Standard universal quantum computing using exchange-only qubits typically requires three physical spins per logical qubit, leading to significant hardware overhead. Conversely, two-spin units offer higher density but rely on local magnetic field gradients for control, increasing integration complexity. In this paper, we propose a resource-efficient quantum machine learning (QML) architecture that achieves high expressibility using minimal two-spin units and Heisenberg exchange interactions alone, without any magnetic gradients. We shift the paradigm from universal gate-based control to utilizing the intrinsic, time-domain dynamics of a spin chain as a learning resource. Numerical simulations on MNIST digit classification demonstrate that the symmetry-protected constraints of isolated spin pairs are bypassed by leveraging inter-pair exchange coupling. This interference-mediated state mixing significantly enhances the expressibility of the Hilbert space. The model reaches a test-set accuracy of 90.9% +/- 0.2% over five independent seeds on the full 10,000-image MNIST test set. Under an identical linear readout, the trained quantum feature map (88.1%) clearly outperforms a classical linear baseline on the same PCA inputs (83.2%) as well as an untrained (reservoir-style) version of the same dynamics (53.0%), demonstrating that the learned, input-dependent exchange pulses implement a genuinely non-linear and trainable feature map. The protocol is also robust to experimentally relevant imperfections: accuracy remains at 89.9% under 10% quasi-static pulse-area noise and at 89.8% when every observable is estimated from 10^3 measurement shots. These findings suggest that competitive QML can be executed on the simplest possible semiconductor spin-chain hardware, bypassing the need for leakage-prone encodings or complex micro-magnet integration.
Comments
9 pages, 10 figures. Code and data: https://github.com/minatoyuichiro/teex-mnist
Standard universal quantum computing using exchange-only qubits typically requires three physical spins per logical qubit, leading to significant hardware overhead. Conversely, two-spin units offer higher density but rely on local magnetic field gradients for control, increasing integration complexity. In this paper, we propose a resource-efficient quantum machine learning (QML) architecture that achieves high expressibility using minimal two-spin units and Heisenberg exchange interactions alone, without any magnetic gradients. We shift the paradigm from universal gate-based control to utilizing the intrinsic, time-domain dynamics of a spin chain as a learning resource. Numerical simulations on MNIST digit classification demonstrate that the symmetry-protected constraints of isolated spin pairs are bypassed by leveraging inter-pair exchange coupling. This interference-mediated state mixing significantly enhances the expressibility of the Hilbert space. The model reaches a test-set accuracy of 90.9% +/- 0.2% over five independent seeds on the full 10,000-image MNIST test set. Under an identical linear readout, the trained quantum feature map (88.1%) clearly outperforms a classical linear baseline on the same PCA inputs (83.2%) as well as an untrained (reservoir-style) version of the same dynamics (53.0%), demonstrating that the learned, input-dependent exchange pulses implement a genuinely non-linear and trainable feature map. The protocol is also robust to experimentally relevant imperfections: accuracy remains at 89.9% under 10% quasi-static pulse-area noise and at 89.8% when every observable is estimated from 10^3 measurement shots. These findings suggest that competitive QML can be executed on the simplest possible semiconductor spin-chain hardware, bypassing the need for leakage-prone encodings or complex micro-magnet integration.
