Training-free channel planning improves convolutional autoencoder efficiency
Training-Free Bottleneck Width Planning for Convolutional Autoencoders
Computer Vision and Pattern Recognition
Summary
Choosing the right size for the middle part of a compression neural network is hard and usually needs training lots of models. The authors present a method that predicts the needed size accurately without training any network by using statistics from example images. This method closely matches results from trained models on many datasets and helps select efficient network designs quickly. It works well even with different network shapes and error limits, making it easier to build image compression tools.
What this means in practice
- •For machine learning engineers: Predict optimal bottleneck sizes for convolutional autoencoders without training, reducing design time for efficient image compression models.
- •For computer vision developers: Quickly select and validate autoencoder architectures at specified error bounds for grayscale image processing without extensive model fitting.
Authors
Guannan Guo
Abstract
Multiscale Spectral Rate-Distortion (MS-SRD) estimates the bottleneck channels required at user-supplied spatial cuts from training images and a normalized mean-squared error (NMSE) bound, without fitting a neural network. Its covariance-tail rule is exact for shared linear block-convolutional autoencoders under squared error. A nested-scale dominance result motivates reporting the activation-parameter Pareto frontier alongside the minimum-latent candidate. At NMSE <= 0.01 on thirteen grayscale datasets, its latent-size prediction has 0.84% mean absolute percentage error against nonlinear patch-autoencoder boundaries; ten predictions are exact and the remaining three differ by one channel. In a four-dataset deployable comparison, MS-SRD matches all retrospective external widths and all four selected models pass, without training a selector; a 46-fit validation grid and four Least-Volume fits each pass on two datasets. In a skip-closed U-shaped autoencoder at the same bound, five predictions are exact, nine are within one channel, and every failing prediction is one channel short. Experiments at looser bounds show progressively larger nonlinear savings.