shenzhen, May 16, 2025 (GLOBE NEWSWIRE) -- Shenzhen, May. 16, 2025––MicroAlgo Inc. (the "Company" or "MicroAlgo") (
NASDAQ: MLGO), today announced the development of a novel quantum entanglement-based training algorithm — the Entanglement-Assisted Training Algorithm for Supervised Quantum Classifiers. They also introduced a cost function based on Bell inequalities, enabling the simultaneous encoding of errors from multiple training samples. This breakthrough surpasses the capability limits of traditional algorithms, offering an efficient and widely applicable solution for supervised quantum classifiers.The core of MicroAlgo's entanglement-assisted training algorithm for supervised quantum classifiers lies in leveraging quantum entanglement to construct a model capable of simultaneously operating on multiple training samples and their corresponding labels. Unlike traditional machine learning methods, quantum classifiers can not only process information from individual samples but also perform parallel processing of multiple samples in quantum states, thereby significantly enhancing training efficiency.The algorithm represents multiple training samples as qubit vectors using quantum superposition, and encodes their label information into quantum states through quantum gate operations. Due to the entangled relationships between qubits, the classifier can simultaneously operate on multiple samples at once. This characteristic breaks away from the conventional sample-by-sample processing paradigm, greatly improving both training speed and classification performance.Furthermore, the algorithm introduces a cost function based on Bell inequalities—an important theorem in quantum mechanics that highlights the distinction between quantum entanglement and classical information processing. By encoding classification errors of multiple samples simultaneously into the cost function, the optimization process is no longer limited to individual sample errors but instead considers the collective performance of multiple samples. This approach overcomes the local optimization issues common in traditional algorithms and significantly enhances classification accuracy.The implementation of MicroAlgo's entanglement-assisted training algorithm for supervised quantum classifiers relies on several core components of current quantum computing technology: qubits, quantum gate operations, and quantum measurement. With these fundamental building blocks, the algorithm can efficiently process input data on a quantum computer.Representation and Initialization of Qubits: at the initial stage of the algorithm, the input training samples are transformed into qubits. Each training sample corresponds to one or more qubits, which are initialized into specific quantum states. To enable entanglement, entangling operations are performed between multiple qubits so that they can collaboratively process sample data in the subsequent steps.Construction of Quantum Entanglement: quantum entanglement is one of the core features of quantum computing. In this algorithm, training samples are arranged into an entangled state, meaning that information between samples is shared and processed through entanglement. This not only improves data processing efficiency but also accelerates convergence during the training process.Application of Bell Inequalities and Cost Function Optimization: a key application of quantum entanglement is in the use of Bell inequalities. In the algorithm, Bell inequalities are employed to construct the cost function, with the objective of minimizing classification errors. Unlike traditional methods, this cost function simultaneously accounts for errors from multiple samples, allowing the optimization process to focus on the collective performance of all samples rather than optimizing on a per-sample basis. Through rapid quantum algorithmic computation, the cost function can be efficiently minimized to achieve optimal classification results.Interpretation and Output of Classification Results: finally, the algorithm outputs the classification results through quantum measurement. In binary classification tasks, the input training samples are divided into two categories, while in multi-class tasks, they are assigned to multiple classes. The advantage of quantum computing lies in its parallel processing capability, enabling the system to complete complex classification tasks in a significantly shorter amount of time.The greatest advantage of this technology lies in its ability to leverage the unique properties of quantum entanglement to parallelize the training process across multiple training samples. This not only accelerates the training speed but also effectively enhances classification accuracy. Especially in problems involving large datasets, traditional methods often face computational bottlenecks, whereas quantum computing can easily overcome these limitations.In addition, the cost function based on Bell's inequality is theoretically more robust than traditional error minimization methods. It can simultaneously handle the errors of multiple training samples, thereby avoiding the local optimum problems that may occur in conventional approaches. This makes the supervised quantum classifier particularly effective in complex classification tasks.However, quantum computing still faces many challenges. For instance, the stability and computational scale of quantum computers remain limiting factors. The number of qubits and their error rates can both impact the practical performance of the algorithms. Therefore, how to implement efficient algorithms on existing quantum computing platforms remains a technical hurdle that needs further breakthroughs.With the continuous advancement of quantum computing technology, quantum machine learning is bound to become a key direction for future technological innovation. The entanglement-assisted training algorithm of the MicroAlgo supervised quantum classifier opens up new possibilities in this field. By integrating quantum entanglement with traditional classification algorithms, this technology demonstrates great potential in improving training efficiency and enhancing classification accuracy. Although quantum computing still faces numerous challenges, with ongoing progress in hardware and deepening theoretical research, we have every reason to believe that quantum computing will bring about a revolution in the field of machine learning. In the future, quantum classifiers may not be limited to traditional binary classification tasks—they could potentially exhibit unparalleled advantages in even more complex domains.
