Quantum

Parameter-Efficient Multilinear Readout for Quantum Reservoir Computing

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 14/10/2026, 12:17:25 UTC.

Briefings

A new method called Multilinear Readout for Quantum Reservoir Computing compresses dense neural readouts by 12.1 times while preserving quantum tensor structure. The approach uses mode-wise projections to reduce parameter counts from 64TnF to d(T+n+F)+64d^3. It achieved the highest full-data TFIM accuracy among tested methods and showed improved low-data stability compared to neural MLPs on Mackey-Glass forecasting.

Key points

• ML-QRC reduces feature-to-hidden parameter count from 64TnF to d(T+n+F)+64d^3, achieving a 12.1x compression at T=50, n=6, d=8, F=n+binom(n,2).
• ML-QRC obtained the numerically highest full-data TFIM accuracy among tested methods while using 12.0x fewer parameters than the full dense MLP.

Context

Quantum Reservoir Computing uses fixed quantum dynamics as a high-dimensional feature map, but standard pipelines flatten reservoir measurements and train a linear readout, ignoring natural tensor structure across time steps, qubits, and observables. This study proposes ML-QRC as a tensor-structured neural readout compressed through mode-wise projections.

Limitations

The study covers Mackey-Glass forecasting and TFIM phase classification. ML-QRC remains statistically competitive with Ridge on Mackey-Glass but does not transfer to TFIM, where Ridge performs best at 50% data. The advantage in low-data stability on Mackey-Glass does not transfer to TFIM. The research is preprint status (arXiv:2610.13974v1) and snapshot limitations apply.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 25/10/2026, 20:13:40 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeifpgy4qxjnepm46ak33uwohcq5w5t65gvkob7knt4bg3bn4gezwey