arXiv subjects

Rui Yang

Publications and source records attributed to Rui Yang.

At least 289 records · Page 16Linked to original sources

arXiv2022

All-optical graph representation learning using integrated diffractive photonic computing units

Photonic neural networks perform brain-inspired computations using photons instead of electrons that can achieve substantially improved computing performance. However, existing architectures can only handle data with regular structures, e.g., images or videos, but fail to generalize to graph-structured data beyond Euclidean space, e.g., social networks or document co-citation networks. Here, we propose an all-optical graph representation learning architecture, termed diffractive graph neural network (DGNN), based on the integrated diffractive photonic computing units (DPUs) to address this limitation. Specifically, DGNN optically encodes node attributes into strip optical waveguides, which are transformed by DPUs and aggregated by on-chip optical couplers to extract their feature representations. Each DPU comprises successive passive layers of metalines to modulate the electromagnetic optical field via diffraction, where the metaline structures are learnable parameters shared across graph nodes. DGNN captures complex dependencies among the node neighborhoods and eliminates the nonlinear transition functions during the light-speed optical message passing over graph structures. We demonstrate the use of DGNN extracted features for node and graph-level classification tasks with benchmark databases and achieve superior performance. Our work opens up a new direction for designing application-specific integrated photonic circuits for high-efficiency processing of large-scale graph data structures using deep learning.

cs.ET↗

arXiv2022

Customized quantum annealing schedules

In a typical quantum annealing protocol, the system starts with a transverse field Hamiltonian which is gradually turned off and replaced by a longitudinal Ising Hamiltonian. The ground state of the Ising Hamiltonian encodes the solution to the computational problem of interest, and the state overlap with this ground state gives the success probability of the annealing protocol. The form of the annealing schedule can have a significant impact on the ground state overlap at the end of the anneal, so precise control over these annealing schedules can be a powerful tool for increasing success probabilities of annealing protocols. Here we show how superconducting circuits, in particular capacitively shunted flux qubits (CSFQs), can be used to construct quantum annealing systems by providing tools for mapping circuit flux biases to Pauli coefficients. We use this mapping to find customized annealing schedules: appropriate circuit control biases that yield a desired annealing schedule, while accounting for the physical limitations of the circuitry. We then provide examples and proposals that utilize this capability to improve quantum annealing performance.

quant-ph↗

arXiv2022

Intelligent Feedback Overhead Reduction (iFOR) in Wi-Fi 7 and Beyond

The IEEE 802.11 standard based wireless local area networks (WLANs) or Wi-Fi networks are critical to provide internet access in today's world. The increasing demand for high data rate in Wi-Fi networks has led to several advancements in the 802.11 standard. Supporting MIMO transmissions with higher number of transmit antennas operating on wider bandwidths is one of the key capabilities for reaching higher throughput. However, the increase in sounding feedback overhead due to higher number of transmit antennas may significantly curb the throughput gain. In this paper, we develop an unsupervised learning-based method to reduce the sounding duration in a Wi-Fi MIMO link. Simulation results show that our method uses approximately only 8% of the number of bits required by the existing feedback mechanism and it can boost the system throughput by up to 52%.

eess.SP↗

arXiv2022

Safe Learning-Based Feedback Linearization Tracking Control for Nonlinear System with Event-Triggered Model Update

Learning-based methods are powerful in handling complex scenarios. However, it is still challenging to use learning-based methods under uncertain environments while stability, safety, and real-time performance of the system are desired to guarantee. In this paper, we propose a learning-based tracking control scheme based on a feedback linearization controller in which uncertain disturbances are approximated online using Gaussian Processes (GPs). Using the predicted distribution of disturbances given by GPs, a Control Lyapunov Function (CLF) and Control Barrier Function (CBF) based Quadratic Program is applied, with which probabilistic stability and safety are guaranteed. In addition, the trajectory is optimized first by Model Predictive Control (MPC) based on the linearized dynamics systems to further reduce the tracking error. We also design an event trigger for GPs updates to improve efficiency while stability and safety of the system are still guaranteed. The effectiveness of the proposed tracking control strategy is illustrated in numerical simulations.

