arXiv subjects

Yang Zhou

Publications and source records attributed to Yang Zhou.

At least 433 records · Page 24Linked to original sources

arXiv2022

Diversity Matters: Fully Exploiting Depth Clues for Reliable Monocular 3D Object Detection

As an inherently ill-posed problem, depth estimation from single images is the most challenging part of monocular 3D object detection (M3OD). Many existing methods rely on preconceived assumptions to bridge the missing spatial information in monocular images, and predict a sole depth value for every object of interest. However, these assumptions do not always hold in practical applications. To tackle this problem, we propose a depth solving system that fully explores the visual clues from the subtasks in M3OD and generates multiple estimations for the depth of each target. Since the depth estimations rely on different assumptions in essence, they present diverse distributions. Even if some assumptions collapse, the estimations established on the remaining assumptions are still reliable. In addition, we develop a depth selection and combination strategy. This strategy is able to remove abnormal estimations caused by collapsed assumptions, and adaptively combine the remaining estimations into a single one. In this way, our depth solving system becomes more precise and robust. Exploiting the clues from multiple subtasks of M3OD and without introducing any extra information, our method surpasses the current best method by more than 20% relatively on the Moderate level of test split in the KITTI 3D object detection benchmark, while still maintaining real-time efficiency.

cs.CV↗

arXiv2022

3D Segmentation Guided Style-based Generative Adversarial Networks for PET Synthesis

Potential radioactive hazards in full-dose positron emission tomography (PET) imaging remain a concern, whereas the quality of low-dose images is never desirable for clinical use. So it is of great interest to translate low-dose PET images into full-dose. Previous studies based on deep learning methods usually directly extract hierarchical features for reconstruction. We notice that the importance of each feature is different and they should be weighted dissimilarly so that tiny information can be captured by the neural network. Furthermore, the synthesis on some regions of interest is important in some applications. Here we propose a novel segmentation guided style-based generative adversarial network (SGSGAN) for PET synthesis. (1) We put forward a style-based generator employing style modulation, which specifically controls the hierarchical features in the translation process, to generate images with more realistic textures. (2) We adopt a task-driven strategy that couples a segmentation task with a generative adversarial network (GAN) framework to improve the translation performance. Extensive experiments show the superiority of our overall framework in PET synthesis, especially on those regions of interest.

eess.IV↗

arXiv2022

None Class Ranking Loss for Document-Level Relation Extraction

Document-level relation extraction (RE) aims at extracting relations among entities expressed across multiple sentences, which can be viewed as a multi-label classification problem. In a typical document, most entity pairs do not express any pre-defined relation and are labeled as "none" or "no relation". For good document-level RE performance, it is crucial to distinguish such none class instances (entity pairs) from those of pre-defined classes (relations). However, most existing methods only estimate the probability of pre-defined relations independently without considering the probability of "no relation". This ignores the context of entity pairs and the label correlations between the none class and pre-defined classes, leading to sub-optimal predictions. To address this problem, we propose a new multi-label loss that encourages large margins of label confidence scores between each pre-defined class and the none class, which enables captured label correlations and context-dependent thresholding for label prediction. To gain further robustness against positive-negative imbalance and mislabeled data that could appear in real-world RE datasets, we propose a margin regularization and a margin shifting technique. Experimental results demonstrate that our method significantly outperforms existing multi-label losses for document-level RE and works well in other multi-label tasks such as emotion classification when none class instances are available for training.

cs.CL↗

arXiv2022

Preserve Pre-trained Knowledge: Transfer Learning With Self-Distillation For Action Recognition

