Computer scienceData scienceMachine learningComputer visionConvolutional neural networks

ResNet

Bottleneck layers

Report a typo

What is the main purpose of the bottleneck residual connection in the deeper variants of ResNet?

Select one option from the list

They are used to prevent the transfer of residuals from previous layers to subsequent layers

They primarily normalize the activation outputs

They primarily reduce the number of parameters through a combination of 1x1 and the 3x3 convolutions

They primarily act a weight regularizer

___

Create a free account to access the full topic