arXiv subjects

Yang Peng

Publications and source records attributed to Yang Peng.

At least 55 records · Page 3Linked to original sources

arXiv2023

Onset mechanism of an inverted U-shaped solar filament eruption revealed by NVST, SDO, and STEREO-A observations

Utilizing observations from the New Vacuum Solar Telescope (NVST), Solar Dynamics Observatory (SDO), and Solar Terrestrial Relations Observatory-Ahead (STEREO-A), we investigate the event from two distinct observational perspectives: on the solar disk using NVST and SDO, and on the solar limb using STEREO-A. We employ both a non-linear force-free field model and a potential field model to reconstruct the coronal magnetic field, aiming to understand its magnetic properties. Two precursor jet-like activities were observed before the eruption, displaying an untwisted rotation. The second activity released an estimated twist of over two turns. During these two jet-like activities, Y-shaped brightenings, newly emerging magnetic flux accompanied by magnetic cancellation, and the formation of newly moving fibrils were identified. Combining these observational features, it can be inferred that these two precursor jet-like activities released the magnetic field constraining the filament and were triggered by newly emerging magnetic flux. Before the filament eruption, it was observed that some moving flows had been ejected from the site as the onset of two jet-like activities, indicating the same physical process as two jet-like activities. Extrapolations revealed that the filament laid under the height of the decay index of 1.0 and had strong magnetic field (540 Gauss) and a high twisted number (2.4 turns) before the eruption. An apparent rotational motion was observed during the filament eruption. We deduce that the solar filament, exhibiting an inverted U-shape, is a significantly twisted flux rope. The eruption of the filament was initiated by the release of constraining magnetic fields through continuous magnetic reconnection. This reconnection process was triggered by the emergence of newly magnetic flux.

astro-ph.SR↗

arXiv2023

Gradient Tracking for High Dimensional Federated Optimization

In this paper, we study the (decentralized) distributed optimization problem with high-dimensional sparse structure. Building upon the FedDA algorithm, we propose a (Decentralized) FedDA-GT algorithm, which combines the \textbf{gradient tracking} technique. It is able to eliminate the heterogeneity among different clients' objective functions while ensuring a dimension-free convergence rate. Compared to the vanilla FedDA approach, (D)FedDA-GT can significantly reduce the communication complexity, from ${O}(s^2\log d/\varepsilon^{3/2})$ to a more efficient ${O}(s^2\log d/\varepsilon)$. In cases where strong convexity is applicable, we introduce a multistep mechanism resulting in the Multistep ReFedDA-GT algorithm, a minor modified version of FedDA-GT. This approach achieves an impressive communication complexity of ${O}\left(s\log d \log \frac{1}{\varepsilon}\right)$ through repeated calls to the ReFedDA-GT algorithm. Finally, we conduct numerical experiments, illustrating that our proposed algorithms enjoy the dual advantage of being dimension-free and heterogeneity-free.

math.OC↗

arXiv2023

BEA: Revisiting anchor-based object detection DNN using Budding Ensemble Architecture

This paper introduces the Budding Ensemble Architecture (BEA), a novel reduced ensemble architecture for anchor-based object detection models. Object detection models are crucial in vision-based tasks, particularly in autonomous systems. They should provide precise bounding box detections while also calibrating their predicted confidence scores, leading to higher-quality uncertainty estimates. However, current models may make erroneous decisions due to false positives receiving high scores or true positives being discarded due to low scores. BEA aims to address these issues. The proposed loss functions in BEA improve the confidence score calibration and lower the uncertainty error, which results in a better distinction of true and false positives and, eventually, higher accuracy of the object detection models. Both Base-YOLOv3 and SSD models were enhanced using the BEA method and its proposed loss functions. The BEA on Base-YOLOv3 trained on the KITTI dataset results in a 6% and 3.7% increase in mAP and AP50, respectively. Utilizing a well-balanced uncertainty estimation threshold to discard samples in real-time even leads to a 9.6% higher AP50 than its base model. This is attributed to a 40% increase in the area under the AP50-based retention curve used to measure the quality of calibration of confidence scores. Furthermore, BEA-YOLOV3 trained on KITTI provides superior out-of-distribution detection on Citypersons, BDD100K, and COCO datasets compared to the ensembles and vanilla models of YOLOv3 and Gaussian-YOLOv3.

