arXiv Science⌕ Search
Home/Archive/arXiv

arXiv · 2402.08523

Investigating the Effect of Noise on the Training Performance of Hybrid Quantum Neural Networks

Muhammad Kashif·Emman Sychiuco·Muhammad Shafique
View original source ↗Read source document ↗Export metadata ↓

Abstract

In this paper, we conduct a comprehensively analyze the influence of different quantum noise gates, including Phase Flip, Bit Flip, Phase Damping, Amplitude Damping, and the Depolarizing Channel, on the performance of HyQNNs. Our results reveal distinct and significant effects on HyQNNs training and validation accuracies across different probabilities of noise. For instance, the Phase Flip gate introduces phase errors, and we observe that HyQNNs exhibit resilience at higher probability (p = 1.0), adapting effectively to consistent noise patterns, whereas at intermediate probabilities, the performance declines. Bit Flip errors, represented by the PauliX gate, impact HyQNNs in a similar way to that Phase Flip error gate. The HyQNNs, can adapt such kind of errors at maximum probability (p = 1.0). Unlike Phase and Bit Flip error gates, Phase Damping and Amplitude Damping gates disrupt quantum information, with HyQNNs demonstrating resilience at lower probabilities but facing challenges at higher probabilities. Amplitude Damping error gate, in particular, poses efficiency and accuracy issues at higher probabilities however with lowest probability (p = 0.1),it has the least effect and the HyQNNs, however not very effectively, but still tends to learn. The Depolarizing Channel proves most detrimental to HyQNNs performance, with limited or no training improvements. There was no training potential observed regardless of the probability of this noise gate. These findings underscore the critical need for advanced quantum error mitigation and resilience strategies in the design and training of HyQNNs, especially in environments prone to depolarizing noise. This paper quantitatively investigate that understanding the impact of quantum noise gates is essential for harnessing the full potential of quantum computing in practical applications.

Explore related subjects

quant-ph

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Muhammad Kashif, Emman Sychiuco, Muhammad Shafique. 2024-09-19. Investigating the Effect of Noise on the Training Performance of Hybrid Quantum Neural Networks. https://doi.org/10.1109/ijcnn60899.2024.10651363

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Designing a Hybrid Digital / Analog Quantum Physics Emulator as Open Hardware

Existing approaches to emulating quantum computing algorithms using classical electronic hardware are limited by exponential scaling limitations in space, such as circuit size, or time, such as runtime or bandwidth. We introduce a scheme for representing quantum information using analog signals that lessens the bandwidth limitation problem (in certain regimes) seen in existing approaches [1, 2] by taking full advantage of the ability of analog signals to encode information using RMS voltage as well as frequency and phase. We introduce the mathematical framework for this representation, which separates the information relevant for measurement in the computational basis from information that is not relevant to it. We introduce circuits that take advantage of this separation of concerns to achieve simplifications, for working with quantum information in this representation. We argue that it is comparatively very inexpensive (as low as ~$5.00 / qubit) to outmatch the computing capabilities of existing FPGA based emulators [3], though scaling beyond tens of qubits is still impractical due to constraints of analog hardware module precision. However, our approach opens the door to a new avenue by which classical emulators can hope to improve: by improving on analog electronic circuit performance.

quant-ph↗
arXiv2026

Measuring a Quantum Measure Exceeding Unity

The history based formalism known as Quantum Measure Theory (QMT) generalizes the concept of probability-measure so as to incorporate quantum interference. The resulting quantum measure $μ$ is defined for arbitrary events (sets of histories), not just for observables at a fixed moment of time. Thanks to interference effects, $μ$ can exceed unity, exhibiting its non-classical nature in a particularly striking manner. Here, in an optical experiment, we illustrate an ancilla based filtering scheme that gives operational meaning to the quantum measure. For a specific photonic event $E$, we report a measured value of $μ(E)=1.172^{+0.013}_{-0.019}$, which within errors agrees with the theoretical value of $5/4$, while exceeding the maximum value permissible for a classical probability (namely $1$) by $13.32$ upper or $8.89$ lower percentile widths. The directly observed quantity is an ordinary detector probability $p_D\le 1$ (or, with laser light, an equivalent power ratio); the value $μ(E)>1$ is inferred via the calibrated relation $μ(E)=2p_D$ for our filter. If an unconventional theoretical concept is to play a role in meeting the foundational challenges of quantum theory, it seems important to bring it into contact with experiment as much as possible. Our experiment does this for the quantum measure.

quant-ph↗
arXiv2026

Complexity Theory for Quantum Promise Problems

We begin by establishing structural results for several fundamental quantum complexity classes: p/mBQP, p/mQ(C)MA, $\text{p/mQSZK}_{\text{hv}}$, p/mQIP, p/mBQP/qpoly, p/mBQP/poly, and p/mPSPACE. This includes identifying complete problems, as well as proving containment and separation results among these classes. Here, p/mC denotes the corresponding quantum promise complexity class with pure (p) or mixed (m) quantum input states for any classical complexity class C. Surprisingly, our findings uncover relationships that diverge from their classical analogues -- specifically, we show unconditionally that p/mQIP$\neq$p/mPSPACE and p/mBQP/qpoly$\neq$p/mBQP/poly. This starkly contrasts the classical setting, where QIP$=$PSPACE and separations such as BQP/qpoly$\neq$BQP/poly are only known relative to oracles. More interestingly, these separation results further connected to the topic of for both quantum property testing and unitary synthesis. This new framework has numerous applications in quantum cryptography, particularly in the contexts of Microcrypt. We provide a better characterization of its primitives; for example, we show that OWSG and PRS can be broken by a p/mQCMA oracle, leading to a natural quantum analogue of Impagliazzo's five worlds by substituting the classical complexity classes in Pessiland, Heuristica, and Algorithmica with mBQP and mQCMA. Moreover, we establish the relativization barrier for proving the existence of EFI, noting that no such barrier currently exists within traditional complexity theory.

quant-ph↗