Adversarial Poisoning Attack on Quantum Machine Learning Models

With the growing interest in Quantum Machine Learning (QML) and the increasing availability of quantum computers through cloud providers, addressing the potential security risks associated with QML has become an urgent priority. One key concern in the QML domain is the threat of data poisoning attacks in the current quantum cloud setting. Adversarial access to training data could severely compromise the integrity and availability of QML models. Classical data poisoning techniques require significant knowledge and training to generate poisoned data, and lack noise resilience, making them ineffective for QML models in the Noisy Intermediate Scale Quantum (NISQ) era. In this work, we first propose a simple yet effective technique to measure intra-class encoder state similarity (ESS) by analyzing the outputs of encoding circuits. Leveraging this approach, we introduce a quantum indiscriminate data poisoning attack, QUID. Through extensive experiments conducted in both noiseless and noisy environments (e.g., IBM\_Brisbane's noise), across various architectures and datasets, QUID achieves up to $92\%$ accuracy degradation in model performance compared to baseline models and up to $75\%$ accuracy degradation compared to random label-flipping. We also tested QUID against state-of-the-art classical defenses, with accuracy degradation still exceeding $50\%$, demonstrating its effectiveness. This work represents the first attempt to reevaluate data poisoning attacks in the context of QML.

AI-driven Reverse Engineering of QML Models

Quantum machine learning (QML) is a rapidly emerging area of research, driven by the capabilities of Noisy Intermediate-Scale Quantum (NISQ) devices. With the progress in the research of QML models, there is a rise in third-party quantum cloud services to cater to the increasing demand for resources. New security concerns surface, specifically regarding the protection of intellectual property (IP) from untrustworthy service providers. One of the most pressing risks is the potential for reverse engineering (RE) by malicious actors who may steal proprietary quantum IPs such as trained parameters and QML architecture, modify them to remove additional watermarks or signatures and re-transpile them for other quantum hardware. Prior work presents a brute force approach to RE the QML parameters which takes exponential time overhead. In this paper, we introduce an autoencoder-based approach to extract the parameters from transpiled QML models deployed on untrusted third-party vendors. We experiment on multi-qubit classifiers and note that they can be reverse-engineered under restricted conditions with a mean error of order 10^-1. The amount of time taken to prepare the dataset and train the model to reverse engineer the QML circuit being of the order 10^3 seconds (which is 10^2x better than the previously reported value for 4-layered 4-qubit classifiers) makes the threat of RE highly potent, underscoring the need for continued development of effective defenses.

Security Concerns in Quantum Machine Learning as a Service

Quantum machine learning (QML) is a category of algorithms that employ variational quantum circuits (VQCs) to tackle machine learning tasks. Recent discoveries have shown that QML models can effectively generalize from limited training data samples. This capability has sparked increased interest in deploying these models to address practical, real-world challenges, resulting in the emergence of Quantum Machine Learning as a Service (QMLaaS). QMLaaS represents a hybrid model that utilizes both classical and quantum computing resources. Classical computers play a crucial role in this setup, handling initial pre-processing and subsequent post-processing of data to compensate for the current limitations of quantum hardware. Since this is a new area, very little work exists to paint the whole picture of QMLaaS in the context of known security threats in the domain of classical and quantum machine learning. This SoK paper is aimed to bridge this gap by outlining the complete QMLaaS workflow, which encompasses both the training and inference phases and highlighting significant security concerns involving untrusted classical or quantum providers. QML models contain several sensitive assets, such as the model architecture, training/testing data, encoding techniques, and trained parameters. Unauthorized access to these components could compromise the model's integrity and lead to intellectual property (IP) theft. We pinpoint the critical security issues that must be considered to pave the way for a secure QMLaaS deployment.