Video-based action recognition is one of the most popular topics in computer vision. With recent advances of selfsupervised video representation learning approaches, action recognition usually follows a two-stage training framework, i.e., self-supervised pre-training on large-scale unlabeled sets and transfer learning on a downstream labeled set. However, catastrophic forgetting of the pre-trained knowledge becomes the main issue in the downstream transfer learning of action recognition, resulting in a sub-optimal solution. In this paper, to alleviate the above issue, we propose a novel transfer learning approach that combines self-distillation in fine-tuning to preserve knowledge from the pre-trained model learned from the large-scale dataset. Specifically, we fix the encoder from the last epoch as the teacher model to guide the training of the encoder from the current epoch in the transfer learning. With such a simple yet effective learning strategy, we outperform state-of-the-art methods on widely used UCF101 and HMDB51 datasets in action recognition task.

cs.CV↗

arXiv2022

Human-Centered Prior-Guided and Task-Dependent Multi-Task Representation Learning for Action Recognition Pre-Training

Recently, much progress has been made for self-supervised action recognition. Most existing approaches emphasize the contrastive relations among videos, including appearance and motion consistency. However, two main issues remain for existing pre-training methods: 1) the learned representation is neutral and not informative for a specific task; 2) multi-task learning-based pre-training sometimes leads to sub-optimal solutions due to inconsistent domains of different tasks. To address the above issues, we propose a novel action recognition pre-training framework, which exploits human-centered prior knowledge that generates more informative representation, and avoids the conflict between multiple tasks by using task-dependent representations. Specifically, we distill knowledge from a human parsing model to enrich the semantic capability of representation. In addition, we combine knowledge distillation with contrastive learning to constitute a task-dependent multi-task framework. We achieve state-of-the-art performance on two popular benchmarks for action recognition task, i.e., UCF101 and HMDB51, verifying the effectiveness of our method.

cs.CV↗

arXiv2022

FedDUAP: Federated Learning with Dynamic Update and Adaptive Pruning Using Shared Data on the Server

Despite achieving remarkable performance, Federated Learning (FL) suffers from two critical challenges, i.e., limited computational resources and low training efficiency. In this paper, we propose a novel FL framework, i.e., FedDUAP, with two original contributions, to exploit the insensitive data on the server and the decentralized data in edge devices to further improve the training efficiency. First, a dynamic server update algorithm is designed to exploit the insensitive data on the server, in order to dynamically determine the optimal steps of the server update for improving the convergence and accuracy of the global model. Second, a layer-adaptive model pruning method is developed to perform unique pruning operations adapted to the different dimensions and importance of multiple layers, to achieve a good balance between efficiency and effectiveness. By integrating the two original techniques together, our proposed FL model, FedDUAP, significantly outperforms baseline approaches in terms of accuracy (up to 4.8% higher), efficiency (up to 2.8 times faster), and computational cost (up to 61.9% smaller).

cs.DC↗

arXiv2022

Lesion Localization in OCT by Semi-Supervised Object Detection

Over 300 million people worldwide are affected by various retinal diseases. By noninvasive Optical Coherence Tomography (OCT) scans, a number of abnormal structural changes in the retina, namely retinal lesions, can be identified. Automated lesion localization in OCT is thus important for detecting retinal diseases at their early stage. To conquer the lack of manual annotation for deep supervised learning, this paper presents a first study on utilizing semi-supervised object detection (SSOD) for lesion localization in OCT images. To that end, we develop a taxonomy to provide a unified and structured viewpoint of the current SSOD methods, and consequently identify key modules in these methods. To evaluate the influence of these modules in the new task, we build OCT-SS, a new dataset consisting of over 1k expert-labeled OCT B-scan images and over 13k unlabeled B-scans. Extensive experiments on OCT-SS identify Unbiased Teacher (UnT) as the best current SSOD method for lesion localization. Moreover, we improve over this strong baseline, with mAP increased from 49.34 to 50.86.

