Statements (20)
Predicate | Object |
---|---|
gptkbp:instanceOf |
problem in deep learning
|
gptkbp:affects |
ReLU activation function
|
gptkbp:alsoKnownAs |
dead ReLU problem
|
gptkbp:cause |
loss of network capacity
neurons output zero for all inputs |
gptkbp:discusses |
deep learning literature
|
https://www.w3.org/2000/01/rdf-schema#label |
dying ReLU problem
|
gptkbp:mitigatedBy |
gptkb:Leaky_ReLU
gptkb:Parametric_ReLU Randomized ReLU lower learning rate proper weight initialization |
gptkbp:occurredIn |
weights update causes neuron to only output zero
|
gptkbp:relatedTo |
vanishing gradient problem
training deep neural networks |
gptkbp:bfsParent |
gptkb:Leaky_ReLU
gptkb:Rectified_Linear_Unit gptkb:LeakyReLU gptkb:ReLU_function |
gptkbp:bfsLayer |
7
|