Real-Time Detection of Charge Jumps in Superconducting Qubits with a Convolutional Neural Network

2026-07-15 · Source: Machine Learning · Field: Science & Research — Physical Sciences & Chemistry, Artificial Intelligence & Machine Learning, Engineering & Applied Sciences · Depth: Expert, quick

Summary

A new online detector for charge jumps in superconducting qubits, based on a dilated causal convolutional neural network (DCCNN), has been developed for real-time deployment on the Quantum Instrumentation Control Kit (QICK) platform. These charge jumps, induced by ionizing radiation, cause correlated errors in quantum computing. The DCCNN is trained on synthetic Ramsey tomography scans derived from qubit templates measured at the Northwestern Experimental Underground Site (NEXUS) at Fermilab. Implemented as FPGA firmware using hls4ml with ap_fixed<16,6> quantization, the detector achieves a per-inference latency of 6.19 μs on the Zynq UltraScale+ RFSoC ZCU216. It matches the detection efficiency of the established offline χ^2 algorithm (0.843 ± 0.022 vs. 0.866 ± 0.020 for |Δq| ∈ [0.1, 0.5] e) without requiring per-qubit hyperparameter tuning. This innovation transforms charge-jump detection into a control-loop primitive, enabling adaptive protocols for quantum-computing error mitigation and particle detection.

Key takeaway

For quantum computing engineers developing fault-tolerant systems, you should integrate this DCCNN-based real-time charge jump detector into your control loops. This shifts detection from post-hoc diagnostics to an in-situ primitive, enabling adaptive error mitigation protocols. Your systems can respond immediately to radiation-induced events, improving qubit stability. This also allows using qubits as particle detectors, all without per-qubit hyperparameter tuning.

Key insights

A DCCNN enables real-time, low-latency detection of superconducting qubit charge jumps, matching offline methods without hyperparameter tuning.

Principles

• Ionizing radiation causes correlated errors in qubits.
• Online detection enables adaptive quantum protocols.

Method

Train a DCCNN on synthetic Ramsey tomography scans, translate to FPGA firmware via hls4ml with ap_fixed<16,6> quantization for in-the-loop deployment.

In practice

• Deploy DCCNN on QICK for real-time error mitigation.
• Use superconducting qubits as particle detectors.

Topics

• Superconducting Qubits
• Charge Jump Detection
• Convolutional Neural Networks
• Quantum Error Mitigation
• FPGA Deployment
• Real-Time Systems

Best for: AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• Analog Quantum Asynchronous Event-Based Graph Neural Network — QA-AEGNNs implement event-based graph neural networks on neutral-atom quantum computers, leveraging Rydberg interactions for message passing.
• Entropy Estimation in Multi-Qutrit Systems via Variational and Classical Neural Networks — VQAs are effective for small multi-qutrit systems, while CNNs offer scalable and robust entropy estimation for larger ones.
• SymQNet: Amortized Acquisition for Low-Latency Adaptive Hamiltonian Learning — Amortized reinforcement learning via SymQNet significantly reduces latency in adaptive Hamiltonian learning by pre-learning experiment acquisition policies.
• Input-Aware Dynamic Backdoor Attack Against Quantum Neural Networks — Q-DIBA introduces the first input-aware dynamic backdoor attack for QNNs, overcoming quantum-specific challenges to achieve stealthy, effective compromise.
• Enhancing Blood Cells Classification using Hybrid Quantum Neural Networks — Hybrid Quantum-Classical Neural Networks enhance blood cell classification by improving feature representation, even under quantum noise.
• Qilimanjaro Pushes Analog Quantum as AI Compute Demands Surge — Analog quantum computing offers a path to reduced errors and accelerated AI by avoiding extensive gate operations.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.