cs.CV↗

arXiv2022

Maximizing Modular plus Non-monotone Submodular Functions

The research problem in this work is the relaxation of maximizing non-negative submodular plus modular with the entire real number domain as its value range over a family of down-closed sets. We seek a feasible point $\mathbf{x}^*$ in the polytope of the given constraint such that $\mathbf{x}^*\in\arg\max_{\mathbf{x}\in\mathcal{P}\subseteq[0,1]^n}F(\mathbf{x})+L(\mathbf{x})$, where $F$, $L$ denote the extensions of the underlying submodular function $f$ and modular function $\ell$. We provide an approximation algorithm named \textsc{Measured Continuous Greedy with Adaptive Weights}, which yields a guarantee $F(\mathbf{x})+L(\mathbf{x})\geq \left(1/e-\mathcal{O}(ε)\right)\cdot f(OPT)+\left(\frac{β-e}{e(β-1)}-\mathcal{O}(ε)\right)\cdot\ell(OPT)$ under the assumption that the ratio of non-negative part within $\ell(OPT)$ to the absolute value of its negative part is demonstrated by a parameter $β\in[0, \infty]$, where $OPT$ is the optimal integral solution for the discrete problem. It is obvious that the factor of $\ell(OPT)$ is $1$ when $β=0$, which means the negative part is completely dominant at this time; otherwise the factor is closed to $1/e$ whe $β\rightarrow\infty$. Our work first breaks the restriction on the specific value range of the modular function without assuming non-positivity or non-negativity as previous results and quantifies the relative variation of the approximation guarantee for optimal solutions with arbitrary structure. Moreover, we also give an analysis for the inapproximability of the problem we consider. We show a hardness result that there exists no polynomial algorithm whose output $S$ satisfies $f(S)+\ell(S)\geq0.478\cdot f(OPT)+\ell(OPT)$.

cs.DS↗

arXiv2022

An improved approximation algorithm for maximizing a DR-submodular function over a convex set

Maximizing a DR-submodular function subject to a general convex set is an NP-hard problem arising from many applications in combinatorial optimization and machine learning. While it is highly desirable to design efficient approximation algorithms under this general setting where neither the objective function is monotonic nor the feasible set is down-closed, our main contribution is to present a 0.25-approximation Frank-Wolfe type of algorithm with a sub-exponential time-complexity under the value oracle model.

cs.DS↗

arXiv2022

From Distributed Machine Learning to Federated Learning: A Survey

In recent years, data and computing resources are typically distributed in the devices of end users, various regions or organizations. Because of laws or regulations, the distributed data and computing resources cannot be directly shared among different regions or organizations for machine learning tasks. Federated learning emerges as an efficient approach to exploit distributed data and computing resources, so as to collaboratively train machine learning models, while obeying the laws and regulations and ensuring data security and data privacy. In this paper, we provide a comprehensive survey of existing works for federated learning. We propose a functional architecture of federated learning systems and a taxonomy of related techniques. Furthermore, we present the distributed training, data communication, and security of FL systems. Finally, we analyze their limitations and propose future research directions.

cs.DC↗

arXiv2022

Variation between Antiferromagnetism and Ferrimagnetism in NiPS3 by Electron Doping

How to electrically control magnetic properties of a magnetic material is promising towards spintronic applications, where the investigation of carrier doping effects on antiferromagnetic (AFM) materials remains challenging due to their zero net magnetization. In this work, we found electron doping dependent variation of magnetic orders of a two-dimensional (2D) AFM insulator NiPS3, where doping concentration is tuned by intercalating various organic cations into the van der Waals gaps of NiPS3 without introduction of defects and impurity phases. The doped NiPS3 shows an AFM-ferrimagnetic (FIM) transition at doping level of 0.2-0.5 electrons/cell and a FIM-AFM transition at doping level of >= 0.6 electrons/cell. We propose that the found phenomenon is due to competition between Stoner exchange dominated inter-chain ferromagnetic order and super-exchange dominated inter-chain AFM order at different doping level. Our studies provide a viable way to exploit correlation between electronic structures and magnetic properties of 2D magnetic materials for realization of magnetoelectric effect.