The Quantum Imitation Game: Reverse Engineering of Quantum Machine Learning Models

Quantum Machine Learning (QML) amalgamates quantum computing paradigms with machine learning models, providing significant prospects for solving complex problems. However, with the expansion of numerous third-party vendors in the Noisy Intermediate-Scale Quantum (NISQ) era of quantum computing, the security of QML models is of prime importance, particularly against reverse engineering, which could expose trained parameters and algorithms of the models. We assume the untrusted quantum cloud provider is an adversary having white-box access to the transpiled user-designed trained QML model during inference. Reverse engineering (RE) to extract the pre-transpiled QML circuit will enable re-transpilation and usage of the model for various hardware with completely different native gate sets and even different qubit technology. Such flexibility may not be obtained from the transpiled circuit which is tied to a particular hardware and qubit technology. The information about the number of parameters, and optimized values can allow further training of the QML model to alter the QML model, tamper with the watermark, and/or embed their own watermark or refine the model for other purposes. In this first effort to investigate the RE of QML circuits, we perform RE and compare the training accuracy of original and reverse-engineered Quantum Neural Networks (QNNs) of various sizes. We note that multi-qubit classifiers can be reverse-engineered under specific conditions with a mean error of order 1e-2 in a reasonable time. We also propose adding dummy fixed parametric gates in the QML models to increase the RE overhead for defense. For instance, adding 2 dummy qubits and 2 layers increases the overhead by ~1.76 times for a classifier with 2 qubits and 3 layers with a performance overhead of less than 9%. We note that RE is a very powerful attack model which warrants further efforts on defenses.

STIQ: Safeguarding Training and Inferencing of Quantum Neural Networks from Untrusted Cloud

The high expenses imposed by current quantum cloud providers, coupled with the escalating need for quantum resources, may incentivize the emergence of cheaper cloud-based quantum services from potentially untrusted providers. Deploying or hosting quantum models, such as Quantum Neural Networks (QNNs), on these untrusted platforms introduces a myriad of security concerns, with the most critical one being model theft. This vulnerability stems from the cloud provider's full access to these circuits during training and/or inference. In this work, we introduce STIQ, a novel ensemble-based strategy designed to safeguard QNNs against such cloud-based adversaries. Our method innovatively trains two distinct QNNs concurrently, hosting them on same or different platforms, in a manner that each network yields obfuscated outputs rendering the individual QNNs ineffective for adversaries operating within cloud environments. However, when these outputs are combined locally (using an aggregate function), they reveal the correct result. Through extensive experiments across various QNNs and datasets, our technique has proven to effectively masks the accuracy and losses of the individually hosted models by upto 76\%, albeit at the expense of $\leq 2\times$ increase in the total computational overhead. This trade-off, however, is a small price to pay for the enhanced security and integrity of QNNs in a cloud-based environment prone to untrusted adversaries. We also demonstrated STIQ's practical application by evaluating it on real 127-qubit IBM\_Sherbrooke hardware, showing that STIQ achieves up to 60\% obfuscation, with combined performance comparable to an unobfuscated model.

Guardians of the Quantum GAN

Quantum Generative Adversarial Networks (qGANs) are at the forefront of image-generating quantum machine learning models. To accommodate the growing demand for Noisy Intermediate-Scale Quantum (NISQ) devices to train and infer quantum machine learning models, the number of third-party vendors offering quantum hardware as a service is expected to rise. This expansion introduces the risk of untrusted vendors potentially stealing proprietary information from the quantum machine learning models. To address this concern we propose a novel watermarking technique that exploits the noise signature embedded during the training phase of qGANs as a non-invasive watermark. The watermark is identifiable in the images generated by the qGAN allowing us to trace the specific quantum hardware used during training hence providing strong proof of ownership. To further enhance the security robustness, we propose the training of qGANs on a sequence of multiple quantum hardware, embedding a complex watermark comprising the noise signatures of all the training hardware that is difficult for adversaries to replicate. We also develop a machine learning classifier to extract this watermark robustly, thereby identifying the training hardware (or the suite of hardware) from the images generated by the qGAN validating the authenticity of the model. We note that the watermark signature is robust against inferencing on hardware different than the hardware that was used for training. We obtain watermark extraction accuracy of 100% and ~90% for training the qGAN on individual and multiple quantum hardware setups (and inferencing on different hardware), respectively. Since parameter evolution during training is strongly modulated by quantum noise, the proposed watermark can be extended to other quantum machine learning models as well.

