Operational Indirect Rate-Distortion Functions for Classification Tasks
Abstract
In goal-oriented communications, indirect source coding involves two sources: a semantic source X and an observation Z, which is the only source available at the encoder. In this paper, we consider this framework for classification, where the decoder only needs to reconstruct X. We argue that the gap between the indirect rate-distortion function and the performance of practical learned coding schemes, such as variational information bottleneck (VIB) and variational autoencoder (VAE) architectures, comes from three effects: the latent representation, the choice of the deep neural network (DNN) architecture and training loss, and whether the encoder is aware of the task. To quantify these effects, we introduce four operational rate-distortion functions that account for the DNN architecture and the training loss, and we establish a hierarchy among them. We then derive these functions for a Gaussian mixture source, where X is the mixture state. Numerical results, both for the Gaussian mixture and for practical coding schemes on the CIFAR-100 dataset, quantify the rate loss incurred by each effect.
Related
- Growth, Distortion, and Schwarzian Norm Estimates for Exponentially Convex Functions
- Cooperative Multi-Task Semantic Communication for Joint Classification and Regression Tasks
- Learning Style, Forgetting Semantics: A Case Study of SFT and RFT on Classification Tasks
- Gaussian Approximation of Besov Functions for Logistic Classification on Manifolds
- Characterizing Nonlinearities in IM-DD Links via the Best Linear Approximation: Distortion Analysis and Modulation Optimization
BibTeX
@misc{miguet2026operational,
title = {{Operational Indirect Rate-Distortion Functions for Classification Tasks}},
author = {Yann Miguet and Elsa Dupraz and Philippe Mary},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.07946},
eprint = {2611.07946},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.07946},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Operational Indirect Rate-Distortion Functions for Classification Tasks
AU - Yann Miguet
AU - Elsa Dupraz
AU - Philippe Mary
PY - 2026
DA - 2026/11/05
PB - arXiv
DO - 10.48550/arxiv.2611.07946
UR - https://arxiv.org/abs/2611.07946
AB - In goal-oriented communications, indirect source coding involves two sources: a semantic source X and an observation Z, which is the only source available at the encoder. In this paper, we consider this framework for classification, where the decoder only needs to reconstruct X. We argue that the gap between the indirect rate-distortion function and the performance of practical learned coding schemes, such as variational information bottleneck (VIB) and variational autoencoder (VAE) architectures, comes from three effects: the latent representation, the choice of the deep neural network (DNN) architecture and training loss, and whether the encoder is aware of the task. To quantify these effects, we introduce four operational rate-distortion functions that account for the DNN architecture and the training loss, and we establish a hierarchy among them. We then derive these functions for a Gaussian mixture source, where X is the mixture state. Numerical results, both for the Gaussian mixture and for practical coding schemes on the CIFAR-100 dataset, quantify the rate loss incurred by each effect.
ER -