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

Genetic and Evolutionary Computation Conference Companion (GECCO Companion)

Quantum Enhancements for AlphaZero

James Chao, Ramiro Rodriguez, Sean Crowe

Published Jul 15, 2023• 3 citations
Publisher

Abstract

Reinforcement learning algorithms including AlphaZero are powerful artificial intelligence (AI) algorithms, but are known to be resource intensive and unable to train within a reasonable budget. Speeding up learning would be valuable to further expand application areas for real world problems. In this paper, we investigate two quantum computing methods to enhance AlphaZero, with the goal of speeding up training. We evaluate the results by playing the board game Othello, an adversarial multi-agent turn based game similar to the game Go. First, parameterized quantum circuits (PQC) have been shown to train hybrid quantum-classical reinforcement learning agents in standard benchmark environments. With this inspiration, we replace the classical neural network with a PQC quantum neural network (QNN) in the AlphaZero architecture. Second, tensor-network quantum circuits have been used to extract important features for convolutional neural networks (CNNs) in image classification tasks. Using this as inspiration, we use a tree tensor network (TTN) to extract features from the Othello game board, generating a new set of feature vectors for a classical neural network to estimate the policy and value. Results show both novel methods converge to master the game and achieve the same level of play compared to the classical AlphaZero agent.

DOI
10.1145/3583133.3596302

Related papers

Quantum agents in the Gym: a variational quantum algorithm for deep Q-learning

Andrea Skolik, Sofiène Jerbi, Vedran Dunjko

Quantum machine learning (QML) has been identified as one of the key fields that could reap advantages from near-term quantum devices, next to optimization and quantum chemistry. Research in this area has focused primarily on variational quantum algorithms (VQAs), and several proposals to enhance supervised, unsupervised and reinforcement learning (RL) algorithms with VQAs have been put forward. Out of the three, RL is the least studied and it is still an open question whether VQAs can be competitive with state-of-the-art classical algorithms based on neural networks (NNs) even on simple benchmark tasks. In this work, we introduce a training method for parametrized quantum circuits (PQCs) that can be used to solve RL tasks for discrete and continuous state spaces based on the deep Q-learning algorithm. We investigate which architectural choices for quantum Q-learning agents are most important for successfully solving certain types of environments by performing ablation studies for a number of different data encoding and readout strategies. We provide insight into why the performance of a VQA-based Q-learning algorithm crucially depends on the observables of the quantum model and show how to choose suitable observables based on the learning task at hand. To compare our model against the classical DQN algorithm, we perform an extensive hyperparameter search of PQCs and NNs with varying numbers of parameters. We confirm that similar to results in classical literature, the architectural choices and hyperparameters contribute more to the agents' success in a RL setting than the number of parameters used in the model. Finally, we show when recent separation results between classical and quantum agents for policy gradient RL can be extended to inferring optimal Q-values in restricted families of environments.

arXiv:2103.15084•Quantum•May 24, 2022•197 citations

A quantum-classical reinforcement learning model to play Atari games

Dominik Freinberger, Julian Lemmel, Radu Grosu +1 more

Recent advances in reinforcement learning have demonstrated the potential of quantum learning models based on parametrized quantum circuits as an alternative to deep learning models. On the one hand, these findings have shown the ultimate exponential speed-ups in learning that full-blown quantum models can offer in certain -- artificially constructed -- environments. On the other hand, they have demonstrated the ability of experimentally accessible PQCs to solve OpenAI Gym benchmarking tasks. However, it remains an open question whether these near-term QRL techniques can be successfully applied to more complex problems exhibiting high-dimensional observation spaces. In this work, we bridge this gap and present a hybrid model combining a PQC with classical feature encoding and post-processing layers that is capable of tackling Atari games. A classical model, subjected to architectural restrictions similar to those present in the hybrid model is constructed to serve as a reference. Our numerical investigation demonstrates that the proposed hybrid model is capable of solving the Pong environment and achieving scores comparable to the classical reference in Breakout. Furthermore, our findings shed light on important hyperparameter settings and design choices that impact the interplay of the quantum and classical components. This work contributes to the understanding of near-term quantum learning models and makes an important step towards their deployment in real-world RL scenarios.

arXiv:2412.08725•Dec 11, 2024•2 citations

QcNet: A Hybrid Quantum-Classical Neural Network for Multi-Class Image Classification

