arXiv subjects

Yuan Liu

Publications and source records attributed to Yuan Liu.

At least 415 records · Page 23Linked to original sources

arXiv2021

A Contract Theory based Incentive Mechanism for Federated Learning

Federated learning (FL) serves as a data privacy-preserved machine learning paradigm, and realizes the collaborative model trained by distributed clients. To accomplish an FL task, the task publisher needs to pay financial incentives to the FL server and FL server offloads the task to the contributing FL clients. It is challenging to design proper incentives for the FL clients due to the fact that the task is privately trained by the clients. This paper aims to propose a contract theory based FL task training model towards minimizing incentive budget subject to clients being individually rational (IR) and incentive compatible (IC) in each FL training round. We design a two-dimensional contract model by formally defining two private types of clients, namely data quality and computation effort. To effectively aggregate the trained models, a contract-based aggregator is proposed. We analyze the feasible and optimal contract solutions to the proposed contract model. %Experimental results demonstrate that the proposed framework and contract model can effective improve the generation accuracy of FL tasks. Experimental results show that the generalization accuracy of the FL tasks can be improved by the proposed incentive mechanism where contract-based aggregation is applied.

cs.DC↗

arXiv2021

AdaFit: Rethinking Learning-based Normal Estimation on Point Clouds

This paper presents a neural network for robust normal estimation on point clouds, named AdaFit, that can deal with point clouds with noise and density variations. Existing works use a network to learn point-wise weights for weighted least squares surface fitting to estimate the normals, which has difficulty in finding accurate normals in complex regions or containing noisy points. By analyzing the step of weighted least squares surface fitting, we find that it is hard to determine the polynomial order of the fitting surface and the fitting surface is sensitive to outliers. To address these problems, we propose a simple yet effective solution that adds an additional offset prediction to improve the quality of normal estimation. Furthermore, in order to take advantage of points from different neighborhood sizes, a novel Cascaded Scale Aggregation layer is proposed to help the network predict more accurate point-wise offsets and weights. Extensive experiments demonstrate that AdaFit achieves state-of-the-art performance on both the synthetic PCPNet dataset and the real-word SceneNN dataset.

cs.CV↗

arXiv2021

How Correlated Adsorbate Dynamics on Realistic Substrates Can Give Rise to 1/ω Electric-Field Noise in Surface Ion Traps

Ion traps are promising architectures for implementing scalable quantum computing, but they suffer from excessive "anomalous" heating that prevents their full potential from being realized. This heating, which is orders of magnitude larger than that expected from Johnson-Nyquist noise, results in ion motion that leads to decoherence and reduced fidelity in quantum logic gates. The exact origin of anomalous heating is an open question, but experiments point to adsorbates on trap electrodes as a likely source. Many different models of anomalous heating have been proposed, but these models have yet to pinpoint the atomistic origin of the experimentally-observed $1/ω$ electric field noise scaling observed in ion traps at frequencies between 0.1-10 MHz. In this work, we perform the first computational study of the ion trap electric field noise produced by the motions of multiple monolayers of adsorbates described by first principles potentials. In so doing, we show that correlated adsorbate motions play a definitive role in producing $1/ω$ noise and identify candidate collective adsorbate motions, including translational and rotational motions of adsorbate patches and multilayer exchanges, that give rise to $1/ω$ scaling at the MHz frequencies typically employed in ion traps. These results demonstrate that multi-adsorbate systems, even simple ones, can give rise to a set of activated motions that can produce the $1/ω$ noise observed in ion traps and that collective, rather than individual, adsorbate motions are much more likely to give rise to low-frequency heating.

quant-ph↗

arXiv2021

Off-axis gyration induces large-area circular motion of anisotropic microparticles in a dynamic magnetic trap

Magnetic tweezers are crucial for single-molecule and atomic characterization, and biomedical isolation of microparticle carriers. The trapping component of magnetic tweezing can be relying on a magnetic potential well that can confine the relevant species to a localized region. Here, we report that magnetic microparticles with tailored anisotropy can transition from localized off-axis gyration to large-area locomotion in a rotating magnetic trap. The microparticles, consisting of assemblies of magnetic cores, are observed to either rotate about its structural geometric center or gyrate about one of the magnetic cores, the switching of which can be modulated by the external field. Raising the magnetic field strength above a threshold, the particles can go beyond the traditional synchronous-rotation and asynchronous-oscillation modes, and into a scenario of large-area circular motion. This results in peculiar retrograde locomotion related to the magnetization maxima of the microparticle. Our finding suggests the important role of the microparticle's magnetic morphology in the controlled transport of microparticles and developing smart micro-actuators and micro-robot devices.

