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

Physical Review A

Compact and classically preprocessed data-loading quantum circuit as a quantum random access memory

Bojia Duan, Chang‐Yu Hsieh

Published Jul 22, 2024• 6 citations
Publisher

Abstract

Quantum random access memory (QRAM) is an essential ingredient for many quantum algorithms, such as quantum machine learning models. The quantum resources (qubit counts and circuit depths) required to load classical data into a circuit severely prohibit many reasonably sophisticated demonstrations in the near term. To this end, we propose a method to construct a compact QRAM circuit, leveraging classical preprocessing to design as shallow a circuit with as few auxiliary qubits as possible. The underlying idea is to map data-encoded quantum states into an orthogonal set of quantum states (characterized by a family of parametrized circuits); then one can straightforwardly entangle the data qubits with the address qubits using cnot gates to realize a data-loading QRAM circuit. The nonunitary transformation to convert the orthogonalized quantum states back to the original data-encoded quantum states can be easily achieved with the help of a small set of auxiliary qubits and postselections. Numerical simulations on handwritten digits and Iris data sets are performed to assess the effectiveness of the proposed method, and we also highlight how noise influences quantum circuit performance across diverse configurations using random data sets. The proposed method for constructing a QRAM circuit can potentially facilitate the development and testing of many quantum machine learning methods in the near term.

DOI
10.1103/physreva.110.012616

Related papers

Resource-efficient simulation of noisy quantum circuits and application to network-enabled QRAM optimization

Luís Bugalho, Emmanuel Zambrini Cruzeiro, Kevin C. Chen +3 more

Abstract Giovannetti, Lloyd, and Maccone (2008) proposed a quantum random access memory (QRAM) architecture to retrieve arbitrary superpositions of N (quantum) memory cells via quantum switches and $$O(\log (N))$$ O ( log ( N ) ) address qubits. Toward physical QRAM implementations, Chen et al. (2021) recently showed that QRAM maps natively onto optically connected quantum networks with $$O(\log (N))$$ O ( log ( N ) ) overhead and built-in error detection. However, modeling QRAM on large networks has been stymied by exponentially rising classical compute requirements. Here, we address this bottleneck by: (1) introducing a resource-efficient method for simulating large-scale noisy entanglement, allowing us to evaluate hundreds and even thousands of qubits under various noise channels; and (2) analyzing Chen et al.’s network-based QRAM as an application at the scale of quantum data centers or near-term quantum internet; and (3) introducing a modified network-based QRAM architecture to improve quantum fidelity and access rate. We conclude that network-based QRAM could be built with existing or near-term technologies leveraging photonic integrated circuits and atomic or atom-like quantum memories.

arXiv:2210.13494•npj Quantum Information•Oct 20, 2023•11 citations

Design of a Superposition-Based Approximate QRAM for Noise-Tolerant Quantum Machine Learning

Sohrab Sajadimanesh, Ehsan Atoofian

Quantum Random Access Memory (QRAM) plays a pivotal role in accelerating quantum algorithms, especially in quantum machine learning. However, its implementation on noisy intermediate-scale quantum (NISQ) computers faces major challenges due to high qubit requirements and large circuit depth. These factors amplify quantum noise, limiting QRAM performance and scalability. In this work, we introduce a gate-pruning strategy that mitigates noise while preserving QRAM functionality. Our approach reduces circuit depth by eliminating small-angle rotation gates which are particularly noise-sensitive on NISQ hardware. This pruning significantly improves QRAM resilience without compromising data-loading efficiency or superposition-based querying. We further show that the pruned QRAM supports a variational quantum classifier capable of parallel training with logarithmic qubit scaling. Experiments on benchmark datasets confirm that our approach enhances noise resilience and classification accuracy, underscoring its potential for practical quantum machine learning in the NISQ era.

International Conference on High Performance Computing in Asia-Pacific Region (HPC Asia)•Jan 9, 2026•0 citations

Quantum Random Access Memory (QRAM) for Accelerating Machine Learning

Jincheng Zhang

The burgeoning field of machine learning (ML) is increasingly reliant on massive datasets for training complex models, particularly deep neural networks. However, the retrieval of data from conventional memory systems represents a significant bottleneck, limiting the scalability and efficiency of ML algorithms. This research proposes a Quantum Random Access Memory (QRAM) architecture designed to overcome this limitation by leveraging quantum mechanical principles. The core concept involves encoding data into qubits and utilizing quantum random access to achieve near-instantaneous data retrieval. This paper details a novel QRAM architecture, focusing on qubit arrangement, control mechanisms, and strategies for minimizing decoherence – a primary challenge in quantum computing. The theoretical framework explores the potential acceleration achievable through QRAM in deep neural network training, with the primary goal of dramatically reducing training times. We present a detailed mathematical model outlining the data encoding process, quantum access protocols, and the expected performance gains. The architecture incorporates error mitigation techniques and discusses the scalability of the design for increasingly complex datasets. The research highlights the transformative potential of QRAM in accelerating ML, addressing a fundamental constraint within the current landscape of computational learning.

