Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

Quantum Computing, Communication, and Simulation

Boson sampling powered image recognition

William J. Munro, Akitada Sakurai, Aoi Hayashi, Kae Nemoto

Published Mar 19, 2025• 0 citations
Publisher

Abstract

Quantum advantage has been demonstrated in a number of physical systems with boson sampling, a well-known restricted non-universal quantum computation model uniquely applicable to photonic systems, being one of the most significant examples. The boson sampling computational processors however has struggled to find real world practical applications they can be applied too. Here we show these processors can be used to perform various image recognition and classification tasks, even for modest sized systems. More specifically we show that these random interferometer-based boson sampling processors generate the complex dynamics necessary to power quantum reservoir computing.

DOI
10.1117/12.3038939

Related papers

Image recognition powered by photonic reservoir computation

William J. Munro, Akitada Sakurai, Aoi Hayashi +2 more

Boson sampling has demonstrated quantum advantage yet struggled to find real world practical applications. We show these processors generate the complex dynamics necessary to power quantum reservoir computing and utilize it for various image recognition and classification tasks.

Quantum 2.0 Conference and Exhibition•Jan 1, 2025•1 citation

Quantum optical reservoir computing powered by boson sampling

Akitada Sakurai, Aoi Hayashi, William J. Munro +1 more

It is well known that boson sampling, a restricted non-universal quantum computation model, enables one to perform tasks that are hard to accomplish with digital computers. Boson sampling is associated with sampling the probability distribution of identical bosons passing through a random interferometer, and its quantum advantage has been demonstrated. It has, however, proven elusive to use such a model for practical applications. In this work, we show that the random interferometer powering boson sampling can be used to generate the complex dynamics necessary for quantum reservoir computing. We use these dynamics to perform various image recognition problems, illustrating the utility of the approach even for modest-size systems.

Optica Quantum•May 9, 2025•12 citations

Modular photonic quantum reservoir computation for image recognition

William J. Munro, Akitada Sakurai, Aoi Hayashi +2 more

We present a modular photonic approach to quantum reservoir computing, leveraging complex optical dynamics for efficient machine learning. Using architectures inspired by boson sampling, we show that random photonic interferometers can serve as powerful quantum reservoirs for tasks such as image classification. Modular connectivity—achieved through long-range and inter-module photonic links—enables scalable and high-performance learning while remaining compatible with integrated photonic platforms. Additionally, we propose a resource-efficient quantum neural network model that uses scale-free photonic connectivity to exploit the Hilbert space as a rich feature space. Our approach requires no optimization, supporting general-purpose use on fixed photonic hardware.

Quantum Computing, Communication, and Simulation•Mar 5, 2026•0 citations

Non-linear Boson Sampling

Nicolò Spagnolo, Daniel J. Brod, Ernesto F. Galvão +1 more

Abstract Boson Sampling is a task that is conjectured to be computationally hard for a classical computer, but which can be efficiently solved by linear-optical interferometers with Fock state inputs. Significant advances have been reported in the last few years, with demonstrations of small- and medium-scale devices, as well as implementations of variants such as Gaussian Boson Sampling. Besides the relevance of this class of computational models in the quest for unambiguous experimental demonstrations of quantum advantage, recent results have also proposed the first applications for hybrid quantum computing. Here, we introduce the adoption of non-linear photon–photon interactions in the Boson Sampling framework, and analyze the enhancement in complexity via an explicit linear-optical simulation scheme. By extending the computational expressivity of Boson Sampling, the introduction of non-linearities promises to disclose novel functionalities for this class of quantum devices. Hence, our results are expected to lead to new applications of near-term, restricted photonic quantum computers.

arXiv:2110.13788•npj Quantum Information•Jan 9, 2023•24 citations

Quantum machine learning with Adaptive Boson Sampling via post-selection

Francesco Hoch, Eugenio Caruccio, Giovanni Rodari +18 more

The implementation of large-scale universal quantum computation represents a challenging and ambitious task on the road to quantum processing of information. In recent years, an intermediate approach has been pursued to demonstrate quantum computational advantage via non-universal computational models. A relevant example for photonic platforms has been provided by the Boson Sampling paradigm and its variants, which are known to be computationally hard while requiring at the same time only the manipulation of the generated photonic resources via linear optics and detection. Beside quantum computational advantage demonstrations, a promising direction towards possibly useful applications can be found in the field of quantum machine learning, considering the currently almost unexplored intermediate scenario between non-adaptive linear optics and universal photonic quantum computation. Here, we report the experimental implementation of quantum machine learning protocols by adding adaptivity via post-selection to a Boson Sampling platform based on universal programmable photonic circuits fabricated via femtosecond laser writing. Our experimental results demonstrate that Adaptive Boson Sampling is a viable route towards dimension-enhanced quantum machine learning with linear optical devices. Extending quantum photonics’ capabilities from simple linear-optics-based schemes to universal quantum computing presents several challenges, but intermediate regimes with some degree of adaptivity might already bring practical advantages. Here, the authors experimentally emulate an adaptive Boson Sampling scheme using post-selection, and apply it to a data classification task.

