Selective Feature Re-Encoded Quantum Convolutional Neural Network with Joint Optimization for Image Classification
Shaswata Mahernob Sarkar, Sheikh Iftekhar Ahmed, Jishnu Mahmud, Shaikh Anowarul Fattah, Gaurav Sharma · 2025
Read Selective Feature Re-Encoded Quantum Convolutional Neural Network with Joint Optimization for Image Classification by Shaswata Mahernob Sarkar, Sheikh Iftekhar Ahmed, Jishnu Mahmud, Shaikh Anowarul Fattah, Gaurav Sharma (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-02086.txt.
Description
Five authors redesign a quantum neural network for image classification around two ideas: feed it the right features twice, and train its two halves together. Quantum convolutional networks suit today's noisy quantum hardware, but their accuracy depends on which input features get encoded. The team's selective re-encoding strategy points the circuit at the most informative features, steering it toward the useful corners of its state space.
The parallel design takes features from two classical squeezers at once, principal component analysis and an autoencoder, inside one training loop. Joint training lets the halves adjust to each other instead of voting after the fact. On two standard image sets, re-encoding clearly sharpened the circuit, and the joint model beat both single models and a late-voting ensemble. The authors claim better accuracy and better generalization.
Catalog record
| CREATOR | Shaswata Mahernob Sarkar, Sheikh Iftekhar Ahmed, Jishnu Mahmud, Shaikh Anowarul Fattah, Gaurav Sharma |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-02086