Sakthi Prabhu Gunasekar, Chatterjee Subarna

Quantum Computing is a very promising paradigm given the ability that Quantum brings in terms of recognizing patterns which their classical counterparts very likely fail at.The intersection of Quantum Computing and Machine Learning gave rise to Quantum Machine Learning, which brings the power of Quantum mechanics to accelerate the progress of Machine Learning. However, Quantum Machine Learning still has its own set of challenges, given the state of Quantum Hardware currently. Here we propose QcNet: A Hybrid Quantum-Classical Neural Network which combines a tensor network with a parameterized quantum circuit. Here we modify the Quantum Neural Network by placing the parametrized quantum circuit convolutional layer at the top followed by some classical CNN layers to make it more feasible for the near term NISQ era and to make it scale to 43 classes. The Quantum Convolutional Layer extracts the higher level features and passes it to the classical layers for further processing. Our model is an end-to-end solution to train a quantum-style model over a multi-class supervised learning problem. This model pushes the limits beyond binary or few class classification and successfully performs well over multiclass classification tasks over 43 classes of the GTSRB dataset. Also it is highly flexible and can be adopted well in the NISQ era, considering the amount of quantum resources available.

International Conference on Inventive Research in Computing Applications (ICIRCA)•Jun 25, 2025•1 citation

Quantum Circuit Structure Optimization for Quantum Reinforcement Learning

Seok Bin Son, Joongheon Kim

Reinforcement learning (RL) enables agents to learn optimal policies through environmental interaction. However, RL suffers from reduced learning efficiency due to the curse of dimensionality in high-dimensional spaces. Quantum reinforcement learning (QRL) addresses this issue by leveraging superposition and entanglement in quantum computing, allowing efficient handling of high-dimensional problems with fewer resources. QRL combines quantum neural networks (QNNs) with RL, where the parameterized quantum circuit (PQC) acts as the core computational module. The PQC performs linear and nonlinear transformations through gate operations, similar to hidden layers in classical neural networks. Previous QRL studies, however, have used fixed PQC structures based on empirical intuition without verifying their optimality. This paper proposes a QRLNAS algorithm that integrates quantum neural architecture search (QNAS) to optimize PQC structures within QRL. Experiments demonstrate that QRL-NAS achieves higher rewards than QRL with fixed circuits, validating its effectiveness and practical utility.

IEEE International Conference on Quantum Computing and Engineering (QCE)•Aug 30, 2025•0 citations

Quantum Circuit Structure Optimization for Quantum Reinforcement Learning

Seok Bin Son, Joongheon Kim

Reinforcement learning (RL) enables agents to learn optimal policies through environmental interaction. However, RL suffers from reduced learning efficiency due to the curse of dimensionality in high-dimensional spaces. Quantum reinforcement learning (QRL) addresses this issue by leveraging superposition and entanglement in quantum computing, allowing efficient handling of high-dimensional problems with fewer resources. QRL combines quantum neural networks (QNNs) with RL, where the parameterized quantum circuit (PQC) acts as the core computational module. The PQC performs linear and nonlinear transformations through gate operations, similar to hidden layers in classical neural networks. Previous QRL studies, however, have used fixed PQC structures based on empirical intuition without verifying their optimality. This paper proposes a QRL-NAS algorithm that integrates quantum neural architecture search (QNAS) to optimize PQC structures within QRL. Experiments demonstrate that QRL-NAS achieves higher rewards than QRL with fixed circuits, validating its effectiveness and practical utility.

arXiv:2507.00589•Jul 1, 2025•0 citations

Exploring the Synergy Between Neuro-Inspired Algorithms and Quantum Computing in Machine Learning

