Closed vandenbergheluke closed 1 year ago
Using the notation and equations from the repository documentation: https://complex-valued-neural-networks.readthedocs.io/en/latest/activations/real_output.html
Current activations implemented in the code are: $out = \frac{\sigma(x)+\sigma(x)}{2}$ $out = \sigma(x) \sigma(x)$ $out = \sigma(\sigma(x) \sigma(x))$ $out = \sigma(\sigma(x)+\sigma(x))$
Proposed change: $out = \frac{\sigma(x)+\sigma(y)}{2}$ $out = \sigma(x) \sigma(y)$ $out = \sigma(\sigma(x) \sigma(y))$ $out = \sigma(\sigma(x)+\sigma(y))$
Using the notation and equations from the repository documentation: https://complex-valued-neural-networks.readthedocs.io/en/latest/activations/real_output.html
Current activations implemented in the code are: $out = \frac{\sigma(x)+\sigma(x)}{2}$ $out = \sigma(x) \sigma(x)$ $out = \sigma(\sigma(x) \sigma(x))$ $out = \sigma(\sigma(x)+\sigma(x))$
Proposed change: $out = \frac{\sigma(x)+\sigma(y)}{2}$ $out = \sigma(x) \sigma(y)$ $out = \sigma(\sigma(x) \sigma(y))$ $out = \sigma(\sigma(x)+\sigma(y))$