Variable rate neural compression for sparse detector data

doi:10.1016/j.patter.2025.101452

Variable rate neural compression for sparse detector data

2026-02-02

0

OA

AI

DOI：10.1016/j.patter.2025.101452

Original

Save

View PDF

Abstract

En 中文

• Sparse-voxel key-point selection enables adaptive compression by signal density • Outstanding reconstruction accuracy at higher compression ratios • Lightweight model with just 382 trainable parameters • Compression throughput scales favorably for very sparse data

Keywords:

deep learning

autoencoder

high-throughput inference

data compression

sparse data

sparse neural network

high-energy and nuclear physics

Variable rate neural compression for sparse detector data

Abstract

AI Summary

Journal

Organization