Machine Learning for Charge State Characterization of Isolated Double Quantum Dots

2026-07-23 · Source: Machine Learning · Field: Science & Research — Physical Sciences & Chemistry, Artificial Intelligence & Machine Learning · Depth: Advanced, quick

Summary

Two convolutional neural networks, CSMClassifier and ChargeLineNet, have been developed to automate the charge state characterization and tuneup of silicon metal-oxide-semiconductor (SiMOS) double-quantum-dot devices. These models, with fewer than one million parameters, were trained on charge stability maps (CSMs) from 32 SiMOS devices measured at approximately 1 K. CSMClassifier achieved 94% macro-averaged accuracy across three quality classes on 2,407 held-out images by identifying charge instability and sensor artifacts. ChargeLineNet achieved 95.3% exact line-count accuracy on 1,131 held-out images by localizing charge-transition lines and determining electron occupancy. Combined, the pipeline correctly determines electron occupancy for 93.8% of clean held-out images. Pre-training on synthetic images significantly improves label efficiency, maintaining over 90% accuracy with limited experimental fine-tuning. The models are compact, occupying only 6.5 MB, and process images in less than 60 ms on standard hardware.

Key takeaway

For Research Scientists and Machine Learning Engineers scaling semiconductor quantum dot arrays, these specialized convolutional neural networks offer a practical path toward automated tuneup. You should consider integrating such compact, efficient ML models into your cryogenic probing systems to streamline charge stability map analysis. This approach can significantly reduce manual effort and accelerate the development of fault-tolerant quantum computing architectures, especially in isolated-mode regimes.

Key insights

Compact ML models automate quantum dot charge state characterization and tuneup in isolated-mode devices.

Principles

• Isolated-mode CSMs are well-suited for compact, task-specific models.
• Pre-training on synthetic data substantially improves label efficiency.
• Fine-tuning pre-trained models maintains high accuracy with limited experimental data.

Method

A pipeline combining CSMClassifier for artifact identification and ChargeLineNet for charge-transition line localization and electron occupancy determination automates quantum dot characterization.

In practice

• Utilize CSMClassifier to identify charge instability and sensor artifacts.
• Employ ChargeLineNet for precise charge-transition line localization.
• Implement synthetic data pre-training to reduce experimental data requirements.

Topics

• Quantum Dots
• Machine Learning
• Convolutional Neural Networks
• Charge Stability Maps
• Quantum Computing
• Automated Tuneup

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

Related on AIssential

• From hyperplanes to hyperellipsoids: characterizing the inherent interpretability of linear and single-qubit mixed-state binary classification models — The single-qubit mixed-state model classifies data using hyperellipsoids, analogous to linear models using hyperplanes.
• Real-Time Detection of Charge Jumps in Superconducting Qubits with a Convolutional Neural Network — A DCCNN enables real-time, low-latency detection of superconducting qubit charge jumps, matching offline methods without hyperparameter tuning.
• Reproducible Reservoir Computing with Thermally Driven Superparamagnets: Controlling Temperature Sensitivity — Heterogeneous superparamagnetic nanodot patterns stabilize reservoir computing performance across wide temperature ranges, mitigating thermal sensitivity.
• The evolution of AI from image interpretation toward scientific inference in nanoparticle electron microscopy — AI is evolving electron microscopy from descriptive imaging to data-driven scientific inference and autonomous materials discovery.
• 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.
• An LLM System for Autonomous Variational Quantum Circuit Design — LLM-driven agentic systems can autonomously design high-performing quantum circuits through iterative, closed-loop workflows.

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.