cs.RO↗

arXiv2022

Banyan: A Scoped Dataflow Engine for Graph Query Service

Graph query services (GQS) are widely used today to interactively answer graph traversal queries on large-scale graph data. Existing graph query engines focus largely on optimizing the latency of a single query. This ignores significant challenges posed by GQS, including fine-grained control and scheduling during query execution, as well as performance isolation and load balancing in various levels from across user to intra-query. To tackle these control and scheduling challenges, we propose a novel scoped dataflow for modeling graph traversal queries, which explicitly exposes concurrent execution and control of any subquery to the finest granularity. We implemented Banyan, an engine based on the scoped dataflow model for GQS. Banyan focuses on scaling up the performance on a single machine, and provides the ability to easily scale out. Extensive experiments on multiple benchmarks show that Banyan improves performance by up to three orders of magnitude over state-of-the-art graph query engines, while providing performance isolation and load balancing.

cs.DB↗

arXiv2022

Safe Learning-based Gradient-free Model Predictive Control Based on Cross-entropy Method

In this paper, a safe and learning-based control framework for model predictive control (MPC) is proposed to optimize nonlinear systems with a non-differentiable objective function under uncertain environmental disturbances. The control framework integrates a learning-based MPC with an auxiliary controller in a way of minimal intervention. The learning-based MPC augments the prior nominal model with incremental Gaussian Processes to learn the uncertain disturbances. The cross-entropy method (CEM) is utilized as the sampling-based optimizer for the MPC with a non-differentiable objective function. A minimal intervention controller is devised with a control Lyapunov function and a control barrier function to guide the sampling process and endow the system with high probabilistic safety. The proposed algorithm shows a safe and adaptive control performance on a simulated quadrotor in the tasks of trajectory tracking and obstacle avoidance under uncertain wind disturbances.

cs.RO↗

arXiv2022

Rethinking Goal-conditioned Supervised Learning and Its Connection to Offline RL

Solving goal-conditioned tasks with sparse rewards using self-supervised learning is promising because of its simplicity and stability over current reinforcement learning (RL) algorithms. A recent work, called Goal-Conditioned Supervised Learning (GCSL), provides a new learning framework by iteratively relabeling and imitating self-generated experiences. In this paper, we revisit the theoretical property of GCSL -- optimizing a lower bound of the goal reaching objective, and extend GCSL as a novel offline goal-conditioned RL algorithm. The proposed method is named Weighted GCSL (WGCSL), in which we introduce an advanced compound weight consisting of three parts (1) discounted weight for goal relabeling, (2) goal-conditioned exponential advantage weight, and (3) best-advantage weight. Theoretically, WGCSL is proved to optimize an equivalent lower bound of the goal-conditioned RL objective and generates monotonically improved policies via an iterated scheme. The monotonic property holds for any behavior policies, and therefore WGCSL can be applied to both online and offline settings. To evaluate algorithms in the offline goal-conditioned RL setting, we provide a benchmark including a range of point and simulated robot domains. Experiments in the introduced benchmark demonstrate that WGCSL can consistently outperform GCSL and existing state-of-the-art offline methods in the fully offline goal-conditioned setting.

cs.LG↗

arXiv2022

A two-qubit entangling gate based on a two-spin gadget

The faster speed and operational convenience of two-qubit gate with flux bias control makes it an important candidate for future large-scale quantum computers based on high coherence flux qubits. Based on a properly designed two-spin gadget which has small gaps during the evolution of energy levels, we build a CNOT-equivalent gate which can reach a fidelity larger than 99.9% within 40ns. Moreover, we also use the Schrieffer-Wolff Transformation to translate the spin model Ising coefficients schedule to circuit model flux bias schedule for realistic flux qubit circuits coupled by a tunable rf-squid. The two-qubit entangling gate scheme introduced here is suitable for realizing efficient two-qubit gates in the large scale flux qubit systems dominated by inductive couplings. Comparing with the current gate-based quantum computation systems dominated by capacitive couplings, it can resolve the conflict between the speed and a high coherence.

quant-ph↗

arXiv2021

Radiomic biomarker extracted from PI-RADS 3 patients support more eìcient and robust prostate cancer diagnosis: a multi-center study

Prostate Imaging Reporting and Data System (PI-RADS) based on multi-parametric MRI classiêes patients into 5 categories (PI-RADS 1-5) for routine clinical diagnosis guidance. However, there is no consensus on whether PI-RADS 3 patients should go through biopsies. Mining features from these hard samples (HS) is meaningful for physicians to achieve accurate diagnoses. Currently, the mining of HS biomarkers is insuìcient, and the eéectiveness and robustness of HS biomarkers for prostate cancer diagnosis have not been explored. In this study, biomarkers from diéerent data distributions are constructed. Results show that HS biomarkers can achieve better performances in diéerent data distributions.

