Sensitivity of Sparse Codes to Image Distortions

DOI10.1162/NECO_A_01513MaRDI QIDQ5095886zbMATH OpenOpenAlexFDO

Authors Kyle Luther, H. Sebastian Seung

Publication date 11 August 2022

Published in Neural Computation (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2204.07466

zbMATH Keywords

visual cortexsparse codinglearning representations

Mathematics Subject Classification ID

Neural networks for/in biological studies, artificial life and related topics (92B20) Memory and learning in psychology (91E40)

Abstract: Sparse coding has been proposed as a theory of visual cortex and as an unsupervised algorithm for learning representations. We show empirically with the MNIST dataset that sparse codes can be very sensitive to image distortions, a behavior that may hinder invariant object recognition. A locally linear analysis suggests that the sensitivity is due to the existence of linear combinations of active dictionary elements with high cancellation. A nearest neighbor classifier is shown to perform worse on sparse codes than original images. For a linear classifier with a sufficiently large number of labeled examples, sparse codes are shown to yield higher accuracy than original images, but no higher than a representation computed by a random feedforward net. Sensitivity to distortions seems to be a basic property of sparse codes, and one should be aware of this property when applying sparse codes to invariant object recognition.

Recommendations

• The problem of sparse image coding
• Sparsity constrained estimation in image processing and computer vision
• Sparse representations and compressive sensing for imaging and vision
• Sparse representation and reconstruction of image based on compressed sensing
• Image Coding Using Generalized Predictors Based on Sparsity and Geometric Transformations
• Distortion Estimators for Bitplane Image Coding
• Image decoding optimization based on compressive sensing
• Convolutional Sparse and Low-Rank Coding-Based Image Decomposition

Cited in

(7)

• Lower bounds for sparse coding
• The problem of sparse image coding
• A sparse coding model based on structural similarity
• Scalable Coding of Plenoptic Images by Using a Sparse Set and Disparities
• Improved sparse coding under the influence of perceptual attention
• Combining Reconstruction and Discrimination with Class-Specific Sparse Coding
• Testing Properties of Sparse Images

This page was built for publication: Sensitivity of Sparse Codes to Image Distortions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5095886)