arXiv subjects

Swaroop Ghosh

Publications and source records attributed to Swaroop Ghosh.

At least 73 records · Page 4Linked to original sources

arXiv2023

A First Order Survey of Quantum Supply Dynamics and Threat Landscapes

Quantum computing, with its transformative computational potential, is gaining prominence in the technological landscape. As a new and exotic technology, quantum computers involve innumerable Intellectual Property (IP) in the form of fabrication recipe, control electronics and software techniques, to name a few. Furthermore, complexity of quantum systems necessitates extensive involvement of third party tools, equipment and services which could risk the IPs and the Quality of Service and enable other attack surfaces. This paper is a first attempt to explore the quantum computing ecosystem, from the fabrication of quantum processors to the development of specialized software tools and hardware components, from a security perspective. By investigating the publicly disclosed information from industry front runners like IBM, Google, Honeywell and more, we piece together various components of quantum computing supply chain. We also uncover some potential vulnerabilities and attack models and suggest defenses. We highlight the need to scrutinize the quantum computing supply chain further through the lens of security.

quant-ph↗

arXiv2023

Randomized Reversible Gate-Based Obfuscation for Secured Compilation of Quantum Circuit

The success of quantum circuits in providing reliable outcomes for a given problem depends on the gate count and depth in near-term noisy quantum computers. Quantum circuit compilers that decompose high-level gates to native gates of the hardware and optimize the circuit play a key role in quantum computing. However, the quality and time complexity of the optimization process can vary significantly especially for practically relevant large-scale quantum circuits. As a result, third-party (often less-trusted/untrusted) compilers have emerged, claiming to provide better and faster optimization of complex quantum circuits than so-called trusted compilers. However, untrusted compilers can pose severe security risks, such as the theft of sensitive intellectual property (IP) embedded within the quantum circuit. We propose an obfuscation technique for quantum circuits using randomized reversible gates to protect them from such attacks during compilation. The idea is to insert a small random circuit into the original circuit and send it to the untrusted compiler. Since the circuit function is corrupted, the adversary may get incorrect IP. However, the user may also get incorrect output post-compilation. To circumvent this issue, we concatenate the inverse of the random circuit in the compiled circuit to recover the original functionality. We demonstrate the practicality of our method by conducting exhaustive experiments on a set of benchmark circuits and measuring the quality of obfuscation by calculating the Total Variation Distance (TVD) metric. Our method achieves TVD of up to 1.92 and performs at least 2X better than a previously reported obfuscation method. We also propose a novel adversarial reverse engineering (RE) approach and show that the proposed obfuscation is resilient against RE attacks. The proposed technique introduces minimal degradation in fidelity (~1% to ~3% on average).

quant-ph↗

arXiv2023

TrojanNet: Detecting Trojans in Quantum Circuits using Machine Learning

Quantum computing holds tremendous potential for various applications, but its security remains a crucial concern. Quantum circuits need high-quality compilers to optimize the depth and gate count to boost the success probability on current noisy quantum computers. There is a rise of efficient but unreliable/untrusted compilers; however, they present a risk of tampering such as Trojan insertion. We propose TrojanNet, a novel approach to enhance the security of quantum circuits by detecting and classifying Trojan-inserted circuits. In particular, we focus on the Quantum Approximate Optimization Algorithm (QAOA) circuit that is popular in solving a wide range of optimization problems. We investigate the impact of Trojan insertion on QAOA circuits and develop a Convolutional Neural Network (CNN) model, referred to as TrojanNet, to identify their presence accurately. Using the Qiskit framework, we generate 12 diverse datasets by introducing variations in Trojan gate types, the number of gates, insertion locations, and compiler backends. These datasets consist of both original Trojan-free QAOA circuits and their corresponding Trojan-inserted counterparts. The generated datasets are then utilized for training and evaluating the TrojanNet model. Experimental results showcase an average accuracy of 98.80% and an average F1-score of 98.53% in effectively detecting and classifying Trojan-inserted QAOA circuits. Finally, we conduct a performance comparison between TrojanNet and existing machine learning-based Trojan detection methods specifically designed for conventional netlists.

quant-ph↗

arXiv2023

A Reference-less Slope Detection Technique in 65nm for Robust Sensing of 1T1R Arrays

Spin-Torque-Transfer RAM (STTRAM) is a promising technology however process variation poses serious challenge to sensing. To eliminate bit-to-bit process variation we propose a reference-less, destructive slope detection technique which exploits the MTJ switching from high to low state to detect memory state. A proof-of-concept fabricated test-chip using 96kb mimicked STTRAM bits in 65nm technology shows that slope sensing reduces failure rate by 120X in 2.5K-5K array@TMR=100% and 162X in 2.5K-5K@TMR=80% array compared to conventional voltage sensing.