eess.IV↗

arXiv2021

MHER: Model-based Hindsight Experience Replay

Solving multi-goal reinforcement learning (RL) problems with sparse rewards is generally challenging. Existing approaches have utilized goal relabeling on collected experiences to alleviate issues raised from sparse rewards. However, these methods are still limited in efficiency and cannot make full use of experiences. In this paper, we propose Model-based Hindsight Experience Replay (MHER), which exploits experiences more efficiently by leveraging environmental dynamics to generate virtual achieved goals. Replacing original goals with virtual goals generated from interaction with a trained dynamics model leads to a novel relabeling method, model-based relabeling (MBR). Based on MBR, MHER performs both reinforcement learning and supervised learning for efficient policy improvement. Theoretically, we also prove the supervised part in MHER, i.e., goal-conditioned supervised learning with MBR data, optimizes a lower bound on the multi-goal RL objective. Experimental results in several point-based tasks and simulated robotics environments show that MHER achieves significantly higher sample efficiency than previous model-free and model-based multi-goal methods.

cs.LG↗

arXiv2021

Demonstration of long-range correlations via susceptibility measurements in a one-dimensional superconducting Josephson spin chain

Spin chains have long been considered an effective medium for long-range interactions, entanglement generation, and quantum state transfer. In this work, we explore the properties of a spin chain implemented with superconducting flux circuits, designed to act as a connectivity medium between two superconducting qubits. The susceptibility of the chain is probed and shown to support long-range, cross chain correlations. In addition, interactions between the two end qubits, mediated by the coupler chain, are demonstrated. This work has direct applicability in near term quantum annealing processors as a means of generating long-range, coherent coupling between qubits.

quant-ph↗

arXiv2021

Annotation-efficient deep learning for automatic medical image segmentation

Automatic medical image segmentation plays a critical role in scientific research and medical care. Existing high-performance deep learning methods typically rely on large training datasets with high-quality manual annotations, which are difficult to obtain in many clinical applications. Here, we introduce Annotation-effIcient Deep lEarning (AIDE), an open-source framework to handle imperfect training datasets. Methodological analyses and empirical evaluations are conducted, and we demonstrate that AIDE surpasses conventional fully-supervised models by presenting better performance on open datasets possessing scarce or noisy annotations. We further test AIDE in a real-life case study for breast tumor segmentation. Three datasets containing 11,852 breast images from three medical centers are employed, and AIDE, utilizing 10% training annotations, consistently produces segmentation maps comparable to those generated by fully-supervised counterparts or provided by independent radiologists. The 10-fold enhanced efficiency in utilizing expert labels has the potential to promote a wide range of biomedical applications.

eess.IV↗

arXiv2021

Experimental Quantum Generative Adversarial Networks for Image Generation

Quantum machine learning is expected to be one of the first practical applications of near-term quantum devices. Pioneer theoretical works suggest that quantum generative adversarial networks (GANs) may exhibit a potential exponential advantage over classical GANs, thus attracting widespread attention. However, it remains elusive whether quantum GANs implemented on near-term quantum devices can actually solve real-world learning tasks. Here, we devise a flexible quantum GAN scheme to narrow this knowledge gap, which could accomplish image generation with arbitrarily high-dimensional features, and could also take advantage of quantum superposition to train multiple examples in parallel. For the first time, we experimentally achieve the learning and generation of real-world hand-written digit images on a superconducting quantum processor. Moreover, we utilize a gray-scale bar dataset to exhibit the competitive performance between quantum GANs and the classical GANs based on multilayer perceptron and convolutional neural network architectures, respectively, benchmarked by the Fréchet Distance score. Our work provides guidance for developing advanced quantum generative models on near-term quantum devices and opens up an avenue for exploring quantum advantages in various GAN-related learning tasks.

quant-ph↗

arXiv2021

Heat transfer in turbulent Rayleigh-Bénard convection within two immiscible fluid layers

We numerically investigate turbulent Rayleigh-Bénard convection within two immiscible fluid layers, aiming to understand how the layer thickness and fluid properties affect the heat transfer (characterized by the Nusselt number $Nu$) in two-layer systems. Both two- and three-dimensional simulations are performed at fixed global Rayleigh number $Ra=10^8$, Prandtl number $Pr=4.38$, and Weber number $We=5$. We vary the relative thickness of the upper layer between $0.01 \le α\le 0.99$ and the thermal conductivity coefficient ratio of the two liquids between $0.1 \le λ_k \le 10$. Two flow regimes are observed: In the first regime at $0.04\leα\le0.96$, convective flows appear in both layers and $Nu$ is not sensitive to $α$. In the second regime at $α\le0.02$ or $α\ge0.98$, convective flow only exists in the thicker layer, while the thinner one is dominated by pure conduction. In this regime, $Nu$ is sensitive to $α$. To predict $Nu$ in the system in which the two layers are separated by a unique interface, we apply the Grossmann-Lohse theory for both individual layers and impose heat flux conservation at the interface. Without introducing any free parameter, the predictions for $Nu$ and for the temperature at the interface well agree with our numerical results and previous experimental data.

