Speaker
Description
In the rapidly developing field of information science,
accurately classifying quantum states is one
of the most important tasks to advance our understanding
and practical applications of quantum
technologies. Our work presents the PGM classifier,
a novel quantum-inspired classification algorithm
designed to address the complex problem of
quantum state discrimination. Unlike classical classifiers,
the PGM classifier leverages the principles
of quantum physics to deliver superior performance,
integrating this knowledge with advanced computational
techniques, making it an adaptive tool for
future quantum computing platforms.
This innovative approach has introduced a general
framework for multiclass classification that
takes advantage of quantum-inspired techniques,
namely through the use of positive operator-valued
measures (POVM) [1], (Just as the Helstrom classifier
uses a fuzzy observable associated with a
POVM, the PGM classifier takes advantage of tensor
copies of quantum representatives to improve
performance). Our research provides an in-depth
evaluation of the PGM classifier in various quantum
state scenarios, including 2, 3, 4, and 5-qubit
systems [2]. The results demonstrate that the
PGM classifier consistently outperforms standard
machine learning algorithms, especially in the complex
task of classifying separable states, an area
where classical methods often fail. With a balanced
accuracy of over 80% on 2-qubit systems, the
PGM classifier maintains its effectiveness even as
the number of qubits increases, although its performance
experiences a gradual decline.
The effectiveness of the PGM classifier is further
validated by comparisons with leading standard
classifiers (Fig.1), showing its superior ability
to handle quantum state discrimination tasks. The
graphical results highlight its effectiveness, demonstrating
its exceptional performance in distinguishing
quantum correlations that classical classifiers
fail to accurately capture.
In addition to state classification, we explore
the PGM Classifier’s application in detecting nonlocality.
Our experiments reveal its high accuracy
in identifying violations of Bell-type inequalities for
2 and 3-qubit systems, underscoring its strength in
revealing genuine quantum correlations.
This work not only presents a significant advancement in quantum state classification but also
establishes a foundation for future research in quantum
information science. The PGM Classifier’s
unique blend of quantum inspiration and practical
applicability positions it as a pivotal tool for exploring
complex quantum phenomena, making it a
valuable asset for the next generation of quantum
technologies and research. References
[1] Roberto Giuntini, Andr´es Camilo Granda Arango, Hector Freytes,
Federico Hernan Holik and Giuseppe Sergioli, Multi-class classification
based on quantum state discrimination, Fuzzy Sets and Systems,
vol. 467, p. 108509, 2023. https://doi.org/10.1016/j.fss.2023.03.012.
[2] Roberto Giuntini, Giuseppe Sergioli, Carlo Cuccu, Andr´es Camilo
Granda Arango and Federico Hernan Holik, Classification of Quantum
Correlations via Quantum-inspired Machine Learning,
