CNN vs Fully connected NN

CNN vs Fully connected NN

CNN has the advantageous properties including sparse interactions with local connectivity, parameter sharing with reduced numbers, and equivariant representation which is invariant to object locations;


Principle

Abstracted features are learned by stacked convolutional and sampling layers.


Pros.

Reduced parameter number,invariance of shift,scale and distortion


Cons.

High computational complexity for high hierarchical model training.



阅读更多

没有更多推荐了,返回首页