Application of Quantum Tensor Networks for Protein Classification

We show that protein sequences can be thought of as sentences in natural language processing and can be parsed using the existing Quantum Natural Language framework into parameterized quantum circuits of reasonable qubits, which can be trained to solve various protein-related machine-learning problems. We classify proteins based on their subcellular locations, a pivotal task in bioinformatics that is key to understanding biological processes and disease mechanisms. Leveraging the quantum-enhanced processing capabilities, we demonstrate that Quantum Tensor Networks (QTN) can effectively handle the complexity and diversity of protein sequences. We present a detailed methodology that adapts QTN architectures to the nuanced requirements of protein data, supported by comprehensive experimental results. We demonstrate two distinct QTNs, inspired by classical recurrent neural networks (RNN) and convolutional neural networks (CNN), to solve the binary classification task mentioned above. Our top-performing quantum model has achieved a 94% accuracy rate, which is comparable to the performance of a classical model that uses the ESM2 protein language model embeddings. It's noteworthy that the ESM2 model is extremely large, containing 8 million parameters in its smallest configuration, whereas our best quantum model requires only around 800 parameters. We demonstrate that these hybrid models exhibit promising performance, showcasing their potential to compete with classical models of similar complexity.

Evaluating Efficacy of Model Stealing Attacks and Defenses on Quantum Neural Networks

Cloud hosting of quantum machine learning (QML) models exposes them to a range of vulnerabilities, the most significant of which is the model stealing attack. In this study, we assess the efficacy of such attacks in the realm of quantum computing. We conducted comprehensive experiments on various datasets with multiple QML model architectures. Our findings revealed that model stealing attacks can produce clone models achieving up to $0.9\times$ and $0.99\times$ clone test accuracy when trained using Top-$1$ and Top-$k$ labels, respectively ($k:$ num\_classes). To defend against these attacks, we leverage the unique properties of current noisy hardware and perturb the victim model outputs and hinder the attacker's training process. In particular, we propose: 1) hardware variation-induced perturbation (HVIP) and 2) hardware and architecture variation-induced perturbation (HAVIP). Although noise and architectural variability can provide up to $\sim16\%$ output obfuscation, our comprehensive analysis revealed that models cloned under noisy conditions tend to be resilient, suffering little to no performance degradation due to such obfuscations. Despite limited success with our defense techniques, this outcome has led to an important discovery: QML models trained on noisy hardwares are naturally resistant to perturbation or obfuscation-based defenses or attacks.

WEPRO: Weight Prediction for Efficient Optimization of Hybrid Quantum-Classical Algorithms

The exponential run time of quantum simulators on classical machines and long queue depths and high costs of real quantum devices present significant challenges in the effective training of Variational Quantum Algorithms (VQAs) like Quantum Neural Networks (QNNs), Variational Quantum Eigensolver (VQE) and Quantum Approximate Optimization Algorithm (QAOA). To address these limitations, we propose a new approach, WEPRO (Weight Prediction), which accelerates the convergence of VQAs by exploiting regular trends in the parameter weights. We introduce two techniques for optimal prediction performance namely, Naive Prediction (NaP) and Adaptive Prediction (AdaP). Through extensive experimentation and training of multiple QNN models on various datasets, we demonstrate that WEPRO offers a speedup of approximately $2.25\times$ compared to standard training methods, while also providing improved accuracy (up to $2.3\%$ higher) and loss (up to $6.1\%$ lower) with low storage and computational overheads. We also evaluate WEPRO's effectiveness in VQE for molecular ground-state energy estimation and in QAOA for graph MaxCut. Our results show that WEPRO leads to speed improvements of up to $3.1\times$ for VQE and $2.91\times$ for QAOA, compared to traditional optimization techniques, while using up to $3.3\times$ less number of shots (i.e., repeated circuit executions) per training iteration.

Random Relabeling for Efficient Machine Unlearning

Learning algorithms and data are the driving forces for machine learning to bring about tremendous transformation of industrial intelligence. However, individuals' right to retract their personal data and relevant data privacy regulations pose great challenges to machine learning: how to design an efficient mechanism to support certified data removals. Removal of previously seen data known as machine unlearning is challenging as these data points were implicitly memorized in training process of learning algorithms. Retraining remaining data from scratch straightforwardly serves such deletion requests, however, this naive method is not often computationally feasible. We propose the unlearning scheme random relabeling, which is applicable to generic supervised learning algorithms, to efficiently deal with sequential data removal requests in the online setting. A less constraining removal certification method based on probability distribution similarity with naive unlearning is further developed for logit-based classifiers.