Quantum machine learning (QML) promises new representational and computational capabilities, yet practical deployment on near-term hardware is hampered by resource overheads, depth constraints, and fragile trainability. This thesis advances resource-aware QML by proposing architectures and kernels that retain expressivity while sharply reducing qubit counts, circuit depth, and entangling-gate budgets. The work is presented in an article-based format with three core contributions. First, I introduce a Coherent Feed-Forward Quantum Neural Network (CFF-QNN) that preserves quantum coherence across all layers, mirrors the flexibility of classical feed-forward networks (adjustable hidden layers and nodes), and decouples qubit requirements from input feature dimension. Compared to prevailing QNN baselines, the CFF-QNN reduces both depth and CNOT count by over 50% while achieving strong performance on standard benchmarks (e.g., 91% accuracy on Wisconsin breast cancer and 85% on credit-card fraud). This contribution is accompanied by an international patent filing (WO2025050205A1). Second, I develop a resource-efficient quantum kernel that enables high-dimensional embeddings with substantially fewer qubits and entanglers, achieving linear scaling of entangling gates in the number of qubits. Empirically, the kernel delivers competitive or superior performance to widely used classical kernels and to popular quantum feature maps (e.g., on the Parkinson's disease dataset), with noisy simulations and small-scale runs on superconducting hardware indicating suitability for near-term devices. This contribution is covered by a companion patent filing (WO2025073041A1). Third, I propose a quantum reservoir computing (QRC) scheme that reuses a fixed quantum feature-map circuit as the reservoir while injecting temporal memory via an explicit feedback loop. The register is recycled across time, so quantum resources remain constant, no quantum parameters are trained, and only a lightweight classical readout is fitted. Experiments on chaotic time series (e.g., Mackey-Glass) show competitive predictive accuracy with strict resource efficiency, and illuminate how feedback delay and entangling structure affect memory and error. Collectively, these results chart a path toward practical QML: coherent architectures and kernels that are trainable, scalable in input dimension, frugal in quantum resources, and viable on noisy intermediate-scale devices—while providing design guidelines for future, larger-scale implementations.
Quantum Machine Learning (QML) has emerged as a promising framework for exploring how quantum dynamics may enhance data processing tasks. Here we investigate Quantum Extreme Learning Machines (QELMs), a quantum analogue of classical Extreme Learning Machines in which training is restricted to the output layer. Our architecture combines dimensionality reduction (via PCA or Autoencoders), quantum state encoding, evolution under an XX Hamiltonian, and projective measurement to produce features for a classical single-layer classifier. By analyzing the classification accuracy as a function of evolution time, we observe a sharp transition between low- and high-accuracy regimes, followed by saturation. Remarkably, the saturated performance is comparable to that obtained using Haar-random unitaries that generate maximally complex dynamics, even though the XX model is integrable and local. Our results indicate that this increase in performance correlates with the onset of entanglement, which improves the embedding of classical data in Hilbert space and leads to more separable clusters in measurement probability space. Thus, moderate entanglement can contribute positively to the structure of the data representation, improving learnability without necessarily implying quantum computational advantage. For the image-classification tasks studied here, namely MNIST, Fashion-MNIST, and CIFAR-10, the relevant evolution time is consistent with information exchange over short distances and, within the explored system sizes, does not show evidence of scaling with the full system size. This suggests that QELM performance in this regime relies only on limited entanglement and remains compatible with efficient classical simulation. Our results clarify how local quantum dynamics and moderate quantum correlations are already sufficient to generate useful feature representations for learning.
Abstract Quantum machine learning (QML) has emerged as an innovative framework that has the potential to uncover complex patterns by leveraging the ability of quantum systems to simulate and exploit high-dimensional latent spaces, particularly in learning tasks. Quantum neural network frameworks are inherently sensitive to the precision of gradient calculations and the computational limitations of current quantum hardware, as unitary rotations introduce overhead from complex number computations, and quantum gate operation speed remains a bottleneck for practical implementations. In this study, we introduce a quantum parallel information exchange hybrid network, a new non-sequential hybrid classical quantum model architecture that leverages quantum transfer learning by feeding pre-trained parameters from classical neural networks into quantum circuits. This enables efficient pattern recognition and temporal series data prediction by utilizing non-Clifford parameterized quantum gates, thereby enhancing both learning efficiency and representational capacity. Additionally, we developed a dynamic gradient selection method that applies the parameter-shift rule to quantum processing units (QPUs) and adjoint differentiation to graphics processing units (GPUs). Our results demonstrate that the model performance exhibits higher accuracy in ad-hoc benchmarks, lowering approximately 88% convergence rate for extra stochasticity time-series data within 100 -steps, and showing a more unbiased eigenvalue spectrum of the Fisher information matrix on the CPU/GPU and IonQ QPU simulators.