physics.app-ph↗

arXiv2021

Large violations in Kochen Specker contextuality and their applications

The Kochen-Specker (KS) theorem is a fundamental result in quantum foundations that has spawned massive interest since its inception. We present state-independent non-contextuality inequalities with large violations, in particular, we exploit a connection between Kochen-Specker proofs and pseudo-telepathy games to show KS proofs in Hilbert spaces of dimension $d \geq 2^{17}$ with the ratio of quantum value to classical bias being $O(\sqrt{d}/\log d)$. We study the properties of this KS set and show applications of the large violation. It has been recently shown that Kochen-Specker proofs always consist of substructures of state-dependent contextuality proofs called $01$-gadgets or bugs. We show a one-to-one connection between $01$-gadgets in $\mathbb{C}^d$ and Hardy paradoxes for the maximally entangled state in $\mathbb{C}^d \otimes \mathbb{C}^d$. We use this connection to construct large violation $01$-gadgets between arbitrary vectors in $\mathbb{C}^d$, as well as novel Hardy paradoxes for the maximally entangled state in $\mathbb{C}^d \otimes \mathbb{C}^d$, and give applications of these constructions. As a technical result, we show that the minimum dimension of the faithful orthogonal representation of a graph in $\mathbb{R}^d$ is not a graph monotone, a result that that may be of independent interest.

quant-ph↗

arXiv2021

A class of adaptive multiresolution ultra-weak discontinuous Galerkin methods for some nonlinear dispersive wave equations

In this paper, we propose a class of adaptive multiresolution (also called adaptive sparse grid) ultra-weak discontinuous Galerkin (UWDG) methods for solving some nonlinear dispersive wave equations including the Korteweg-de Vries (KdV) equation and its two dimensional generalization, the Zakharov-Kuznetsov (ZK) equation. The UWDG formulation, which relies on repeated integration by parts, was proposed for KdV equation in \cite{cheng2008discontinuous}. For the ZK equation which contains mixed derivative terms, we develop a new UWDG formulation. The $L^2$ stability and the optimal error estimate with a novel local projection are established for this new scheme on regular meshes. Adaptivity is achieved based on multiresolution and is particularly effective for capturing solitary wave structures. Various numerical examples are presented to demonstrate the accuracy and capability of our methods.

math.NA↗

arXiv2021

Does Your Dermatology Classifier Know What It Doesn't Know? Detecting the Long-Tail of Unseen Conditions

We develop and rigorously evaluate a deep learning based system that can accurately classify skin conditions while detecting rare conditions for which there is not enough data available for training a confident classifier. We frame this task as an out-of-distribution (OOD) detection problem. Our novel approach, hierarchical outlier detection (HOD) assigns multiple abstention classes for each training outlier class and jointly performs a coarse classification of inliers vs. outliers, along with fine-grained classification of the individual classes. We demonstrate the effectiveness of the HOD loss in conjunction with modern representation learning approaches (BiT, SimCLR, MICLe) and explore different ensembling strategies for further improving the results. We perform an extensive subgroup analysis over conditions of varying risk levels and different skin types to investigate how the OOD detection performance changes over each subgroup and demonstrate the gains of our framework in comparison to baselines. Finally, we introduce a cost metric to approximate downstream clinical impact. We use this cost metric to compare the proposed method against a baseline system, thereby making a stronger case for the overall system effectiveness in a real-world deployment scenario.

cs.CV↗

arXiv2021

Point2Skeleton: Learning Skeletal Representations from Point Clouds

We introduce Point2Skeleton, an unsupervised method to learn skeletal representations from point clouds. Existing skeletonization methods are limited to tubular shapes and the stringent requirement of watertight input, while our method aims to produce more generalized skeletal representations for complex structures and handle point clouds. Our key idea is to use the insights of the medial axis transform (MAT) to capture the intrinsic geometric and topological natures of the original input points. We first predict a set of skeletal points by learning a geometric transformation, and then analyze the connectivity of the skeletal points to form skeletal mesh structures. Extensive evaluations and comparisons show our method has superior performance and robustness. The learned skeletal representation will benefit several unsupervised tasks for point clouds, such as surface reconstruction and segmentation.

