Haiyue Kang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Haiyue Kang has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Martin E. Sevior
University of Melbourne
3 shared publications - Muhammad Usman
University of Melbourne
3 shared publications - Gary J. Mooney
University of Melbourne
2 shared publications - Benjamin Harper
The University of Melbourne, Melbourne, Australia and Data 61, CSIRO, Melbourne, Australia
1 shared publication - Charles D. Hill
Silicon Quantum Computing, The University of New South Wales, Sydney, New South Wales 2052, Australia
1 shared publication - Eromanga Adermann
Technology , CSIRO , Marsfield New South Wales, 2122 , Australia , csiro.au
1 shared publication
Latest Publications
- Open Access
- 19 November 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (4) , 0450d7
Fourier analysis of quantum neural network with non-linear data embedding
Fourier analysis has become a crucial tool for understanding the expressivity of variational quantum circuit (VQC) models, as well as an important indicator of barren plateaus. While existing literature has only studied angle-embedded VQCs in a noiseless environment, here we develop the Fourier analysis of VQCs with non-linear data embedding, with particular focus on amplitude embedding, which provides a naturally compact encoding scheme. We first investigate a subtle difference in the domain of input features within amplitude embedding that leads to a distinct expressivity of the zero-frequency Fourier coefficient. Assuming that the ensemble of unitaries generated from the parameter space forms at least a unitary 2-design, we derive that, in integer-valued frequencies, the mean of the Fourier coefficients is concentrated at zero, and the variance is upper bounded by an inverse power-law function with respect to the infinity norm of the multi-dimensional frequency. A numerical evidence of exponentially decaying pattern with respect to the frequency magnitude is also established, providing a frequency-perspective of viewing VQC model expressivity and its trainability. When a noise channel with unitary Kraus operators and probabilities { p k } is taken into account, the variance is further suppressed by a factor ( ∑ k p k 2 ) Q < 1 , where Q is the number of channel instances applied. Furthermore, we demonstrate and validate the analytical results through simulations, both noiseless and noisy, and contrast its expressivity against the angle-embedded model Fourier features. We also examine the model’s performance where target functions are decomposed into non-integer frequencies, highlighting the practical utility of the approach. Our results establish a rigorous Fourier framework for amplitude-encoded VQCs, offering both theoretical guarantees on expressivity, hence trainability scaling in the frequency domain, as well as practical simulations for deployment on noisy quantum devices.
- 19 November 2026
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (4) , 0450d7
Open AccessFourier analysis of quantum neural network with non-linear data embedding
Fourier analysis has become a crucial tool for understanding the expressivity of variational quantum circuit (VQC) models, as well as an important indicator of barren plateaus. While existing literature has only studied angle-embedded VQCs in a noiseless environment, here we develop the Fourier analysis of VQCs with non-linear data embedding, with particular focus on amplitude embedding, which provides a naturally compact encoding scheme. We first investigate a subtle difference in the domain of input features within amplitude embedding that leads to a distinct expressivity of the zero-frequency Fourier coefficient. Assuming that the ensemble of unitaries generated from the parameter space forms at least a unitary 2-design, we derive that, in integer-valued frequencies, the mean of the Fourier coefficients is concentrated at zero, and the variance is upper bounded by an inverse power-law function with respect to the infinity norm of the multi-dimensional frequency. A numerical evidence of exponentially decaying pattern with respect to the frequency magnitude is also established, providing a frequency-perspective of viewing VQC model expressivity and its trainability. When a noise channel with unitary Kraus operators and probabilities { p k } is taken into account, the variance is further suppressed by a factor ( ∑ k p k 2 ) Q < 1 , where Q is the number of channel instances applied. Furthermore, we demonstrate and validate the analytical results through simulations, both noiseless and noisy, and contrast its expressivity against the angle-embedded model Fourier features. We also examine the model’s performance where target functions are decomposed into non-integer frequencies, highlighting the practical utility of the approach. Our results establish a rigorous Fourier framework for amplitude-encoded VQCs, offering both theoretical guarantees on expressivity, hence trainability scaling in the frequency domain, as well as practical simulations for deployment on noisy quantum devices.
- Open Access
- 19 November 2025
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (1)
Almost fault-tolerant quantum machine learning with drastic overhead reduction
Errors in the current generation of quantum processors pose a significant challenge towards practical-scale implementations of quantum machine learning (QML) as they lead to trainability issues arising from noise-induced barren plateaus, as well as performance degradations due to the noise accumulation in deep circuits even when QML models are free from barren plateaus. Quantum error correction (QEC) protocols are being developed to overcome hardware noise, but their extremely high spacetime overheads, mainly due to magic state distillation, make them infeasible for near-term practical implementation. This work proposes the idea of partial quantum error correction (QEC) for quantum machine learning (QML) models and ident ifies a sweet spot where distillations are omitted to significantly reduce overhead. By assuming error-corrected two-qubit Controlled- Z s (Clifford operations), we demonstrate that the QML models remain trainable even when single-qubit gates are subjected to ≈0.2% depolarizing noise, corresponding to a gate error rate of ≈0.13% under randomized benchmarking. Further analysis based on various noise models, such as phase-damping and thermal-dissipation channels at low temperature, indicates that the QML models are trainable independent of the mean angle of over-rotation, or can even be improved by thermal damping that purifies a quantum state away from depolarizations. While it may take several years to build quantum processors capable of fully fault-tolerant QML, our work proposes a resource-efficient solution for trainable and high-accuracy QML implementations in noisy environments.