A two-level quantum system is the building block of a quantum computer. This pair of quantum states defines the computational unit of a quantum computer, and it is called a quantum bit or qubit in analogy to the bit that is the binary unit of a classical computer. The spin of a single electron naturally defines such a two-level quantum system. A quantum dot can be tuned to the single-electron regime, and an array of singly occupied quantum dots provides a multi-qubit register. This thesis describes the multielectron encoding of a spin qubit using a pair or a trio of quantum dots, and it analyzes the coherence properties and manipulation protocols of this system. The singlet-triplet qubit, coded using the singlet state and the spinless triplet state of a pair of singly occupied quantum dots, can be controlled all electrically when applying voltages to gates close to the quantum dot structure. Rapid, subnanosecond modifications of the double quantum dot's charge configuration have been realized successfully. A small magnetic field gradient across the double quantum dot enables universal qubit control. Two modifications of the normal singlet-triplet encoding are introduced. (1) The six-electron configuration of a double quantum dot encodes a singlet-triplet qubit in the same way as for the two-electron double quantum dot. Two electrons at each quantum dot are irrelevant for the qubit manipulations because they are paired in a singlet state; the remaining two electrons encode the qubit. The qubit's wave function is immune to charge noise at moderate out-of-plane magnetic fields. (2) A singlet-triplet qubit encoded using two quantum dots of different sizes has an orbital state degeneracy of the singlet state and the spinless triplet state at finite out-of-plane magnetic fields. Spin-orbit interactions lift this state degeneracy, while the magnitude of the state coupling is determined by the size difference of the QDs. This setup enables the manipulations of singlet-triplet qubits without the need for magnetic field gradients. Finally, two-qubit gates between singlet-triplet qubits are proposed that use mediated exchange interactions via one quantum state. These operations are well controlled and highly noise insensitive. Orbital interactions alone can control spin qubits coded in a three electron Hilbert space. For example, the exchange-only qubit is encoded using three singly occupied quantum dots. The exchange interactions of two quantum dot pairs need to be modified to manipulate this qubit. The noise sensitivity of the exchange-only qubit is discussed. Alternatively, a three-electron qubit at a double quantum dot can be operated when single electrons are transferred between the quantum dots. Fast, subnanosecond manipulations of the double quantum dot's charge configuration are required to realize single-qubit gates. A novel two-qubit pulse gate for the three-electron double quantum dot qubit is proposed.
We analyze a simple exchange-based two-qubit gate for singlet-triplet qubits in gate-defined semiconductor quantum dots that can be implemented in a single exchange pulse. Excitations from the logical subspace are suppressed by a magnetic field gradient that causes spin-flip transitions to be non-energy-conserving. We show that the use of adiabatic pulses greatly reduces leakage processes comapred to square pulses. We also characterize the effect of charge noise on the entanglement fidelity of the gate both analytically and in simulations; demonstrating high entanglement fidelities for physically realistic experimental parameters. Specifically we find that it is possible to achieve fidelities and gate times that are comparable to single-qubit states using realistic magnetic field gradients.
We propose a singlet-only always-on gapless exchange (SAGE) spin qubit that encodes a single qubit in the spins of four electrons while allowing universal baseband control. While conventional exchange-only qubits suffer from magnetic-field-gradient-induced leakage and coherent errors due to local nuclear environments and variations in the g-factor, the SAGE qubit subspace is protected from coherent errors due to local magnetic field gradients, and leakage out of the computational subspace is energetically suppressed due to the exchange interactions between electrons being always on. Consequently, we find that when magnetic gradient noise dominates over charge noise, coherence times and single-qubit gate infidelities of the SAGE qubit improve by an order of magnitude compared to conventional exchange-only qubits. Moreover, using realistic parameters, two-qubit gates can be performed with a single interqubit exchange pulse with times comparable in duration to conventional exchange-only qubits but with a significantly simplified pulse sequence.