cs.CV↗

arXiv2023

Deep learning for 3D Object Detection and Tracking in Autonomous Driving: A Brief Survey

Object detection and tracking are vital and fundamental tasks for autonomous driving, aiming at identifying and locating objects from those predefined categories in a scene. 3D point cloud learning has been attracting more and more attention among all other forms of self-driving data. Currently, there are many deep learning methods for 3D object detection. However, the tasks of object detection and tracking for point clouds still need intensive study due to the unique characteristics of point cloud data. To help get a good grasp of the present situation of this research, this paper shows recent advances in deep learning methods for 3D object detection and tracking.

cs.CV↗

arXiv2023

Envisioning a Safety Island to Enable HPC Devices in Safety-Critical Domains

HPC (High Performance Computing) devices increasingly become the only alternative to deliver the performance needed in safety-critical autonomous systems (e.g., autonomous cars, unmanned planes) due to deploying large and powerful multicores along with accelerators such as GPUs. However, the support that those HPC devices offer to realize safety-critical systems on top is heterogeneous. Safety islands have been devised to be coupled to HPC devices and complement them to meet the safety requirements of an increased set of applications, yet the variety of concepts and realizations is large. This paper presents our own concept of a safety island with two goals in mind: (1) offering a wide set of features to enable the broadest set of safety applications for each HPC device, and (2) being realized with open source components based on RISC-V ISA to ease its use and adoption. In particular, we present our safety island concept, the key features we foresee it should include, and its potential application beyond safety.

cs.AR↗

arXiv2023

Finding Lookalike Customers for E-Commerce Marketing

Customer-centric marketing campaigns generate a large portion of e-commerce website traffic for Walmart. As the scale of customer data grows larger, expanding the marketing audience to reach more customers is becoming more critical for e-commerce companies to drive business growth and bring more value to customers. In this paper, we present a scalable and efficient system to expand targeted audience of marketing campaigns, which can handle hundreds of millions of customers. We use a deep learning based embedding model to represent customers and an approximate nearest neighbor search method to quickly find lookalike customers of interest. The model can deal with various business interests by constructing interpretable and meaningful customer similarity metrics. We conduct extensive experiments to demonstrate the great performance of our system and customer embedding model.

cs.LG↗

arXiv2023

Distinct Floquet topological classifications from color-decorated frequency lattices with space-time symmetries

We consider nontrivial topological phases in Floquet systems using unitary loops and stroboscopic evolutions under a static Floquet Hamiltonian $H_F$ in the presence of dynamical space-time symmetries $G$. While the latter has been subject of out-of-equilibrium classifications that extend the ten-fold way and systems with additional crystalline symmetries to periodically driven systems, we explore the anomalous topological zero modes that arise in $H_F$ from the coexistence of a dynamical space-time symmetry $M$ and antisymmetry $A$ of $G$, and classify them using a frequency-domain formulation. Moreover, we provide an interpretation of the resulting Floquet topological phases using a frequency lattice with a decoration represented by color degrees of freedom on the lattice vertices. These colors correspond to the coefficient $N$ of the group extension $\tilde{G}$ of $G$ along the frequency lattice, given by $N=Z\rtimes H^1[A,M]$. The distinct topological classifications that arise at different energy gaps in its quasi-energy spectrum are described by the torsion product of the cohomology group $H^{2}[G,N]$ classifying the group extension.

cond-mat.mes-hall↗

arXiv2023

Query-Driven Knowledge Base Completion using Multimodal Path Fusion over Multimodal Knowledge Graph

