inception-v2.html.pm

#lang pollen

◊define-meta[page-title]{What is Inception-V2?}

◊title{What is Inception-v2?}  
◊heading{It's Inception plus Batch Normalization and some 5x5 factoring}
  
In ◊a[#:href "https://arxiv.org/abs/1602.07261"]{Inception-v4}, Szegedy et al. describe Inception-v2:

◊q{"Later the Inception architecture was refined in various ways, first
by the introduction of batch normalization (Inception-v2) by Ioffe et
al."}

That paper by Ioffe et al. describes their model like this:

◊q{"The main difference to [Inception-v1] is that the 5 × 5
convolutional layers are replaced by two consecutive layers
of 3 × 3 convolutions with up to 128 filters."}

◊heading{It's an intermediate variant of Inception before -v3, and different than batch-normalized Inception}
  
◊a[#:href "https://arxiv.org/abs/1512.00567"]{Rethinking the Inception
Architecture for Computer Vision} has a section (Section 6) titled
"Inception-v2". It describes a network with many more changes than the
Batch Normalization paper listed compared to Inception-v1:
  
◊q{"Here we are connecting the dots from above and propose a new
architecture with improved performance on the ILSVRC 2012
classification benchmark. The layout of our network is given in table
1. Note that we have factorized the traditional 7 × 7 convolution into
three 3 × 3 convolutions based on the same ideas as described in
section 3.1.  For the Inception part of the network, we have 3
traditional inception modules at the 35×35 with 288 filters each. This
is reduced to a 17 × 17 grid with 768 filters using the grid reduction
technique described in section 5. This is is followed by 5 instances
of the factorized inception modules as depicted in figure 5. This is
reduced to a 8 × 8 × 1280 grid with the grid reduction technique
depicted in figure 10. At the coarsest 8 × 8 level, we have two
Inception modules as depicted in figure 6, with a concatenated output
filter bank size of 2048 for each tile. The detailed structure of the
network, including the sizes of filter banks inside the Inception
modules, is given in the supplementary material, given in
the ◊em{model.txt} that is in the tar-file of this submission."}
  
I can't find this model.txt file anywhere on the internet, but they present this table describing Inception-v2:

◊fig[#:src "assets/possibly-inception-v2.png"]{}

"Rethinking Inception" also includes a table that has separate entries for:

◊itemize{
BN-GoogLeNet

BN-Inception

Inception-v2
}