G Nithya, Praveen Kumar R, V. Dineshbabu +2 more

The integration of neuro-inspired algorithms and quantum computing in machine learning presents a promising frontier for addressing complex computational challenges in modern AI. Neuro-inspired algorithms, such as artificial neural networks (ANNs), deep learning (DL), and spiking neural networks (SNNs), have demonstrated impressive performance in various domains, including image recognition, natural language processing, and autonomous systems. This research explores the synergy between neuro-inspired algorithms and quantum computing, focusing on how quantum-enhanced machine learning models can accelerate training and inference processes in neuro-inspired systems. Quantum neural networks (QNNs) leverage quantum principles, such as superposition and entanglement, to represent and manipulate data in ways that classical systems cannot. By combining quantum computing's parallelism with the flexibility and learning capability of neuro-inspired algorithms, the proposed hybrid models can provide exponential speedups in tasks involving large-scale data processing and optimization. To evaluate the performance of these hybrid models, experiments were conducted using a quantum-enhanced deep learning model applied to image classification and a neuro-inspired algorithm augmented by quantum optimization techniques for optimization tasks. The quantum-enhanced deep learning model achieved a 45% reduction in training time compared to classical deep learning models while maintaining similar accuracy levels. These findings highlight the significant potential of combining quantum computing with neuro-inspired algorithms, opening new avenues for faster, more efficient machine learning models capable of solving previously unsolvable problems. The synergy between these two domains could lead to breakthroughs in areas like artificial general intelligence (AGI), drug discovery, and autonomous systems, where large-scale optimization and pattern recognition are critical.