Over the past few years, large knowledge bases have been constructed to store massive amounts of knowledge. However, these knowledge bases are highly incomplete, for example, over 70% of people in Freebase have no known place of birth. To solve this problem, we propose a query-driven knowledge base completion system with multimodal fusion of unstructured and structured information. To effectively fuse unstructured information from the Web and structured information in knowledge bases to achieve good performance, our system builds multimodal knowledge graphs based on question answering and rule inference. We propose a multimodal path fusion algorithm to rank candidate answers based on different paths in the multimodal knowledge graphs, achieving much better performance than question answering, rule inference and a baseline fusion algorithm. To improve system efficiency, query-driven techniques are utilized to reduce the runtime of our system, providing fast responses to user queries. Extensive experiments have been conducted to demonstrate the effectiveness and efficiency of our system.

cs.DB↗

arXiv2023

Knowledge Base Completion using Web-Based Question Answering and Multimodal Fusion

Over the past few years, large knowledge bases have been constructed to store massive amounts of knowledge. However, these knowledge bases are highly incomplete. To solve this problem, we propose a web-based question answering system system with multimodal fusion of unstructured and structured information, to fill in missing information for knowledge bases. To utilize unstructured information from the Web for knowledge base completion, we design a web-based question answering system using multimodal features and question templates to extract missing facts, which can achieve good performance with very few questions. To help improve extraction quality, the question answering system employs structured information from knowledge bases, such as entity types and entity-to-entity relatedness.

cs.AI↗

arXiv2023

Semi-Infinitely Constrained Markov Decision Processes and Efficient Reinforcement Learning

We propose a novel generalization of constrained Markov decision processes (CMDPs) that we call the \emph{semi-infinitely constrained Markov decision process} (SICMDP). Particularly, we consider a continuum of constraints instead of a finite number of constraints as in the case of ordinary CMDPs. We also devise two reinforcement learning algorithms for SICMDPs that we call SI-CRL and SI-CPO. SI-CRL is a model-based reinforcement learning algorithm. Given an estimate of the transition model, we first transform the reinforcement learning problem into a linear semi-infinitely programming (LSIP) problem and then use the dual exchange method in the LSIP literature to solve it. SI-CPO is a policy optimization algorithm. Borrowing the ideas from the cooperative stochastic approximation approach, we make alternative updates to the policy parameters to maximize the reward or minimize the cost. To the best of our knowledge, we are the first to apply tools from semi-infinitely programming (SIP) to solve constrained reinforcement learning problems. We present theoretical analysis for SI-CRL and SI-CPO, identifying their iteration complexity and sample complexity. We also conduct extensive numerical examples to illustrate the SICMDP model and demonstrate that our proposed algorithms are able to solve complex sequential decision-making tasks leveraging modern deep reinforcement learning techniques.

cs.LG↗

arXiv2022

Semi-Supervised Specific Emitter Identification Method Using Metric-Adversarial Training

Specific emitter identification (SEI) plays an increasingly crucial and potential role in both military and civilian scenarios. It refers to a process to discriminate individual emitters from each other by analyzing extracted characteristics from given radio signals. Deep learning (DL) and deep neural networks (DNNs) can learn the hidden features of data and build the classifier automatically for decision making, which have been widely used in the SEI research. Considering the insufficiently labeled training samples and large unlabeled training samples, semi-supervised learning-based SEI (SS-SEI) methods have been proposed. However, there are few SS-SEI methods focusing on extracting the discriminative and generalized semantic features of radio signals. In this paper, we propose an SS-SEI method using metric-adversarial training (MAT). Specifically, pseudo labels are innovatively introduced into metric learning to enable semi-supervised metric learning (SSML), and an objective function alternatively regularized by SSML and virtual adversarial training (VAT) is designed to extract discriminative and generalized semantic features of radio signals. The proposed MAT-based SS-SEI method is evaluated on an open-source large-scale real-world automatic-dependent surveillance-broadcast (ADS-B) dataset and WiFi dataset and is compared with state-of-the-art methods. The simulation results show that the proposed method achieves better identification performance than existing state-of-the-art methods. Specifically, when the ratio of the number of labeled training samples to the number of all training samples is 10\%, the identification accuracy is 84.80\% under the ADS-B dataset and 80.70\% under the WiFi dataset. Our code can be downloaded from https://github.com/lovelymimola/MAT-based-SS-SEI.

