Welcome to The Daily Qubit!

Get the latest in top quantum news and research Monday through Friday, summarized for quick reading so you stay informed without missing a qubit.

Have questions, feedback, or ideas? Fill out the survey at the end of the issue or email me directly at [email protected].

And remember—friends don’t let friends miss out on the quantum era. If you enjoy The Daily Qubit, pass it along to others who’d appreciate it too.

Happy reading and onward!

Cierra

Today’s issue includes:

• A quantum-inspired convolutional neural network is used for classifying thyroid nodules in ultrasound images.

• The projected quantum kernel is used to analyze data generated by Internet-of-Things sensors.

• A comparative analysis of classical and quantum neural networks is completed to provide insights into stock price prediction accuracy.

QUANTUM APPLICATION HEADLINES

🩻 Quantum Filters Transform Thyroid Nodule Detection

APPLICATION: Researchers from Vardhaman College of Engineering propose a quantum-inspired convolutional neural network for classifying thyroid nodules in ultrasound images as benign or malignant.

SIGNIFICANCE: The thyroid gland regulates critical functions like neural activity and organ maintenance, making its health vital. Thyroid nodules, common especially in women and older adults, pose a diagnostic challenge as a subset may become cancerous. Early and accurate identification through imaging is necessary in order to receive effective treatment. Ultrasound, while non-invasive, can provide real-time analysis of nodule features like size and structure but struggles with subjectivity and variability in diagnosis. The proposed QCNN model uses quantum mechanics-inspired data processing to extract high-dimensional features from limited datasets. This innovation improves diagnostic reliability, reducing the risk of misdiagnosis, supporting early treatment of thyroid cancer, and inspiring further quantum-based solutions for diagnosis.

HOW: The QCNN uses a three-stage quantum processing pipeline to achieve its results. First, classical ultrasound image data is encoded into quantum states using rotation gates in an embedding circuit, which paints a richer representation of the image features. This encoded quantum data is then processed using random quantum circuits, which identify intricate patterns within the data that classical methods often overlook. Finally, the processed quantum states undergo decoding through a Quantum Fourier Transform to extract essential global and localized features while maintaining the structure of the input data. This output, now optimized for key patterns, is fed into a classical neural network for accurate classification of thyroid nodules.

BY THE NUMBERS:

• 98.59% accuracy — Demonstrates strong performance compared to traditional and advanced deep learning models in thyroid nodule classification.

• 0.0545 final loss value — Reflects improvement during model training, ensuring reliability and precision.

• 10% error reduction — The quantum approach outperformed conventional CNNs in handling small datasets and complex features.

📱 Enhancing IoT Insights with Projected Quantum Kernels

APPLICATION: Researchers from ICAR-CNR and the University of Calabria have proposed using the Projected Quantum Kernel to analyze data generated by internet-of-things sensors.

SIGNIFICANCE: IoT sensor data analysis is relevant to applications ranging from smart buildings to resource optimization. Existing machine learning models for this domain often rely on conventional feature engineering, which may overlook intricate relationships in high-dimensional data. The PQK approach incorporates both quantum and classical resources which simplifies data processing while preserving essential information. This effectively enables improved predictions for tasks like office occupancy detection.

HOW: The PQK framework uses a three-step process to analyze IoT data. First, classical sensor data, including metrics like CO₂ levels, illuminance, and temperature, are encoded into quantum states using a feature mapping circuit. The encoded quantum states are then processed through a quantum circuit to extract features, leveraging quantum mechanics' probabilistic nature. Finally, the quantum data is measured and projected back into classical space, where it is input into a Gaussian kernel for binary classification.

BY THE NUMBERS

• 2865 observations — The dataset comprises nearly 3000 samples of IoT sensor data collected over a month.

• 0.8368 accuracy — Achieved using PQK with entanglement-free encoding, surpassing the 0.8089 accuracy of traditional SVM models.

• 6 quantum features — The system processes six key environmental parameters, such as CO₂ and temperature, using quantum encoding to capture complex correlations.

📈 Quantum Neural Networks Revolutionize Stock Price Prediction

APPLICATION: Researchers from the University of Dodoma conducted a comparative analysis of classical and quantum neural networks to improve stock price prediction accuracy for the Cooperative Rural Development Bank and National Microfinance Bank in Tanzania.