cs.CV↗

arXiv2021

The Blessings of Unlabeled Background in Untrimmed Videos

Weakly-supervised Temporal Action Localization (WTAL) aims to detect the action segments with only video-level action labels in training. The key challenge is how to distinguish the action of interest segments from the background, which is unlabelled even on the video-level. While previous works treat the background as "curses", we consider it as "blessings". Specifically, we first use causal analysis to point out that the common localization errors are due to the unobserved confounder that resides ubiquitously in visual recognition. Then, we propose a Temporal Smoothing PCA-based (TS-PCA) deconfounder, which exploits the unlabelled background to model an observed substitute for the unobserved confounder, to remove the confounding effect. Note that the proposed deconfounder is model-agnostic and non-intrusive, and hence can be applied in any WTAL method without model re-designs. Through extensive experiments on four state-of-the-art WTAL methods, we show that the deconfounder can improve all of them on the public datasets: THUMOS-14 and ActivityNet-1.3.

cs.CV↗

arXiv2021

The first photometric analysis of the totally eclipsing contact binary V811 Cep

The first photometric analysis of V811 Cep was carried out. The first complete light curves of V, R and I bands are given. The analysis was carried out by Wilson-Devinney (W-D) program, and the results show that V811 Cep is a median-contact binary ($f=33.9(\pm4.9)\%$) with a mass ratio of 0.285. It is a W-subtype contact binary, that is, the component with less mass is hotter than the component with more mass, and the light curves are asymmetric (O'Connell effect), which can be explained by the existence of a hot spot on the component with less mass. The orbital inclination is $i=88.3^{\circ}$, indicating that it is a totally eclipsing binary, so the parameters obtained are reliable. Through the O-C analyzing, it is found that the orbital period decreases at the rate of $\dot{P}=-3.90(\pm0.06)\times 10^{-7}d \cdot yr^{-1}$, which indicates that the mass transfer occurs from the more massive component to the less massive one.

astro-ph.SR↗

arXiv2021

Multi-Cell Mobile Edge Computing: Joint Service Migration and Resource Allocation

Mobile-edge computing (MEC) enhances the capacities and features of mobile devices by offloading computation-intensive tasks over wireless networks to edge servers. One challenge faced by the deployment of MEC in cellular networks is to support user mobility. As a result, offloaded tasks can be seamlessly migrated between base stations (BSs) without compromising the resource-utilization efficiency and link reliability. In this paper, we tackle the challenge by optimizing the policy for migration/handover between BSs by jointly managing computation-and-radio resources. The objectives are twofold: maximizing the sum offloading rate, quantifying MEC throughput, and minimizing the migration cost. The policy design is formulated as a decision-optimization problem that accounts for virtualization, I/O interference between virtual machines (VMs), and wireless multi-access. To solve the complex combinatorial problem, we develop an efficient relaxation-and-rounding based solution approach. The approach relies on an optimal iterative algorithm for solving the integer-relaxed problem and a novel integer-recovery design. The latter outperforms the traditional rounding method by exploiting the derived problem properties and applying matching theory. In addition, we also consider the design for a special case of "hotspot mitigation", referring to alleviating an overloaded server/BS by migrating its load to the nearby idle servers/BSs. From simulation results, we observed close-to-optimal performance of the proposed migration policies under various settings. This demonstrates their efficiency in computation-and-radio resource management for joint service migration and BS handover in multi-cell MEC networks.

cs.IT↗

arXiv2021

Supervised Transfer Learning at Scale for Medical Imaging

Transfer learning is a standard technique to improve performance on tasks with limited data. However, for medical imaging, the value of transfer learning is less clear. This is likely due to the large domain mismatch between the usual natural-image pre-training (e.g. ImageNet) and medical images. However, recent advances in transfer learning have shown substantial improvements from scale. We investigate whether modern methods can change the fortune of transfer learning for medical imaging. For this, we study the class of large-scale pre-trained networks presented by Kolesnikov et al. on three diverse imaging tasks: chest radiography, mammography, and dermatology. We study both transfer performance and critical properties for the deployment in the medical domain, including: out-of-distribution generalization, data-efficiency, sub-group fairness, and uncertainty estimation. Interestingly, we find that for some of these properties transfer from natural to medical images is indeed extremely effective, but only when performed at sufficient scale.

cs.CV↗

arXiv2021

Noise Is Useful: Exploiting Data Diversity for Edge Intelligence

Edge intelligence requires to fast access distributed data samples generated by edge devices. The challenge is using limited radio resource to acquire massive data samples for training machine learning models at edge server. In this article, we propose a new communication-efficient edge intelligence scheme where the most useful data samples are selected to train the model. Here the usefulness or values of data samples is measured by data diversity which is defined as the difference between data samples. We derive a close-form expression of data diversity that combines data informativeness and channel quality. Then a joint data-and-channel diversity aware multiuser scheduling algorithm is proposed. We find that noise is useful for enhancing data diversity under some conditions.

