Quantum-LSTM hybrid cuts physics model training data by 100×
Federated learning with quantum-enhanced LSTM achieves classical accuracy on SUSY classification using 20K samples instead of 2M, with under 300 parameters.
Source: arxiv/cs.LG · Abhishek Sawaika, Durga Pritam Suggisetti, Udaya Parampalli, Rajkumar Buyya · open original ↗
Hybrid quantum-classical LSTM in federated setup matches classical deep learning on high-energy physics tasks with 100× fewer data points.
- — Combines quantum variational circuits with LSTM to capture complex feature relationships and temporal correlations.
- — Federated architecture distributes training across nodes, reducing computational burden on individual NISQ devices.
- — Achieves ±1% accuracy parity with classical benchmarks on SUSY dataset using only 20K samples versus 2M baseline.
- — Model footprint under 300 parameters enables deployment on resource-constrained distributed infrastructure.
- — Outperforms standalone variational quantum circuit approaches in both accuracy and data efficiency.
- — Addresses practical NISQ hardware limitations by avoiding reliance on single powerful quantum processor.
Astrobobo tool mapping
- Knowledge Capture Record the three key design choices: federated topology (how many nodes, data split), quantum circuit ansatz (depth, gates), and LSTM architecture (hidden size, layers). Link to the arxiv paper and your baseline results.
- Focus Brief Summarize the 100× data efficiency claim with caveats: applies to SUSY classification, synthetic data, assumes noise model matches your hardware. Note open questions (convergence speed, real-world noise robustness) for follow-up.
- Reading Queue Queue related papers on federated quantum ML and NISQ optimization to understand communication costs and noise mitigation strategies not covered in this abstract.
Frequently asked
- Current quantum computers (NISQ devices) are noisy, expensive, and limited in qubit count. Federated learning distributes the quantum computation across multiple smaller nodes and classical servers, reducing the burden on any single quantum processor. This makes the approach practical for organizations with distributed data and limited quantum hardware access.
cite ▸
APA
Abhishek Sawaika, Durga Pritam Suggisetti, Udaya Parampalli, Rajkumar Buyya. (2026, April 20). Quantum-LSTM hybrid cuts physics model training data by 100×. Astrobobo Content Engine (rewrite of arxiv/cs.LG). https://astrobobo-content-engine.vercel.app/article/quantum-lstm-hybrid-cuts-physics-model-training-data-by-100-8fa3b8
MLA
Abhishek Sawaika, Durga Pritam Suggisetti, Udaya Parampalli, Rajkumar Buyya. "Quantum-LSTM hybrid cuts physics model training data by 100×." Astrobobo Content Engine, 20 Apr 2026, https://astrobobo-content-engine.vercel.app/article/quantum-lstm-hybrid-cuts-physics-model-training-data-by-100-8fa3b8. Based on "arxiv/cs.LG", https://arxiv.org/abs/2604.15775.
BibTeX
@misc{astrobobo_quantum-lstm-hybrid-cuts-physics-model-training-data-by-100-8fa3b8_2026,
author = {Abhishek Sawaika, Durga Pritam Suggisetti, Udaya Parampalli, Rajkumar Buyya},
title = {Quantum-LSTM hybrid cuts physics model training data by 100×},
year = {2026},
url = {https://astrobobo-content-engine.vercel.app/article/quantum-lstm-hybrid-cuts-physics-model-training-data-by-100-8fa3b8},
note = {Astrobobo rewrite of arxiv/cs.LG, https://arxiv.org/abs/2604.15775},
}