= Inception
{disambiguate=deep learning architecture}
{c}
<Inception (deep learning architecture)> is a family of <convolutional neural network> architectures built around modules with parallel processing branches. An early module combines several convolution sizes and a <pooling layer>, then joins their outputs along the channel dimension. <Pointwise convolution> can reduce channel counts before more expensive spatial operations.<Going Deeper with Convolutions by Christian Szegedy and others>[§4 and Fig. 2, parallel branches and dimensionality reduction]{ref}
For a stipulated shape example, let every branch produce a $28\times28$ grid, with output channel counts $64$, $128$, $32$ and $32$. Joining channels gives a $28\times28\times256$ output because
$$
64+128+32+32=256.
$$
The equal spatial grid sizes are an assumption ensured by the branch design. Channel joining does not add corresponding values into one channel; it retains the distinct branch outputs as separate channels of the combined <neural network feature map>.
A cost example explains the reduction branch. Ignoring biases, a dense three-by-three <convolutional layer> from $192$ input channels to $128$ output channels has
$$
3^2(192)(128)=221184
$$
weights. Reducing first from $192$ to $32$ channels with a one-by-one <pointwise convolution>, then applying the three-by-three layer, uses
$$
(192)(32)+3^2(32)(128)=43008
$$
weights. This own example isolates channel dimensions; it is not a reported layer configuration or a claim that the two computations represent identical functions.
The original paper presents this design in the GoogLeNet network and evaluates image classification and detection. Later <Rethinking the Inception Architecture for Computer Vision by Christian Szegedy and others> develops additional factorization and training choices. The family name therefore does not identify one immutable list of layers or one universal performance result.<Going Deeper with Convolutions by Christian Szegedy and others>[Abstract; §§4–5, modules and network]{ref}<Rethinking the Inception Architecture for Computer Vision by Christian Szegedy and others>[§§2–3, design principles and factorization]{ref}
Back to article page