International Journal of Computational and Experimental Science and Engineering•Jun 14, 2025•5 citations
Reinforcement learning algorithms including AlphaZero are powerful artificial intelligence (AI) algorithms, but are known to be resource intensive and unable to train within a reasonable budget. Speeding up learning would be valuable to further expand application areas for real world problems. In this paper, we investigate two quantum computing methods to enhance AlphaZero, with the goal of speeding up training. We evaluate the results by playing the board game Othello, an adversarial multi-agent turn based game similar to the game Go. First, parameterized quantum circuits (PQC) have been shown to train hybrid quantum-classical reinforcement learning agents in standard benchmark environments. With this inspiration, we replace the classical neural network with a PQC quantum neural network (QNN) in the AlphaZero architecture. Second, tensor-network quantum circuits have been used to extract important features for convolutional neural networks (CNNs) in image classification tasks. Using this as inspiration, we use a tree tensor network (TTN) to extract features from the Othello game board, generating a new set of feature vectors for a classical neural network to estimate the policy and value. Results show both novel methods converge to master the game and achieve the same level of play compared to the classical AlphaZero agent.
Quantum machine learning (QML) has been identified as one of the key fields that could reap advantages from near-term quantum devices, next to optimization and quantum chemistry. Research in this area has focused primarily on variational quantum algorithms (VQAs), and several proposals to enhance supervised, unsupervised and reinforcement learning (RL) algorithms with VQAs have been put forward. Out of the three, RL is the least studied and it is still an open question whether VQAs can be competitive with state-of-the-art classical algorithms based on neural networks (NNs) even on simple benchmark tasks. In this work, we introduce a training method for parametrized quantum circuits (PQCs) that can be used to solve RL tasks for discrete and continuous state spaces based on the deep Q-learning algorithm. We investigate which architectural choices for quantum Q-learning agents are most important for successfully solving certain types of environments by performing ablation studies for a number of different data encoding and readout strategies. We provide insight into why the performance of a VQA-based Q-learning algorithm crucially depends on the observables of the quantum model and show how to choose suitable observables based on the learning task at hand. To compare our model against the classical DQN algorithm, we perform an extensive hyperparameter search of PQCs and NNs with varying numbers of parameters. We confirm that similar to results in classical literature, the architectural choices and hyperparameters contribute more to the agents' success in a RL setting than the number of parameters used in the model. Finally, we show when recent separation results between classical and quantum agents for policy gradient RL can be extended to inferring optimal Q-values in restricted families of environments.
Recent advances in reinforcement learning have demonstrated the potential of quantum learning models based on parametrized quantum circuits as an alternative to deep learning models. On the one hand, these findings have shown the ultimate exponential speed-ups in learning that full-blown quantum models can offer in certain -- artificially constructed -- environments. On the other hand, they have demonstrated the ability of experimentally accessible PQCs to solve OpenAI Gym benchmarking tasks. However, it remains an open question whether these near-term QRL techniques can be successfully applied to more complex problems exhibiting high-dimensional observation spaces. In this work, we bridge this gap and present a hybrid model combining a PQC with classical feature encoding and post-processing layers that is capable of tackling Atari games. A classical model, subjected to architectural restrictions similar to those present in the hybrid model is constructed to serve as a reference. Our numerical investigation demonstrates that the proposed hybrid model is capable of solving the Pong environment and achieving scores comparable to the classical reference in Breakout. Furthermore, our findings shed light on important hyperparameter settings and design choices that impact the interplay of the quantum and classical components. This work contributes to the understanding of near-term quantum learning models and makes an important step towards their deployment in real-world RL scenarios.
Quantum Computing is a very promising paradigm given the ability that Quantum brings in terms of recognizing patterns which their classical counterparts very likely fail at.The intersection of Quantum Computing and Machine Learning gave rise to Quantum Machine Learning, which brings the power of Quantum mechanics to accelerate the progress of Machine Learning. However, Quantum Machine Learning still has its own set of challenges, given the state of Quantum Hardware currently. Here we propose QcNet: A Hybrid Quantum-Classical Neural Network which combines a tensor network with a parameterized quantum circuit. Here we modify the Quantum Neural Network by placing the parametrized quantum circuit convolutional layer at the top followed by some classical CNN layers to make it more feasible for the near term NISQ era and to make it scale to 43 classes. The Quantum Convolutional Layer extracts the higher level features and passes it to the classical layers for further processing. Our model is an end-to-end solution to train a quantum-style model over a multi-class supervised learning problem. This model pushes the limits beyond binary or few class classification and successfully performs well over multiclass classification tasks over 43 classes of the GTSRB dataset. Also it is highly flexible and can be adopted well in the NISQ era, considering the amount of quantum resources available.
Reinforcement learning (RL) enables agents to learn optimal policies through environmental interaction. However, RL suffers from reduced learning efficiency due to the curse of dimensionality in high-dimensional spaces. Quantum reinforcement learning (QRL) addresses this issue by leveraging superposition and entanglement in quantum computing, allowing efficient handling of high-dimensional problems with fewer resources. QRL combines quantum neural networks (QNNs) with RL, where the parameterized quantum circuit (PQC) acts as the core computational module. The PQC performs linear and nonlinear transformations through gate operations, similar to hidden layers in classical neural networks. Previous QRL studies, however, have used fixed PQC structures based on empirical intuition without verifying their optimality. This paper proposes a QRLNAS algorithm that integrates quantum neural architecture search (QNAS) to optimize PQC structures within QRL. Experiments demonstrate that QRL-NAS achieves higher rewards than QRL with fixed circuits, validating its effectiveness and practical utility.
Reinforcement learning (RL) enables agents to learn optimal policies through environmental interaction. However, RL suffers from reduced learning efficiency due to the curse of dimensionality in high-dimensional spaces. Quantum reinforcement learning (QRL) addresses this issue by leveraging superposition and entanglement in quantum computing, allowing efficient handling of high-dimensional problems with fewer resources. QRL combines quantum neural networks (QNNs) with RL, where the parameterized quantum circuit (PQC) acts as the core computational module. The PQC performs linear and nonlinear transformations through gate operations, similar to hidden layers in classical neural networks. Previous QRL studies, however, have used fixed PQC structures based on empirical intuition without verifying their optimality. This paper proposes a QRL-NAS algorithm that integrates quantum neural architecture search (QNAS) to optimize PQC structures within QRL. Experiments demonstrate that QRL-NAS achieves higher rewards than QRL with fixed circuits, validating its effectiveness and practical utility.
The integration of neuro-inspired algorithms and quantum computing in machine learning presents a promising frontier for addressing complex computational challenges in modern AI. Neuro-inspired algorithms, such as artificial neural networks (ANNs), deep learning (DL), and spiking neural networks (SNNs), have demonstrated impressive performance in various domains, including image recognition, natural language processing, and autonomous systems. This research explores the synergy between neuro-inspired algorithms and quantum computing, focusing on how quantum-enhanced machine learning models can accelerate training and inference processes in neuro-inspired systems. Quantum neural networks (QNNs) leverage quantum principles, such as superposition and entanglement, to represent and manipulate data in ways that classical systems cannot. By combining quantum computing's parallelism with the flexibility and learning capability of neuro-inspired algorithms, the proposed hybrid models can provide exponential speedups in tasks involving large-scale data processing and optimization. To evaluate the performance of these hybrid models, experiments were conducted using a quantum-enhanced deep learning model applied to image classification and a neuro-inspired algorithm augmented by quantum optimization techniques for optimization tasks. The quantum-enhanced deep learning model achieved a 45% reduction in training time compared to classical deep learning models while maintaining similar accuracy levels. These findings highlight the significant potential of combining quantum computing with neuro-inspired algorithms, opening new avenues for faster, more efficient machine learning models capable of solving previously unsolvable problems. The synergy between these two domains could lead to breakthroughs in areas like artificial general intelligence (AGI), drug discovery, and autonomous systems, where large-scale optimization and pattern recognition are critical.