cs.ET↗

arXiv2023

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.

cs.LG↗

arXiv2023

Obfuscating Quantum Hybrid-Classical Algorithms for Security and Privacy

As the quantum computing ecosystem grows in popularity and utility it is important to identify and address the security and privacy vulnerabilities before they can be widely exploited. One major concern is the involvement of third party tools and hardware. Usage of untrusted hardware could present the risk of intellectual property (IP) theft. For example the hybrid quantum classical algorithms like QAOA encodes the graph properties e.g. number of nodes edges and connectivity in the parameterized quantum circuit to solve a graph maxcut problem. QAOA employs a classical computer which optimizes the parameters of a parametric quantum circuit (which encodes graph structure) iteratively by executing the circuit on a quantum hardware and measuring the output. The graph properties can be readily retrieved by analyzing the QAOA circuit by the untrusted quantum hardware provider. To mitigate this risk, we propose an edge pruning obfuscation method for QAOA along with a split iteration methodology. The basic idea is to (i) create two flavors of QAOA circuit each with few distinct edges eliminated from the problem graph for obfuscation (ii) iterate the circuits alternately during optimization process to uphold the optimization quality and (iii) send the circuits to two different untrusted hardware provider so that the adversary has access to partial graph protecting the IP. We demonstrate that combining edge pruning obfuscation with split iteration on two different hardware secures the IP and increases the difficulty of reconstruction while limiting performance degradation to a maximum of 10 percent (approximately 5 percent on average) and maintaining low overhead costs (less than 0.5X for QAOA with single layer implementation).

quant-ph↗

arXiv2023

Quantum Random Access Memory For Dummies

Quantum Random Access Memory (QRAM) has the potential to revolutionize the area of quantum computing. QRAM uses quantum computing principles to store and modify quantum or classical data efficiently, greatly accelerating a wide range of computer processes. Despite its importance, there is a lack of comprehensive surveys that cover the entire spectrum of QRAM architectures. We fill this gap by providing a comprehensive review of QRAM, emphasizing its significance and viability in existing noisy quantum computers. By drawing comparisons with conventional RAM for ease of understanding, this survey clarifies the fundamental ideas and actions of QRAM.

quant-ph↗

arXiv2023

Shot Optimization in Quantum Machine Learning Architectures to Accelerate Training

In this paper, we propose shot optimization method for QML models at the expense of minimal impact on model performance. We use classification task as a test case for MNIST and FMNIST datasets using a hybrid quantum-classical QML model. First, we sweep the number of shots for short and full versions of the dataset. We observe that training the full version provides 5-6% higher testing accuracy than short version of dataset with up to 10X higher number of shots for training. Therefore, one can reduce the dataset size to accelerate the training time. Next, we propose adaptive shot allocation on short version dataset to optimize the number of shots over training epochs and evaluate the impact on classification accuracy. We use a (a) linear function where the number of shots reduce linearly with epochs, and (b) step function where the number of shots reduce in step with epochs. We note around 0.01 increase in loss and around 4% (1%) reduction in testing accuracy for reduction in shots by up to 100X (10X) for linear (step) shot function compared to conventional constant shot function for MNIST dataset, and 0.05 increase in loss and around 5-7% (5-7%) reduction in testing accuracy with similar reduction in shots using linear (step) shot function on FMNIST dataset. For comparison, we also use the proposed shot optimization methods to perform ground state energy estimation of different molecules and observe that step function gives the best and most stable ground state energy prediction at 1000X less number of shots.

quant-ph↗

arXiv2023

Quantum Error Correction For Dummies

In the current Noisy Intermediate Scale Quantum (NISQ) era of quantum computing, qubit technologies are prone to imperfections, giving rise to various errors such as gate errors, decoherence/dephasing, measurement errors, leakage, and crosstalk. These errors present challenges in achieving error-free computation within NISQ devices. A proposed solution to this issue is Quantum Error Correction (QEC), which aims to rectify the corrupted qubit state through a three-step process: (i) detection: identifying the presence of an error, (ii) decoding: pinpointing the location(s) of the affected qubit(s), and (iii) correction: restoring the faulty qubits to their original states. QEC is an expanding field of research that encompasses intricate concepts. In this paper, we aim to provide a comprehensive review of the historical context, current state, and future prospects of Quantum Error Correction, tailored to cater to computer scientists with limited familiarity with quantum physics and its associated mathematical concepts. In this work, we, (a) explain the foundational principles of QEC and explore existing Quantum Error Correction Codes (QECC) designed to correct errors in qubits, (b) explore the practicality of these QECCs concerning implementation and error correction quality, and (c) highlight the challenges associated with implementing QEC within the context of the current landscape of NISQ computers.

