Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Learning on JPEG-LDPC Compressed Images: Classifying with Syndromes

Ahcen Aliouat, Elsa Dupraz

TL;DR

This work tackles the problem of performing image-related learning directly on compressed data in goal-oriented communications by replacing traditional entropy coders with LDPC-based coding and feeding the resulting syndromes to a GRU classifier. The JPEG-like pipeline preserves more of the image structure than Huffman/Arithmetic coding, enabling higher classification accuracy with significantly smaller models. The results show that LDPC-coded syndromes yield better accuracy than baselines and reduce learning complexity, enabling decoding-free inference. The approach offers practical impact for fast, end-to-end learning on compressed data without full decompression in communication and sensing systems.

Abstract

In goal-oriented communications, the objective of the receiver is often to apply a Deep-Learning model, rather than reconstructing the original data. In this context, direct learning over compressed data, without any prior decoding, holds promise for enhancing the time-efficient execution of inference models at the receiver. However, conventional entropic-coding methods like Huffman and Arithmetic break data structure, rendering them unsuitable for learning without decoding. In this paper, we propose an alternative approach in which entropic coding is realized with Low-Density Parity Check (LDPC) codes. We hypothesize that Deep Learning models can more effectively exploit the internal code structure of LDPC codes. At the receiver, we leverage a specific class of Recurrent Neural Networks (RNNs), specifically Gated Recurrent Unit (GRU), trained for image classification. Our numerical results indicate that classification based on LDPC-coded bit-planes surpasses Huffman and Arithmetic coding, while necessitating a significantly smaller learning model. This demonstrates the efficiency of classification directly from LDPC-coded data, eliminating the need for any form of decompression, even partial, prior to applying the learning model.

Learning on JPEG-LDPC Compressed Images: Classifying with Syndromes

TL;DR

This work tackles the problem of performing image-related learning directly on compressed data in goal-oriented communications by replacing traditional entropy coders with LDPC-based coding and feeding the resulting syndromes to a GRU classifier. The JPEG-like pipeline preserves more of the image structure than Huffman/Arithmetic coding, enabling higher classification accuracy with significantly smaller models. The results show that LDPC-coded syndromes yield better accuracy than baselines and reduce learning complexity, enabling decoding-free inference. The approach offers practical impact for fast, end-to-end learning on compressed data without full decompression in communication and sensing systems.

Abstract

In goal-oriented communications, the objective of the receiver is often to apply a Deep-Learning model, rather than reconstructing the original data. In this context, direct learning over compressed data, without any prior decoding, holds promise for enhancing the time-efficient execution of inference models at the receiver. However, conventional entropic-coding methods like Huffman and Arithmetic break data structure, rendering them unsuitable for learning without decoding. In this paper, we propose an alternative approach in which entropic coding is realized with Low-Density Parity Check (LDPC) codes. We hypothesize that Deep Learning models can more effectively exploit the internal code structure of LDPC codes. At the receiver, we leverage a specific class of Recurrent Neural Networks (RNNs), specifically Gated Recurrent Unit (GRU), trained for image classification. Our numerical results indicate that classification based on LDPC-coded bit-planes surpasses Huffman and Arithmetic coding, while necessitating a significantly smaller learning model. This demonstrates the efficiency of classification directly from LDPC-coded data, eliminating the need for any form of decompression, even partial, prior to applying the learning model.
Paper Structure (17 sections, 8 equations, 3 figures, 3 tables)

This paper contains 17 sections, 8 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. JPEG Coding with LDPC Codes
  3. JPEG Coding
  4. JPEG Coding with LDPC Codes
  5. Classification over LDPC-coded Images
  6. GRU for compressed image sequences classification
  7. Learning setups
  8. Data and Experiments
  9. Datasets
  10. Code parameters
  11. Model parameters
  12. Rate gain
  13. Results and Discussion
  14. Accuracy comparison
  15. Accuracy-Complexity balance
  16. ...and 2 more sections

Figures (3)

  • Figure 1: First setup: Syndromes obtained from the LDPC-coded bit-planes are fed as input of a GRU model for classification
  • Figure 2: Second setup: Syndromes of the DCT-LDPC coefficients bit-planes are fed as inputs of a GRU model for classification
  • Figure 3: Considered learning model for classification