A5008277800- Quantum Information and Cryptography
- Neural Networks and Reservoir Computing
- Quantum Computing Algorithms and Architecture
- Optical Network Technologies
- Neural Networks and Applications
- Particle accelerators and beam dynamics
- Gyrotron and Vacuum Electronics Research
- Pulsed Power Technology Applications
- Blockchain Technology Applications and Security
- Industrial Technology and Control Systems
- Statistical Mechanics and Entropy
- Machine Learning and ELM
- Quantum and electron transport phenomena
- Machine Learning in Materials Science
- Quantum Mechanics and Applications
- Model Reduction and Neural Networks
- Particle Accelerators and Free-Electron Lasers
Princeton University
2023-2024
Tsinghua University
2022-2024
The practical implementation of many quantum algorithms known today is limited by the coherence time executing hardware and sampling noise. Here we present a machine learning algorithm, NISQRC, for qubit-based systems that enables inference on temporal data over durations unconstrained decoherence. NISQRC leverages mid-circuit measurements deterministic reset operations to reduce circuit executions, while still maintaining an appropriate length persistent memory in system, confirmed through...
The expressive capacity of physical systems employed for learning is limited by the unavoidable presence noise in their extracted outputs. Though present across both classical and quantum regimes, precise impact on remains poorly understood. Focusing supervised learning, we a mathematical framework evaluating resolvable (REC) general under finite sampling noise, provide methodology extracting its extrema, eigentasks. Eigentasks are native set functions that given system can approximate with...
A compact $X$-band two-stage rf pulse compression system has been successfully designed, fabricated, and tested at Tsinghua high-power test stand. The consists of a correction cavity chain, first-stage, second-stage storage cavity. chain adopts new design whose transmission loss length are reduced by half compared with the old one. detuning device is applied to so that can work in one-stage or alternatively mode. In mode, 150-ns, 70-MW flattop output, standard deviation 1.5% amplitude...
The paradigm of reservoir computing exploits the nonlinear dynamics a physical to perform complex time-series processing tasks such as speech recognition and forecasting. Unlike other machine-learning approaches, relaxes need for optimization intra-network parameters, is thus particularly attractive near-term hardware-efficient quantum implementations. However, complete description practical computers requires accounting their placement in measurement chain, its conditional evolution under...
Although linear quantum amplification has proven essential to the processing of weak signals, extracting higher-order features such as correlations in principle demands nonlinear operations. However, signals is often associated with non-idealities and excess noise, absent a general framework harness nonlinearity, regimes are typically avoided. Here we present uncover signal principles broad class bosonic processors (QNPs), inspired by remarkably analogous paradigm nature: environmental...
Tackling output sampling noise due to finite shots of quantum measurement is an unavoidable challenge when extracting information in machine learning with physical systems. A technique called Eigentask Learning was developed recently as a framework for infinite input training data the presence noise. In work Learning, numerical evidence presented that low-noise contributions features can practically improve performance tasks, displaying robustness overfitting and increasing generalization...
Duality quantum computing (DQC) offers the use of linear combination unitaries (LCU), or generalized gates, in designing algorithms. DQC contains wave divider and combiner operations. The function a computer is split into several subwaves after division operation. Then different unitary operations are performed on parallel. A combines final function, so that state. In this paper, we study properties duality with projections subwaves. subwave-projection (SWP-DQC), can realize combinations...
The practical implementation of many quantum algorithms known today is believed to be limited by the coherence time executing hardware and sampling noise. Here we present a machine learning algorithm, NISQRC, for qubit-based systems that enables processing temporal data over durations unconstrained finite times constituent qubits. NISQRC strikes balance between input encoding steps mid-circuit measurements with reset endow system an appropriate-length persistent memory capture time-domain...
The expressive capacity of quantum systems for machine learning is limited by sampling noise incurred during measurement. Although it generally believed that limits the resolvable systems, precise impact on not yet fully understood. We present a mathematical framework evaluating available general from finite number measurements, and provide methodology extracting extrema this capacity, its eigentasks. Eigentasks are native set functions given system can approximate with minimal error. show...