SIGNIFICANCE: Accurate stock price predictions are important for optimizing investment decisions, particularly in developing markets where financial volatility and geopolitical factors complicate forecasting. Classical machine learning models like LSTM and MLR have shown improvements over traditional statistical methods by capturing long-term dependencies and adapting to complex relationships in financial data. The VQNN introduces quantum-inspired advantages, such as resistance to overfitting and the ability to process high-dimensional data which may lead to more sophisticated financial modeling tools, even for NISQ devices.

HOW: The study used three neural network architectures to predict low stock prices for CRDB and NMB using data from the Dar es Salaam Stock Exchange. The LSTM model relied on sequential processing with dropout layers to reduce overfitting and capture temporal dependencies in the data. The MLR model used multiple hidden layers to improve weight optimization and adaptability to complex stock price patterns. The VQNN used quantum circuits to encode financial data into quantum states, with variational parameters improving learning speed and reducing sensitivity to data sparsity. By comparing the models' performance using metrics such as the mean absolute percentage error and root mean square error, the researchers were able to single out the strengths and limitations of each.

BY THE NUMBERS: 

• 2525 data points — The dataset included stock prices for CRDB and NMB, divided into 80% training and 20% testing for sufficient analysis.

• 0.1742 RMSE — The VQNN demonstrated the lowest error rate among the models, highlighting its ability to reduce overfitting and accurately predict stock prices.

• 0.9940 R² — The MLR model achieved the highest correlation coefficient, indicating strong predictive accuracy for stock price trends.

RESEARCH HIGHLIGHTS

✅ Researchers from NIST and UCLA have developed Absorption-Emission codes for atomic and molecular quantum information systems to address noise such as spontaneous emission and Raman scattering, which diatomic molecular codes fail to correct. These A-E codes use angular momentum states to robustly encode quantum information while minimizing resource requirements, making them suitable for current and near-term quantum devices. The study demonstrates the codes' compatibility with atomic and molecular platforms with potential for improving error correction in quantum systems.

⛓️ Scientist from the University of Arizona, Cisco Quantum Labs, and the University of Massachusetts explore probabilistic entanglement between sensors in quantum networks to improve the precision of distributed sensing tasks, such as estimating global parameters. By modeling a star network, the study examines different sensing protocols and determines the conditions under which entanglement is beneficial, including thresholds for fidelity and success probabilities. Results demonstrate that incorporating entanglement distillation and optimized protocols improves sensing precision, with potential applications in quantum-enhanced metrology and distributed quantum systems.

🧪 A team from the University of Innsbruck establishes a connection between measurement-based quantum computation and Clifford quantum cellular automata as well as provides framework for understanding MBQC. By mapping MBQC processes onto the Heisenberg picture of CQCAs, the study demonstrates a practical method for constructing parameterized quantum circuits optimized for specific hardware and tasks.

NEWS QUICK BYTES

🚀 Rigetti Computing has launched its 84-qubit Ankaa-3 quantum computer with a redesigned hardware architecture that acheives high median two-qubit gate fidelities of 99.0% for iSWAP gates and 99.5% for fSim gates. The system, designed for faster, high-fidelity operations, features improved qubit coherence, advanced control technologies, and a modular architecture,.

🎁 Hermes Germany, in collaboration with D-Wave and QuantumBasel, is testing quantum annealing to optimize delivery logistics by addressing complex routing, scheduling, and constraint management challenges. By encoding logistics problems into a quantum system, this approach seeks to outperform classical methods, offering faster and more efficient solutions. The ongoing proof of concept phase compares quantum and classical systems on metrics such as route length, delivery time, and manual intervention.

🪢 Northwestern University researchers have achieved the first demonstration of quantum teleportation over fiber optic cables carrying active Internet traffic. By identifying low-interference wavelengths and employing noise-reducing filters, they successfully transmitted quantum information alongside high-speed classical data over a 30-kilometer cable.

QUANTUM MEDIA

LISTEN

Simon Borger, a physicist turned patent attorney, is interviewed by Yuval Boger. They discuss the evolving intellectual property landscape in quantum technologies. Simon shares insights on global patent trends, the role of startups versus large companies, and the challenges of filing patents for quantum innovations. They explore how to balance academic publishing with IP protection, the rising influence of European filings, and the surprising nuances of Chinese patent strategies, and much more.

THAT’S A WRAP.