•Sep 7, 2026•0 citations

Loading classical data

Alexander M. Dalzell, Sam McArdle, Mario Berta +10 more

This chapter covers quantum algorithmic primitives for loading classical data into a quantum algorithm. These primitives are important in many quantum algorithms, and they are especially essential for algorithms for big-data problems in the area of machine learning. We cover quantum random access memory (QRAM), an operation that allows a quantum algorithm to query a classical database in superposition. We carefully detail caveats and nuances that appear for realizing fast large-scale QRAM and what this means for algorithms that rely upon QRAM. We also cover primitives for preparing arbitrary quantum states given a list of the amplitudes stored in a classical database, and for performing a block-encoding of a matrix, given a list of its entries stored in a classical database.

Cambridge University Press eBooks•Apr 24, 2025•0 citations

Optimization of Approximate Quantum Random Access Memory for NISQ Devices

Sohrab Sajadimanesh, Jean Paul Latyr Faye, Ehsan Atoofian

An integral aspect of quantum supremacy is the implementation of quantum queries which enable quantum circuits to process quantum data. Quantum random access memory (QRAM) is a promising architecture for realizing these queries. However, implementing QRAM on contemporary quantum computers presents significant challenges due to the substantial noise it introduces. In this paper, we propose a technique that identifies gates with small rotation angles and removes them to simplify QRAM circuits, addressing the challenge of quantum noise on NISQ devices. Among various QRAM architectures, approximate QRAMs exhibit unique characteristics that make them particularly suitable for quantum neural networks (QNNs). Approximate QRAMs leverage trainable QNNs to provide approximate data for a given address that closely matches the exact data. Since QNNs are inherently resilient to errors, the approximate nature of these QRAMs enables them to achieve high accuracy. Additionally, the approximate nature of QRAMs creates opportunities for reducing the number of gates and overall circuit depth. To mitigate potential accuracy loss, we retrain the pruned QRAM to maintain the performance of the original QRAM. Our evaluation on a real quantum computer demonstrates that pruning enhances the robustness of approximate QRAMs, allowing them to function effectively on NISQ devices.

IEEE International Conference on Quantum Artificial Intelligence (QAI)•Nov 2, 2025•0 citations

A Randomized Data Structure for Efficient Quantum Encoding

Johann Maximilian Zollner, Carla Rieger, Martin Werner

Near-term quantum computers are constrained by limited qubit counts and shallow circuit depth, making them unsuitable for large-scale data processing. To address this challenge, we introduce a framework for randomized data representation that enables the efficient encoding of low-cardinality classical data for quantum circuits, as presented in [12]. Using Bloom filters and fragment-based ensemble learning, our method distributes information on multiple compact quantum circuits to ensure low hardware requirements. The data structure provides the flexibility to trade representation accuracy for representation size. Our evaluation suggests that combining lossy data compression with ensemble-based quantum learning offers a promising direction toward practical quantum machine learning on near-term devices.

