= Christian Szegedy
{c}
<Christian Szegedy> is a researcher whose coauthored work includes <Inception (deep learning architecture)> and methods for training <convolutional neural networks>. <Rethinking the Inception Architecture for Computer Vision by Christian Szegedy and others> examines computationally efficient convolutional design and <regularization>, including <label smoothing>. The original paper's byline identifies him as a coauthor; these are collaborative contributions.<Rethinking the Inception Architecture for Computer Vision by Christian Szegedy and others>[First-page byline and abstract; §§2–3 and 7]{ref}
One idea is to change training targets rather than demand probability one for a single class. For $K$ classes and smoothing amount $\varepsilon$, the paper mixes the original target with a uniform distribution. In an illustrative three-class problem with target $(1,0,0)$ and $\varepsilon=0.1$, this gives
$$
q'=(1-\varepsilon)(1,0,0)+\varepsilon(1/3,1/3,1/3)
=(14/15,1/30,1/30).
$$
The target still favors the labeled class, but allocates some mass to the others. This changes the <cross-entropy> training objective; it is not the same operation as fitting <temperature scaling> after training.<Rethinking the Inception Architecture for Computer Vision by Christian Szegedy and others>[§7, mixture target and uniform-label distribution]{ref}
The architectural work also motivates checking computation counts. In a simplified same-channel comparison, a five-by-five convolution uses $25C^2$ weights, while two successive three-by-three convolutions use $18C^2$, ignoring biases and boundary effects. The reduction is $7/25=28\%$. This own arithmetic illustrates the paper's factorization motivation; the two networks need not represent identical functions, particularly when a nonlinearity intervenes.<Rethinking the Inception Architecture for Computer Vision by Christian Szegedy and others>[§3.1, replacing larger spatial convolutions by smaller ones]{ref}
The scientific value is the combination of testable architecture and training choices: compare classification behavior, parameter counts and computation under stated conditions rather than inferring that one named architecture is always best.
Back to article page
