiments
benchmarks |
gptkbp:offersServices |
weights based on fan-in and fan-out
|
gptkbp:previouslyKnownAs |
Glorot initialization
|
gptkbp:usedFor |
neural networks
|
gptkbp:works |
exploding gradients
vanishing gradients |