Inception

The Inception model was introduced by researchers at Google in 2014.

It made use of 1x1 convolutions that are used to increase or shrink the number of channels without affecting the height and width of the image.

The concept of 1x1 convolutions was introduced in the paper for the Network in Network model.

A single inception block allows the network to use a combination of 1x1, 3x3, 5x5 convolutions and pooling. Inception blocks usually use 1x1 convolutions to reduce the input data volume’s size before applying 3x3 and 5x5 convolutions.

The Inception model consists of multiple such inception blocks.S