arXiv:2502.20305•Nature Communications•Jan 21, 2025•34 citations

Photonic Quantum-Accelerated Machine Learning

Markus Rambach, Abhishek Roy, Alexei Gilchrist +4 more

Machine learning is widely applied in modern society, but has yet to capitalise on the unique benefits offered by quantum resources. Boson sampling -- a quantum-interference based sampling protocol -- is a resource that is classically hard to simulate and can be implemented on current quantum hardware. Here, we present a quantum accelerator for classical machine learning, using boson sampling to provide a high-dimensional quantum fingerprint for reservoir computing. We show robust performance improvements under various conditions: imperfect photon sources down to complete distinguishability; scenarios with severe class imbalances, classifying both handwritten digits and biomedical images; and sparse data, maintaining model accuracy with twenty times less training data. Crucially, we demonstrate the acceleration of our scheme on a photonic quantum processing unit, providing experimental validation that boson-sampling-enhanced learning with Fock states delivers real performance gains on actual quantum hardware.

quant-pharXiv:2512.08318•Dec 10, 2025•0 citations
Quantum advantage has been demonstrated in a number of physical systems with boson sampling, a well-known restricted non-universal quantum computation model uniquely applicable to photonic systems, being one of the most significant examples. The boson sampling computational processors however has struggled to find real world practical applications they can be applied too. Here we show these processors can be used to perform various image recognition and classification tasks, even for modest sized systems. More specifically we show that these random interferometer-based boson sampling processors generate the complex dynamics necessary to power quantum reservoir computing.
It is well known that boson sampling, a restricted non-universal quantum computation model, enables one to perform tasks that are hard to accomplish with digital computers. Boson sampling is associated with sampling the probability distribution of identical bosons passing through a random interferometer, and its quantum advantage has been demonstrated. It has, however, proven elusive to use such a model for practical applications. In this work, we show that the random interferometer powering boson sampling can be used to generate the complex dynamics necessary for quantum reservoir computing. We use these dynamics to perform various image recognition problems, illustrating the utility of the approach even for modest-size systems.
We present a modular photonic approach to quantum reservoir computing, leveraging complex optical dynamics for efficient machine learning. Using architectures inspired by boson sampling, we show that random photonic interferometers can serve as powerful quantum reservoirs for tasks such as image classification. Modular connectivity—achieved through long-range and inter-module photonic links—enables scalable and high-performance learning while remaining compatible with integrated photonic platforms. Additionally, we propose a resource-efficient quantum neural network model that uses scale-free photonic connectivity to exploit the Hilbert space as a rich feature space. Our approach requires no optimization, supporting general-purpose use on fixed photonic hardware.
Abstract Boson Sampling is a task that is conjectured to be computationally hard for a classical computer, but which can be efficiently solved by linear-optical interferometers with Fock state inputs. Significant advances have been reported in the last few years, with demonstrations of small- and medium-scale devices, as well as implementations of variants such as Gaussian Boson Sampling. Besides the relevance of this class of computational models in the quest for unambiguous experimental demonstrations of quantum advantage, recent results have also proposed the first applications for hybrid quantum computing. Here, we introduce the adoption of non-linear photon–photon interactions in the Boson Sampling framework, and analyze the enhancement in complexity via an explicit linear-optical simulation scheme. By extending the computational expressivity of Boson Sampling, the introduction of non-linearities promises to disclose novel functionalities for this class of quantum devices. Hence, our results are expected to lead to new applications of near-term, restricted photonic quantum computers.
The implementation of large-scale universal quantum computation represents a challenging and ambitious task on the road to quantum processing of information. In recent years, an intermediate approach has been pursued to demonstrate quantum computational advantage via non-universal computational models. A relevant example for photonic platforms has been provided by the Boson Sampling paradigm and its variants, which are known to be computationally hard while requiring at the same time only the manipulation of the generated photonic resources via linear optics and detection. Beside quantum computational advantage demonstrations, a promising direction towards possibly useful applications can be found in the field of quantum machine learning, considering the currently almost unexplored intermediate scenario between non-adaptive linear optics and universal photonic quantum computation. Here, we report the experimental implementation of quantum machine learning protocols by adding adaptivity via post-selection to a Boson Sampling platform based on universal programmable photonic circuits fabricated via femtosecond laser writing. Our experimental results demonstrate that Adaptive Boson Sampling is a viable route towards dimension-enhanced quantum machine learning with linear optical devices. Extending quantum photonics’ capabilities from simple linear-optics-based schemes to universal quantum computing presents several challenges, but intermediate regimes with some degree of adaptivity might already bring practical advantages. Here, the authors experimentally emulate an adaptive Boson Sampling scheme using post-selection, and apply it to a data classification task.
Machine learning is widely applied in modern society, but has yet to capitalise on the unique benefits offered by quantum resources. Boson sampling -- a quantum-interference based sampling protocol -- is a resource that is classically hard to simulate and can be implemented on current quantum hardware. Here, we present a quantum accelerator for classical machine learning, using boson sampling to provide a high-dimensional quantum fingerprint for reservoir computing. We show robust performance improvements under various conditions: imperfect photon sources down to complete distinguishability; scenarios with severe class imbalances, classifying both handwritten digits and biomedical images; and sparse data, maintaining model accuracy with twenty times less training data. Crucially, we demonstrate the acceleration of our scheme on a photonic quantum processing unit, providing experimental validation that boson-sampling-enhanced learning with Fock states delivers real performance gains on actual quantum hardware.