Quantum Data Classification with 90% Accuracy

Quantum-secure deep learning that protects both model and data Cloud-based deep learning creates a fundamental privacy problem. Clients with sensitive data—hospitals classifying patient scans, banks detecting fraud—need powerful models they can't run locally. Server operators have proprietary models worth protecting. Standard approaches like homomorphic encryption carry massive computational overhead and rely on computational assumptions that future adversaries might break. Yet quantum mechanics has long hinted that if you encode information into the right physical states, you can get security guarantees that no amount of computing power can overcome. Sulimany and coauthors take that idea and build a working protocol around it. Using standard telecommunications components—coherent laser pulses, homodyne detection, optical modulators—they encode neural network weights into the complex amplitudes of weak coherent light states. The client performs optical transformations using their private data to compute inner products, then returns the residual light to the server as a security certificate. Shot noise—the fundamental quantum uncertainty in measuring light's amplitude—masks information in both directions, while the Holevo bound limits what the client can learn about weights and the Cramér-Rao bound limits what the server can learn about data. Applied to MNIST classification, the protocol achieves greater than 95% accuracy while leaking less than 0.1 bits per weight and per data element—an order of magnitude below the minimum bit precision required for accurate inference, meaning leaked information is genuinely useless for reconstruction. Remarkably, leakage decreases with network size, making the approach increasingly favorable for state-of-the-art architectures. The upshot is striking: by encoding computation into quantum optical states and leveraging fundamental physical bounds on information extraction, it becomes possible to achieve what classical cryptography cannot—provable, information-theoretic security for multiparty deep learning, implemented with present-day photonics and ready for deployment in finance, healthcare, and anywhere privacy is paramount. Paper: https://lnkd.in/eRy7zm6W #QuantumComputing #DeepLearning #Cybersecurity #QuantumCryptography #MachineLearning #PrivacyPreservingAI #Photonics #QuantumInformation #CloudComputing #AIforScience #DataPrivacy #SecureComputation #QuantumOptics #NeuralNetworks #ScienceCommunication

I have been exploring how classical deep learning models and quantum circuits can be combined to solve machine learning problems more effectively. Rather than replacing classical approaches, this work focuses on leveraging the strengths of both paradigms. Below is a high-level overview of a hybrid quantum–classical classifier trained on the MNIST dataset (binary classification: digits 0 vs 1). 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 𝗢𝘃𝗲𝗿𝘃𝗶𝗲𝘄 𝗖𝗹𝗮𝘀𝘀𝗶𝗰𝗮𝗹 𝗙𝗲𝗮𝘁𝘂𝗿𝗲 𝗘𝘅𝘁𝗿𝗮𝗰𝘁𝗶𝗼𝗻 (𝗣𝘆𝗧𝗼𝗿𝗰𝗵) A standard convolutional neural network processes the 28×28 MNIST images and extracts high-level features. This step reduces the dimensionality of the input while preserving essential information. 𝗙𝗲𝗮𝘁𝘂𝗿𝗲 𝗛𝗮𝗻𝗱𝗼𝗳𝗳 The CNN outputs two features, chosen to match the number of qubits used in the quantum circuit. 𝗤𝘂𝗮𝗻𝘁𝘂𝗺 𝗣𝗿𝗼𝗰𝗲𝘀𝘀𝗶𝗻𝗴 𝗟𝗮𝘆𝗲𝗿 (𝗤𝗶𝘀𝗸𝗶𝘁) The features are encoded into a quantum state using a ZZFeatureMap. A parameterized variational circuit (RealAmplitudes) transforms this state, and the expectation value of a measurement operator is used for classification. 𝗘𝗻𝗱-𝘁𝗼-𝗘𝗻𝗱 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 Using Qiskit’s TorchConnector, the quantum and classical components are trained jointly. Gradients from the quantum circuit are computed using the parameter-shift rule and integrated with PyTorch’s automatic differentiation. 𝗧𝗼𝗼𝗹𝘀 𝗮𝗻𝗱 𝗦𝗲𝘁𝘂𝗽 • PyTorch and Qiskit Machine Learning • Training performed on the Aer simulator, with compatibility for quantum hardware via primitives On small subsets of the dataset, this hybrid approach achieves perfect classification accuracy, demonstrating how classical feature extraction combined with quantum decision layers can be effective even with limited qubit resources. Hybrid quantum–classical models provide a practical direction for near-term quantum machine learning, especially when classical preprocessing reduces problem complexity before quantum evaluation. 𝗥𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲𝘀 𝗮𝗻𝗱 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗥𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀 https://lnkd.in/gkwjwp-r https://lnkd.in/g7ge8MHD https://lnkd.in/gApuMGHE ♻️ Repost if you found this valuable! ➕ Follow me https://lnkd.in/gGhxx66A for more insights on the cutting edge of AI and Quantum. #QuantumComputing #QuantumMachineLearning #MachineLearning #PyTorch #Qiskit #ArtificialIntelligence #HybridModels

*Quantum-limited stochastic optical neural networks using just ~1 photon per neuron activation* The energy efficiency of computing is ultimately limited by noise, with quantum noise as the fundamental floor. What happens if we operate an optical neural network with such low power that each neuron activation is caused by just a single photon? In this regime, quantum noise dominates and the system is very stochastic - so much so that you might imagine it would be wildly inaccurate at any deterministic classification task. In our paper out in Nature Communications today, https://lnkd.in/e4RvuVbB , we show that it is possible to train such a system to perform MNIST handwritten-digit classification with accuracy >98% while using only 0.038 detected photons per multiply-accumulate (MAC) operation. In total, a few thousand photons are detected per inference, so there is clearly enough information to convey which of 10 digits the input image contained, but it's still a bit mind-bending to think about what happens at the level of individual MACs. Congratulations to Shi-Yuan Ma and co-authors Tianyu Wang, Jérémie Laydevant, and Logan Wright!

Check out this research on the scalable parameterized quantum circuits classifier (SPQCC) by Ding, X., Song, Z., Xu, J. et al., published in Scientific Reports #Nature. This approach addresses the limitations of conventional parameterized quantum circuits (PQC) in multi-category classification tasks, achieving state-of-the-art results. Key Highlights: 🔹 Fast convergence of the classifier 🔹 Parallel execution on identical quantum machines 🔹 Reduced complexity in classifier design 🔹 State-of-the-art simulation results on the MNIST Dataset 🔹 Comparable performance to classical classifiers with fewer parameters Their novel methodology involves per-channel PQC execution, combining measurements as outputs, and optimizing trainable parameters to minimize cross-entropy loss, leading to rapid convergence and enhanced classification accuracy. Explore the open-source research paper and the methodology leading to the results: https://lnkd.in/dT68bR8Q Let’s continue pushing the boundaries of what’s possible with quantum technology! #QuantumComputing #MachineLearning #Innovation #QuantumTech #Research #OpenSource #AI #MNIST Citation: Ding, X., Song, Z., Xu, J. et al. Scalable parameterized quantum circuits classifier. Sci Rep14, 15886 (2024). https://lnkd.in/dT68bR8Q