eess.SP↗

arXiv2022

Hardware faults that matter: Understanding and Estimating the safety impact of hardware faults on object detection DNNs

Object detection neural network models need to perform reliably in highly dynamic and safety-critical environments like automated driving or robotics. Therefore, it is paramount to verify the robustness of the detection under unexpected hardware faults like soft errors that can impact a systems perception module. Standard metrics based on average precision produce model vulnerability estimates at the object level rather than at an image level. As we show in this paper, this does not provide an intuitive or representative indicator of the safety-related impact of silent data corruption caused by bit flips in the underlying memory but can lead to an over- or underestimation of typical fault-induced hazards. With an eye towards safety-related real-time applications, we propose a new metric IVMOD (Image-wise Vulnerability Metric for Object Detection) to quantify vulnerability based on an incorrect image-wise object detection due to false positive (FPs) or false negative (FNs) objects, combined with a severity analysis. The evaluation of several representative object detection models shows that even a single bit flip can lead to a severe silent data corruption event with potentially critical safety implications, with e.g., up to (much greater than) 100 FPs generated, or up to approx. 90% of true positives (TPs) are lost in an image. Furthermore, with a single stuck-at-1 fault, an entire sequence of images can be affected, causing temporally persistent ghost detections that can be mistaken for actual objects (covering up to approx. 83% of the image). Furthermore, actual objects in the scene are continuously missed (up to approx. 64% of TPs are lost). Our work establishes a detailed understanding of the safety-related vulnerability of such critical workloads against hardware faults.

cs.CV↗

arXiv2022

Hierarchy of Symmetry Breaking Correlated Phases in Twisted Bilayer Graphene

Twisted bilayer graphene (TBG) near the magic twist angle of $\sim1.1^{o}$ exhibits a rich phase diagram. However, the interplay between different phases and their dependence on twist angle is still elusive. Here, we explore the stability of various TBG phases and demonstrate that superconductivity near filling of two electrons per moiré unit cell alongside Fermi surface reconstructions, as well as entropy-driven high-temperature phase transitions and linear-in-T resistance occur over a range of twist angles which extends far beyond those exhibiting correlated insulating phases. In the vicinity of the magic angle, we also find a metallic phase that displays a hysteretic anomalous Hall effect and incipient Chern insulating behaviour. Such a metallic phase can be rationalized in terms of the interplay between interaction-driven deformations of TBG bands leading to Berry curvature redistribution and Fermi surface reconstruction. Our results provide an extensive perspective on the hierarchy of correlated phases in TBG as classified by their robustness against deviations from the magic angle or, equivalently, their electronic interaction requirements.

cond-mat.str-el↗

arXiv2022

Topological space-time crystal

We introduce a new class of out-of-equilibrium noninteracting topological phases, the topological space-time crystals. These are time-dependent quantum systems which do not have discrete spatial translation symmetries, but instead are invariant under discrete space-time translations. Similar to the Floquet-Bloch systems, the space-time crystals can be described by a frequency-domain-enlarged Hamiltonian, which is used to classify topologically distinct space-time crystals. We show that these space-time crystals can be engineered from conventional crystals with an additional time-dependent drive that behaves like a traveling wave moving across the crystal. Interestingly, we are able to construct 1D and 2D examples of topological space-time crystals based on tight-binding models which involve only one orbital, in contrast to the two-orbital minimal models for any previous discovered static or Floquet topological phases with crystalline structures.