physics.flu-dyn↗

arXiv2021

A Generic Complementary Sequence Construction and Associated Encoder/Decoder Design

In this study, we propose a flexible construction of complementary sequences (CSs) that can contain zero-valued elements. To derive the construction, we use Boolean functions to represent a polynomial generated with a recursion. By applying this representation to recursive CS constructions, we show the impact of construction parameters such as sign, amplitude, phase rotation used in the recursion on the elements of the synthesized CS. As a result, we extend Davis and Jedwab's CS construction by obtaining independent functions for the amplitude and phase of each element of the CS, and the seed sequence positions in the CS. The proposed construction shows that a set of distinct CSs compatible with non-contiguous resource allocations for orthogonal frequency-division multiplexing (OFDM) and various constellations can be synthesized systematically. It also leads to a low peak-to-mean-envelope-power ratio (PMEPR) multiple accessing scheme in the uplink and a low-complexity recursive decoder. We demonstrate the performance of the proposed encoder and decoder through comprehensive simulations.

cs.IT↗

arXiv2021

Efficient Online Transfer Learning for 3D Object Classification in Autonomous Driving

Autonomous driving has achieved rapid development over the last few decades, including the machine perception as an important issue of it. Although object detection based on conventional cameras has achieved remarkable results in 2D/3D, non-visual sensors such as 3D LiDAR still have incomparable advantages in the accuracy of object position detection. However, the challenge also exists with the difficulty in properly interpreting point cloud generated by LiDAR. This paper presents a multi-modal-based online learning system for 3D LiDAR-based object classification in urban environments, including cars, cyclists and pedestrians. The proposed system aims to effectively transfer the mature detection capabilities based on visual sensors to the new model learning based on non-visual sensors through a multi-target tracker (i.e. using one sensor to train another). In particular, it integrates the Online Random Forests (ORF) [1] method, which inherently has the abilities of fast and multi-class learning. Through experiments, we show that our system is capable of learning a high-performance model for LiDAR-based 3D object classification on-the-fly, which is especially suitable for robotics in-situ deployment while responding to the widespread challenge of insufficient detector generalization capabilities.

cs.RO↗

arXiv2021

Atomic transitions for adaptive optics

This paper reviews atoms and ions in the upper atmosphere, including the mesospheric metals Na, Fe, Mg$^+$, Si$^+$, Ca$^+$, K and also non-metallic species N, N$^+$, O, H, considering their potential for astronomical adaptive optics. Na and Fe are the best candidates for the creation of polychromatic laser guide stars, with the strongest returns coming from transitions that can be reached by excitation at two wavelengths. Ca$^+$ and Si$^+$ have strong visible-light transitions, but require short wavelengths, beyond the atmospheric cutoff, for excitation from the ground state. Atomic O, N and N$^+$ have strong transitions and high abundances in the mesosphere. The product of column density and cross section for these species can be as high as $10^5$ for O and several hundred for N and N$^+$, making them potential candidates for amplified spontaneous emission. However they require vacuum-ultraviolet wavelengths for excitation.

astro-ph.IM↗

arXiv2021

Thermoelectric response from grain boundaries and lattice distortions in crystalline gold devices

The electronic Seebeck response in a conductor involves the energy-dependent mean free path of the charge carriers and is affected by crystal structure, scattering from boundaries and defects, and strain. Previous photothermoelectric (PTE) studies have suggested that the thermoelectric properties of polycrystalline metal nanowires are related to grain structure, though direct evidence linking crystal microstructure to the PTE response is difficult to elucidate. Here, we show that room temperature scanning PTE measurements are sensitive probes that can detect subtle changes in the local Seebeck coefficient of gold tied to the underlying defects and strain that mediate crystal deformation. This connection is revealed through a combination of scanning PTE and electron microscopy measurements of single crystal and bicrystal gold microscale devices. Unexpectedly, the photovoltage maps strongly correlate with gradually varying crystallographic misorientations detected by electron backscatter diffraction. The effects of individual grain boundaries and differing grain orientations on the PTE signal are minimal. This scanning PTE technique shows promise for identifying minor structural distortions in nanoscale materials and devices.

cond-mat.mtrl-sci↗