cs.IT↗

arXiv2020

The first photometric analysis and period investigation of the K-type W UMa type binary system V0842 Cep

V0842 Cep is a W UMa-type binary star that has been neglected since its discovery. We analysed the VR$_c$I$_c$ light curves, obtained by the 1 m telescope at the Weihai Observatory of Shandong University, using the Wilson-Devinney code. V0842 Cep was found to be a shallow contact binary system (f=8.7$\%$) with a mass ratio of 2.281. Because its orbital inclination is greater than 80$^\circ$, the photometric results are reliable. A period study is included which reveals a continually decreasing orbital period ($\frac{\mathrm{d}p}{\mathrm{d}t}$=1.50($\pm$0.42)$\times$10$^{-7}$dyr$^{-1}$). This trend could be attributed to the angular momentum loss via stellar wind.

astro-ph.SR↗

arXiv2020

A software decoder implementation for H.266/VVC video coding standard

Versatile Video Coding Standard (H.266/VVC) was completed by Joint Video Expert Team (JVET) of ITU-T and ISO/IEC, in July 2020. This new ITU recommendation/international standard is a successor to the well-known H.265/HEVC video coding standard with roughly doubled compression efficiency, but also at the cost of an increased computational complexity. The complexity of H.266/VVC decoder processing modules is studied in this paper. An optimized decoder implementation using SIMD instruction extensions and additional parallel processing including data and task level parallelism is presented, which can achieve real-time decoding of 4K 60fps VVC bitstreams on an x86 based CPU.

eess.IV↗

arXiv2020

Accelerated Deep Reinforcement Learning Based Load Shedding for Emergency Voltage Control

Load shedding has been one of the most widely used and effective emergency control approaches against voltage instability. With increased uncertainties and rapidly changing operational conditions in power systems, existing methods have outstanding issues in terms of either speed, adaptiveness, or scalability. Deep reinforcement learning (DRL) was regarded and adopted as a promising approach for fast and adaptive grid stability control in recent years. However, existing DRL algorithms show two outstanding issues when being applied to power system control problems: 1) computational inefficiency that requires extensive training and tuning time; and 2) poor scalability making it difficult to scale to high dimensional control problems. To overcome these issues, an accelerated DRL algorithm named PARS was developed and tailored for power system voltage stability control via load shedding. PARS features high scalability and is easy to tune with only five main hyperparameters. The method was tested on both the IEEE 39-bus and IEEE 300-bus systems, and the latter is by far the largest scale for such a study. Test results show that, compared to other methods including model-predictive control (MPC) and proximal policy optimization(PPO) methods, PARS shows better computational efficiency (faster convergence), more robustness in learning, excellent scalability and generalization capability.

eess.SY↗

arXiv2020

Learnable Motion Coherence for Correspondence Pruning

Motion coherence is an important clue for distinguishing true correspondences from false ones. Modeling motion coherence on sparse putative correspondences is challenging due to their sparsity and uneven distributions. Existing works on motion coherence are sensitive to parameter settings and have difficulty in dealing with complex motion patterns. In this paper, we introduce a network called Laplacian Motion Coherence Network (LMCNet) to learn motion coherence property for correspondence pruning. We propose a novel formulation of fitting coherent motions with a smooth function on a graph of correspondences and show that this formulation allows a closed-form solution by graph Laplacian. This closed-form solution enables us to design a differentiable layer in a learning framework to capture global motion coherence from putative correspondences. The global motion coherence is further combined with local coherence extracted by another local layer to robustly detect inlier correspondences. Experiments demonstrate that LMCNet has superior performances to the state of the art in relative camera pose estimation and correspondences pruning of dynamic scenes.

cs.CV↗

arXiv2020

Observation of two-dimensional superconductivity at the LaAlO3/KTaO3(110) heterointerface

We report on the observation of a Tc ~0.9 K superconductivity at the interface between LaAlO3 film and the 5d transition-metal oxide KTaO3(110) single crystal. The interface shows a large anisotropy of the upper critical field, and its superconducting transition is consistent with a Berezinskii-Kosterlitz-Thouless transition. Both facts suggest that the superconductivity is two-dimensional (2D) in nature. The carrier density measured at 5 K is ~7 time 10^13 cm-2. The superconducting layer thickness and coherence length are estimated to be ~8 and ~30 nm, respectively. Our result provides a new platform for the study of 2D superconductivity at oxide interfaces.

cond-mat.supr-con↗