cond-mat.mes-hall↗

arXiv2022

Federated Reinforcement Learning with Environment Heterogeneity

We study a Federated Reinforcement Learning (FedRL) problem in which $n$ agents collaboratively learn a single policy without sharing the trajectories they collected during agent-environment interaction. We stress the constraint of environment heterogeneity, which means $n$ environments corresponding to these $n$ agents have different state transitions. To obtain a value function or a policy function which optimizes the overall performance in all environments, we propose two federated RL algorithms, \texttt{QAvg} and \texttt{PAvg}. We theoretically prove that these algorithms converge to suboptimal solutions, while such suboptimality depends on how heterogeneous these $n$ environments are. Moreover, we propose a heuristic that achieves personalization by embedding the $n$ environments into $n$ vectors. The personalization heuristic not only improves the training but also allows for better generalization to new environments.

cs.LG↗

arXiv2021

Ascendance of Superconductivity in Magic-Angle Graphene Multilayers

Graphene moire superlattices have emerged as a platform hosting and abundance of correlated insulating, topological, and superconducting phases. While the origins of strong correlations and non-trivial topology are shown to be directly linked to flat moire bands, the nature and mechanism of superconductivity remain enigmatic. In particular, only alternating twisted stacking geometries of bilayer and trilayer graphene are found to exhibit robust superconductivity manifesting as zero resistance and Fraunhofer interference patterns. Here we demonstrate that magic-angle twisted tri-, quadri-, and pentalayers placed on monolayer tungsten diselenide exhibit flavour polarization and superconductivity. We also observe insulating states in the trilayer and quadrilayer arising at finite electric displacement fields, despite the presence of dispersive bands introduced by additional graphene layers. Moreover, the three multilayer geometries allow us to identify universal features in the family of graphene moire structures arising from the intricate relations between superconducting states, symmetry-breaking transitions, and van Hove singularities. Remarkably, as the number of layers increases, superconductivity emerges over a dramatically enhanced filling-factor range. In particular, in twisted pentalayers, superconductivity extends well beyond the filling of four electrons per moire unit cell, demonstrating the non-trivial role of the additional bands. Our results highlight the importance of the interplay between flat and dispersive bands in extending superconducting regions in graphene moire superlattices and open new frontiers for developing graphene-based superconductors.

cond-mat.supr-con↗

arXiv2021

Universal nonadiabatic energy pumping in a quasiperiodically driven extended system

The paradigm of Floquet engineering of topological states of matter can be generalized into the time-quasiperiodic scenario, where a lower dimensional time-dependent system maps into a higher dimensional one by combining the physical dimensions with additional synthetic dimensions generated by multiple incommensurate driving frequencies. Different than most previous works in which gapped topological phases were considered, we propose an experimentally realizable, one dimensional chain driven by two frequencies, which maps into a gapless Weyl semimetal in synthetic dimension. Based on analytical reasoning and numerical simulations, we found the nonadiabatic quantum dynamics of this system exhibit energy pumping behaviors characterized by universal functions. We also numerically found such behaviors are robust against a considerable amount of spatial disorder.

cond-mat.mes-hall↗

arXiv2021

Unsupervised Domain Adaptive Person Re-Identification via Human Learning Imitation

Unsupervised domain adaptive person re-identification has received significant attention due to its high practical value. In past years, by following the clustering and finetuning paradigm, researchers propose to utilize the teacher-student framework in their methods to decrease the domain gap between different person re-identification datasets. Inspired by recent teacher-student framework based methods, which try to mimic the human learning process either by making the student directly copy behavior from the teacher or selecting reliable learning materials, we propose to conduct further exploration to imitate the human learning process from different aspects, \textit{i.e.}, adaptively updating learning materials, selectively imitating teacher behaviors, and analyzing learning materials structures. The explored three components, collaborate together to constitute a new method for unsupervised domain adaptive person re-identification, which is called Human Learning Imitation framework. The experimental results on three benchmark datasets demonstrate the efficacy of our proposed method.

cs.CV↗