quant-ph↗

arXiv2022

Energy-based Generative Models for Target-specific Drug Discovery

Drug targets are the main focus of drug discovery due to their key role in disease pathogenesis. Computational approaches are widely applied to drug development because of the increasing availability of biological molecular datasets. Popular generative approaches can create new drug molecules by learning the given molecule distributions. However, these approaches are mostly not for target-specific drug discovery. We developed an energy-based probabilistic model for computational target-specific drug discovery. Results show that our proposed TagMol can generate molecules with similar binding affinity scores as real molecules. GAT-based models showed faster and better learning relative to GCN baseline models.

cs.LG↗

arXiv2022

Architectures for Quantum Information Processing

Quantum computing is changing the way we think about computing. Significant strides in research and development for managing and harnessing the power of quantum systems has been made in recent years, demonstrating the potential for transformative quantum technology. Quantum phenomena like superposition, entanglement, and interference can be exploited to solve issues that are difficult for traditional computers. IBM's first public access to true quantum computers through the cloud, as well as Google's demonstration of quantum supremacy, are among the accomplishments. Besides, a slew of other commercial, government, and academic projects are in the works to create next-generation hardware, a software stack to support the hardware ecosystem, and viable quantum algorithms. This chapter covers various quantum computing architectures including many hardware technologies that are being investigated. We also discuss a variety of challenges, including numerous errors/noise that plague the quantum computers. An overview of literature investigating noise-resilience approaches is also presented.

quant-ph↗

arXiv2022

Approximate Quantum Random Access Memory Architectures

Quantum supremacy in many applications using well-known quantum algorithms rely on availability of data in quantum format. Quantum Random Access Memory (QRAM), an equivalent of classical Random Access Memory (RAM), fulfills this requirement. However, the existing QRAM proposals either require qutrit technology and/or incur access challenges. We propose an approximate Parametric Quantum Circuit (PQC) based QRAM which takes address lines as input and gives out the corresponding data in these address lines as the output. We present two applications of the proposed PQC-based QRAM namely, storage of binary data and storage of machine learning (ML) dataset for classification.

quant-ph↗

arXiv2022

Robust and Secure Hybrid Quantum-Classical Computation on Untrusted Cloud-Based Quantum Hardware

Quantum computers are currently accessible through a cloud-based platform that allows users to run their programs on a suite of quantum hardware. As the quantum computing ecosystem grows in popularity and utility, it is reasonable to expect more companies, including untrustworthy or untrustworthy or unreliable vendors, to begin offering quantum computers as hardware as a service at various price or performance points. Since computing time on quantum hardware is expensive and the access queue may be long, users will be enticed to use less expensive but less reliable or trustworthy hardware. Less trusted vendors may tamper with the results and or parameters of quantum circuits, providing the user with a sub-optimal solution or incurring a cost of higher iterations. In this paper, we model and simulate adversarial tampering of input parameters and measurement outcomes on an exemplary hybrid quantum classical algorithm namely, Quantum Approximate Optimization Algorithm (QAOA). We observe a maximum performance degradation of approximately 40%. To achieve comparable performance with minimal parameter tampering, the user incurs a minimum cost of 20X higher iteration. We propose distributing the computation (iterations) equally among the various hardware options to ensure trustworthy computing for a mix of trusted and untrusted hardware. In the chosen performance metrics, we observe a maximum improvement of approximately 30%. In addition, we propose re-initialization of the parameters after a few initial iterations to fully recover the original program performance and an intelligent run adaptive split heuristic, which allows users to identify tampered/untrustworthy hardware at runtime and allocate more iterations to the reliable hardware, resulting in a maximum improvement of approximately 45%.

quant-ph↗

arXiv2022

Quantum Machine Learning for Material Synthesis and Hardware Security

Using quantum computing, this paper addresses two scientifically pressing and day-to-day relevant problems, namely, chemical retrosynthesis which is an important step in drug/material discovery and security of the semiconductor supply chain. We show that Quantum Long Short-Term Memory (QLSTM) is a viable tool for retrosynthesis. We achieve 65% training accuracy with QLSTM, whereas classical LSTM can achieve 100%. However, in testing, we achieve 80% accuracy with the QLSTM while classical LSTM peaks at only 70% accuracy! We also demonstrate an application of Quantum Neural Network (QNN) in the hardware security domain, specifically in Hardware Trojan (HT) detection using a set of power and area Trojan features. The QNN model achieves detection accuracy as high as 97.27%.

