The Impact of Architecture and Cost Function on Dissipative Quantum Neural Networks

Name: The Impact of Architecture and Cost Function on Dissipative Quantum Neural Networks
Uploaded: 2026-10-21T09:23:10.094066+00:00
Duration: 211 s
Description: The work proposes a novel architecture for dissipative quantum neural networks (DQNNs) using isometries instead of unitaries to reduce parameters while maintaining universality. It numerically investigates how different cost functions affect training, revealing significant differences in trainability and suggesting the Hilbert-Schmidt distance as a preferred metric for Choi training.

The work proposes a novel architecture for dissipative quantum neural networks (DQNNs) using isometries instead of unitaries to reduce parameters while maintaining universality. It numerically investigates how different cost functions affect training, revealing significant differences in trainability and suggesting the Hilbert-Schmidt distance as a preferred metric for Choi training.

Open the full episode