Flatten-T Swish: a thresholded ReLU-Swish-like activation function for deep learning.
Hock Hung ChiengNoorhaniza WahidPauline OngSai Raj Kishore PerlaPublished in: CoRR (2018)
Keyphrases
- deep learning
- activation function
- neural network
- feed forward
- artificial neural networks
- unsupervised learning
- neural nets
- hidden layer
- learning rate
- back propagation
- machine learning
- multilayer perceptron
- radial basis function
- network architecture
- basis functions
- weakly supervised
- multi layer perceptron
- pattern recognition
- rbf neural network
- mental models
- training phase
- feature space
- natural language processing
- multiscale
- fuzzy neural network
- pairwise
- prior knowledge
- supervised learning