quant-ph↗

arXiv2022

Knowledge Distillation in Quantum Neural Network using Approximate Synthesis

Recent assertions of a potential advantage of Quantum Neural Network (QNN) for specific Machine Learning (ML) tasks have sparked the curiosity of a sizable number of application researchers. The parameterized quantum circuit (PQC), a major building block of a QNN, consists of several layers of single-qubit rotations and multi-qubit entanglement operations. The optimum number of PQC layers for a particular ML task is generally unknown. A larger network often provides better performance in noiseless simulations. However, it may perform poorly on hardware compared to a shallower network. Because the amount of noise varies amongst quantum devices, the optimal depth of PQC can vary significantly. Additionally, the gates chosen for the PQC may be suitable for one type of hardware but not for another due to compilation overhead. This makes it difficult to generalize a QNN design to wide range of hardware and noise levels. An alternate approach is to build and train multiple QNN models targeted for each hardware which can be expensive. To circumvent these issues, we introduce the concept of knowledge distillation in QNN using approximate synthesis. The proposed approach will create a new QNN network with (i) a reduced number of layers or (ii) a different gate set without having to train it from scratch. Training the new network for a few epochs can compensate for the loss caused by approximation error. Through empirical analysis, we demonstrate ~71.4% reduction in circuit layers, and still achieve ~16.2% better accuracy under noise.

quant-ph↗

arXiv2022

Analysis of Power-Oriented Fault Injection Attacks on Spiking Neural Networks

Spiking Neural Networks (SNN) are quickly gaining traction as a viable alternative to Deep Neural Networks (DNN). In comparison to DNNs, SNNs are more computationally powerful and provide superior energy efficiency. SNNs, while exciting at first appearance, contain security-sensitive assets (e.g., neuron threshold voltage) and vulnerabilities (e.g., sensitivity of classification accuracy to neuron threshold voltage change) that adversaries can exploit. We investigate global fault injection attacks by employing external power supplies and laser-induced local power glitches to corrupt crucial training parameters such as spike amplitude and neuron's membrane threshold potential on SNNs developed using common analog neurons. We also evaluate the impact of power-based attacks on individual SNN layers for 0% (i.e., no attack) to 100% (i.e., whole layer under attack). We investigate the impact of the attacks on digit classification tasks and find that in the worst-case scenario, classification accuracy is reduced by 85.65%. We also propose defenses e.g., a robust current driver design that is immune to power-oriented attacks, improved circuit sizing of neuron components to reduce/recover the adversarial accuracy degradation at the cost of negligible area and 25% power overhead. We also present a dummy neuron-based voltage fault injection detection system with 1% power and area overhead.

cs.AI↗

arXiv2022

A Shuttle-Efficient Qubit Mapper for Trapped-Ion Quantum Computers

Trapped-ion (TI) quantum computer is one of the forerunner quantum technologies. However, TI systems can have a limited number of qubits in a single trap. Execution of meaningful quantum algorithms requires a multiple trap system. In such systems, the computation may frequently involve ions from two different traps for which the qubits must be co-located in the same trap, hence one of the ions needs to be shuttled (moved) between traps, increasing the vibrational energy, degrading fidelity, and increasing the program execution time. The choice of initial mapping influences the number of shuttles. The existing Greedy policy counts the number of gates occurring between each pair of qubits and assigns edge weight. The qubits with high edge weights are placed close to each other. However, it neglects the stage of the program at which the gate is occurring. Intuitively, the contribution of the late-occurring gates to the initial mapping reduces since the ions might have already shuttled to a different trap to satisfy other gate operations. In this paper, we target this gap and propose a new policy especially for programs with considerable depth and high number of qubits (valid for practical-scale quantum programs). Our policy is program adaptive and prioritizes the gates re-occurring at the initial stages of the program over late occurring gates. Our technique achieves an average reduction of 9% shuttles/program (with 21.3% at best) for 120 random circuits and enhances the program fidelity up to 3.3X (1.41X on average).

quant-ph↗

arXiv2022

Security Aspects of Quantum Machine Learning: Opportunities, Threats and Defenses

In the last few years, quantum computing has experienced a growth spurt. One exciting avenue of quantum computing is quantum machine learning (QML) which can exploit the high dimensional Hilbert space to learn richer representations from limited data and thus can efficiently solve complex learning tasks. Despite the increased interest in QML, there have not been many studies that discuss the security aspects of QML. In this work, we explored the possible future applications of QML in the hardware security domain. We also expose the security vulnerabilities of QML and emerging attack models, and corresponding countermeasures.

cs.CR↗