LRN
Input
x:
- Required: yes
- Type: float16, float32
- Description: performs a kind of lateral inhibition by normalizing over local input regions, power(k + (alpha/n) * sum(square(x)),beta)
- Restriction: Must be of type float16.
Attribute
- local_size:
- Required: no
- Type: uint
- Description: Must be an odd value up to 255 (default = 5).
- Restriction: Must be an odd value up to 255.
- alpha:
- Required: no
- Type: float
- Description: default = 1.0
- Restrictions: none
- beta:
- Required: no
- Type: float
- Description: default = 0.75
- Restrictions: none
- norm_region:
- k:
- Required: no
- Type: float
- Description: default = 1.0
- Restrictions: none
Output
y:
- Required: yes
- Type: float16, float32
- Description: performs a kind of lateral inhibition by normalizing over local input regions, power(k + (alpha/n) * sum(square(x)),beta)
- Restriction: Must be of type float16.
Parent topic: Supported Caffe Operators