•Nov 3, 2025•1 citation
Quantum random access memory (QRAM) is an essential ingredient for many quantum algorithms, such as quantum machine learning models. The quantum resources (qubit counts and circuit depths) required to load classical data into a circuit severely prohibit many reasonably sophisticated demonstrations in the near term. To this end, we propose a method to construct a compact QRAM circuit, leveraging classical preprocessing to design as shallow a circuit with as few auxiliary qubits as possible. The underlying idea is to map data-encoded quantum states into an orthogonal set of quantum states (characterized by a family of parametrized circuits); then one can straightforwardly entangle the data qubits with the address qubits using cnot gates to realize a data-loading QRAM circuit. The nonunitary transformation to convert the orthogonalized quantum states back to the original data-encoded quantum states can be easily achieved with the help of a small set of auxiliary qubits and postselections. Numerical simulations on handwritten digits and Iris data sets are performed to assess the effectiveness of the proposed method, and we also highlight how noise influences quantum circuit performance across diverse configurations using random data sets. The proposed method for constructing a QRAM circuit can potentially facilitate the development and testing of many quantum machine learning methods in the near term.
Abstract Giovannetti, Lloyd, and Maccone (2008) proposed a quantum random access memory (QRAM) architecture to retrieve arbitrary superpositions of N (quantum) memory cells via quantum switches and O(log⁡(N))O(\log (N))O(log(N)) O ( log ( N ) ) address qubits. Toward physical QRAM implementations, Chen et al. (2021) recently showed that QRAM maps natively onto optically connected quantum networks with O(log⁡(N))O(\log (N))O(log(N)) O ( log ( N ) ) overhead and built-in error detection. However, modeling QRAM on large networks has been stymied by exponentially rising classical compute requirements. Here, we address this bottleneck by: (1) introducing a resource-efficient method for simulating large-scale noisy entanglement, allowing us to evaluate hundreds and even thousands of qubits under various noise channels; and (2) analyzing Chen et al.’s network-based QRAM as an application at the scale of quantum data centers or near-term quantum internet; and (3) introducing a modified network-based QRAM architecture to improve quantum fidelity and access rate. We conclude that network-based QRAM could be built with existing or near-term technologies leveraging photonic integrated circuits and atomic or atom-like quantum memories.
Quantum Random Access Memory (QRAM) plays a pivotal role in accelerating quantum algorithms, especially in quantum machine learning. However, its implementation on noisy intermediate-scale quantum (NISQ) computers faces major challenges due to high qubit requirements and large circuit depth. These factors amplify quantum noise, limiting QRAM performance and scalability. In this work, we introduce a gate-pruning strategy that mitigates noise while preserving QRAM functionality. Our approach reduces circuit depth by eliminating small-angle rotation gates which are particularly noise-sensitive on NISQ hardware. This pruning significantly improves QRAM resilience without compromising data-loading efficiency or superposition-based querying. We further show that the pruned QRAM supports a variational quantum classifier capable of parallel training with logarithmic qubit scaling. Experiments on benchmark datasets confirm that our approach enhances noise resilience and classification accuracy, underscoring its potential for practical quantum machine learning in the NISQ era.
The burgeoning field of machine learning (ML) is increasingly reliant on massive datasets for training complex models, particularly deep neural networks. However, the retrieval of data from conventional memory systems represents a significant bottleneck, limiting the scalability and efficiency of ML algorithms. This research proposes a Quantum Random Access Memory (QRAM) architecture designed to overcome this limitation by leveraging quantum mechanical principles. The core concept involves encoding data into qubits and utilizing quantum random access to achieve near-instantaneous data retrieval. This paper details a novel QRAM architecture, focusing on qubit arrangement, control mechanisms, and strategies for minimizing decoherence – a primary challenge in quantum computing. The theoretical framework explores the potential acceleration achievable through QRAM in deep neural network training, with the primary goal of dramatically reducing training times. We present a detailed mathematical model outlining the data encoding process, quantum access protocols, and the expected performance gains. The architecture incorporates error mitigation techniques and discusses the scalability of the design for increasingly complex datasets. The research highlights the transformative potential of QRAM in accelerating ML, addressing a fundamental constraint within the current landscape of computational learning.
This chapter covers quantum algorithmic primitives for loading classical data into a quantum algorithm. These primitives are important in many quantum algorithms, and they are especially essential for algorithms for big-data problems in the area of machine learning. We cover quantum random access memory (QRAM), an operation that allows a quantum algorithm to query a classical database in superposition. We carefully detail caveats and nuances that appear for realizing fast large-scale QRAM and what this means for algorithms that rely upon QRAM. We also cover primitives for preparing arbitrary quantum states given a list of the amplitudes stored in a classical database, and for performing a block-encoding of a matrix, given a list of its entries stored in a classical database.
An integral aspect of quantum supremacy is the implementation of quantum queries which enable quantum circuits to process quantum data. Quantum random access memory (QRAM) is a promising architecture for realizing these queries. However, implementing QRAM on contemporary quantum computers presents significant challenges due to the substantial noise it introduces. In this paper, we propose a technique that identifies gates with small rotation angles and removes them to simplify QRAM circuits, addressing the challenge of quantum noise on NISQ devices. Among various QRAM architectures, approximate QRAMs exhibit unique characteristics that make them particularly suitable for quantum neural networks (QNNs). Approximate QRAMs leverage trainable QNNs to provide approximate data for a given address that closely matches the exact data. Since QNNs are inherently resilient to errors, the approximate nature of these QRAMs enables them to achieve high accuracy. Additionally, the approximate nature of QRAMs creates opportunities for reducing the number of gates and overall circuit depth. To mitigate potential accuracy loss, we retrain the pruned QRAM to maintain the performance of the original QRAM. Our evaluation on a real quantum computer demonstrates that pruning enhances the robustness of approximate QRAMs, allowing them to function effectively on NISQ devices.
Near-term quantum computers are constrained by limited qubit counts and shallow circuit depth, making them unsuitable for large-scale data processing. To address this challenge, we introduce a framework for randomized data representation that enables the efficient encoding of low-cardinality classical data for quantum circuits, as presented in [12]. Using Bloom filters and fragment-based ensemble learning, our method distributes information on multiple compact quantum circuits to ensure low hardware requirements. The data structure provides the flexibility to trade representation accuracy for representation size. Our evaluation suggests that combining lossy data compression with ensemble-based quantum learning offers a promising direction toward practical quantum machine learning on near-term devices.