- 19 November 2025
- Published by IOP Publishing in Quantum Science and Technology
- Vol. 11 (1)
Open AccessAlmost fault-tolerant quantum machine learning with drastic overhead reduction
Errors in the current generation of quantum processors pose a significant challenge towards practical-scale implementations of quantum machine learning (QML) as they lead to trainability issues arising from noise-induced barren plateaus, as well as performance degradations due to the noise accumulation in deep circuits even when QML models are free from barren plateaus. Quantum error correction (QEC) protocols are being developed to overcome hardware noise, but their extremely high spacetime overheads, mainly due to magic state distillation, make them infeasible for near-term practical implementation. This work proposes the idea of partial quantum error correction (QEC) for quantum machine learning (QML) models and ident ifies a sweet spot where distillations are omitted to significantly reduce overhead. By assuming error-corrected two-qubit Controlled- Z s (Clifford operations), we demonstrate that the QML models remain trainable even when single-qubit gates are subjected to ≈0.2% depolarizing noise, corresponding to a gate error rate of ≈0.13% under randomized benchmarking. Further analysis based on various noise models, such as phase-damping and thermal-dissipation channels at low temperature, indicates that the QML models are trainable independent of the mean angle of over-rotation, or can even be improved by thermal damping that purifies a quantum state away from depolarizations. While it may take several years to build quantum processors capable of fully fault-tolerant QML, our work proposes a resource-efficient solution for trainable and high-accuracy QML implementations in noisy environments.
- 16 July 2025
- Published by American Physical Society (APS) in Physical Review Applied
- Vol. 24 (2)
Time-adaptive single-shot crosstalk detector on a superconducting quantum computer
- 16 July 2025
- Published by American Physical Society (APS) in Physical Review Applied
- Vol. 24 (2)
Time-adaptive single-shot crosstalk detector on a superconducting quantum computer
- 24 January 2025
- Published by American Physical Society (APS) in Physical Review Applied
- Vol. 23 (1)
Teleporting two-qubit entanglement across 19 qubits on a superconducting quantum computer
- 24 January 2025
- Published by American Physical Society (APS) in Physical Review Applied
- Vol. 23 (1)
Teleporting two-qubit entanglement across 19 qubits on a superconducting quantum computer
- Open Access
- 12 August 2024
- Published by American Physical Society (APS) in Physical Review Research
- Vol. 6 (3) , 033155
Characterization of entanglement on superconducting quantum computers of up to 414 qubits
As quantum technology advances and the size of quantum computers grow, it becomes increasingly important to understand the extent of quality in the devices. As large-scale entanglement is a quantum resource crucial for achieving quantum advantage, the challenge in its generation makes it a valuable benchmark for measuring the performance of universal quantum devices. In this paper, we study entanglement in Greenberger-Horne-Zeilinger (GHZ) and graph states prepared on the range of IBM Quantum devices. We generate GHZ states and investigate their coherence times with respect to state size and dynamical decoupling techniques. A GHZ fidelity of is measured on a 32-qubit GHZ state, certifying its genuine multipartite entanglement (GME). We show a substantial improvement in GHZ decoherence rates for a seven-qubit GHZ state after implementing dynamical decoupling, and observe a linear trend in the decoherence rate of for up to qubits, confirming the absence of superdecoherence. Additionally, we prepare and characterize fully bipartite entangled native-graph states on 22 superconducting quantum devices with qubit counts as high as 414 qubits, all active qubits of the 433-qubit IBM Osprey device. Analysis of the decay of two-qubit entanglement within the prepared states shows suppression of coherent noise signals with the implementation of dynamical decoupling techniques. Additionally, we observe that the entanglement in some qubit pairs oscillates over time, which is likely caused by residual ZZ-interactions. Characterizing entanglement in native-graph states, along with detecting entanglement oscillations, can be an effective approach to low-level device benchmarking that encapsulates two-qubit error rates along with additional sources of noise, with possible applications to quantum circuit compilation. We develop several tools to automate the preparation and entanglement characterization of GHZ and graph states. In particular, a method to characterize graph state bipartite entanglement using just 36 circuits, constant with respect to the number of qubits.- 12 August 2024
- Published by American Physical Society (APS) in Physical Review Research
- Vol. 6 (3) , 033155
Open AccessCharacterization of entanglement on superconducting quantum computers of up to 414 qubits
As quantum technology advances and the size of quantum computers grow, it becomes increasingly important to understand the extent of quality in the devices. As large-scale entanglement is a quantum resource crucial for achieving quantum advantage, the challenge in its generation makes it a valuable benchmark for measuring the performance of universal quantum devices. In this paper, we study entanglement in Greenberger-Horne-Zeilinger (GHZ) and graph states prepared on the range of IBM Quantum devices. We generate GHZ states and investigate their coherence times with respect to state size and dynamical decoupling techniques. A GHZ fidelity of is measured on a 32-qubit GHZ state, certifying its genuine multipartite entanglement (GME). We show a substantial improvement in GHZ decoherence rates for a seven-qubit GHZ state after implementing dynamical decoupling, and observe a linear trend in the decoherence rate of for up to qubits, confirming the absence of superdecoherence. Additionally, we prepare and characterize fully bipartite entangled native-graph states on 22 superconducting quantum devices with qubit counts as high as 414 qubits, all active qubits of the 433-qubit IBM Osprey device. Analysis of the decay of two-qubit entanglement within the prepared states shows suppression of coherent noise signals with the implementation of dynamical decoupling techniques. Additionally, we observe that the entanglement in some qubit pairs oscillates over time, which is likely caused by residual ZZ-interactions. Characterizing entanglement in native-graph states, along with detecting entanglement oscillations, can be an effective approach to low-level device benchmarking that encapsulates two-qubit error rates along with additional sources of noise, with possible applications to quantum circuit compilation. We develop several tools to automate the preparation and entanglement characterization of GHZ and graph states. In particular, a method to characterize graph state bipartite entanglement using just 36 circuits, constant with respect to the number of qubits.