cond-mat.mtrl-sci↗

arXiv2022

Multi-Robot Collaborative Perception with Graph Neural Networks

Multi-robot systems such as swarms of aerial robots are naturally suited to offer additional flexibility, resilience, and robustness in several tasks compared to a single robot by enabling cooperation among the agents. To enhance the autonomous robot decision-making process and situational awareness, multi-robot systems have to coordinate their perception capabilities to collect, share, and fuse environment information among the agents in an efficient and meaningful way such to accurately obtain context-appropriate information or gain resilience to sensor noise or failures. In this paper, we propose a general-purpose Graph Neural Network (GNN) with the main goal to increase, in multi-robot perception tasks, single robots' inference perception accuracy as well as resilience to sensor failures and disturbances. We show that the proposed framework can address multi-view visual perception problems such as monocular depth estimation and semantic segmentation. Several experiments both using photo-realistic and real data gathered from multiple aerial robots' viewpoints show the effectiveness of the proposed approach in challenging inference conditions including images corrupted by heavy noise and camera occlusions or failures.

cs.RO↗

arXiv2022

Connected and Automated Vehicle Distributed Control for On-ramp Merging Scenario: A Virtual Rotation Approach

In this study, we propose a rotation-based connected automated vehicle (CAV) distributed cooperative control strategy for an on-ramp merging scenario. By assuming the mainline and ramp line are straight, we firstly design a virtual rotation approach that transfers the merging problem to a virtual car following (CF) problem to reduce the complexity and dimension of the cooperative CAVs merging control. Based on this concept, a multiple-predecessor virtual CF model and a unidirectional multi-leader communication topology are developed to determine the longitudinal behavior of each CAV. Specifically, we exploit a distributed feedback and feedforward longitudinal controller in preparation for actively generating gaps for merging CAVs, reducing the voids caused by merging, and ensuring safety and traffic efficiency during the process. To ensure the disturbance attenuation property of this system, practical string stability is mathematically proved for the virtual CF controllers to prohibit the traffic oscillation amplification through the traffic stream. Moreover, as a provision for extending the virtual CF application scenarios of any curvy ramp geometry, we utilize a curvilinear coordinate to model the two-dimensional merging control, and further design a local lateral controller based on an extended linear-quadratic regulator to regulate the position deviation and angular deviation of the lane centerlines. For the purpose of systematically evaluating the control performance of the proposed methods, numerical simulation experiments are conducted. As the results indicate, the proposed controllers can actively reduce the void and meanwhile guarantee the damping of traffic oscillations in the merging control area.

eess.SY↗

arXiv2021

Partial Reduction and Cosmology at Defect Brane

Partial reduction is a Randall-Sundrum reduction for only part of the AdS region between finite tension brane and zero tension brane. This is interesting in AdS/BCFT where the AdS bulk contains a defect brane. We employ partial reduction for a AdS bulk with a brane evolving as a $2d$ Friedmann-Robertson-Walker (FRW) cosmology and demonstrate the equivalence between defect extremal surface and island formula for a large subregion fine grained entropy in boundary CFT. We then move to higher dimensions and demonstrate the existence of $4d$ massless graviton on AdS$_4$ brane in partial reduction. We also propose a partial reduction for a $4d$ FRW cosmology at defect brane and obtain the Newton constant by computing boundary entropy.

hep-th↗

arXiv2021

Efficient Device Scheduling with Multi-Job Federated Learning

Recent years have witnessed a large amount of decentralized data in multiple (edge) devices of end-users, while the aggregation of the decentralized data remains difficult for machine learning jobs due to laws or regulations. Federated Learning (FL) emerges as an effective approach to handling decentralized data without sharing the sensitive raw data, while collaboratively training global machine learning models. The servers in FL need to select (and schedule) devices during the training process. However, the scheduling of devices for multiple jobs with FL remains a critical and open problem. In this paper, we propose a novel multi-job FL framework to enable the parallel training process of multiple jobs. The framework consists of a system model and two scheduling methods. In the system model, we propose a parallel training process of multiple jobs, and construct a cost model based on the training time and the data fairness of various devices during the training process of diverse jobs. We propose a reinforcement learning-based method and a Bayesian optimization-based method to schedule devices for multiple jobs while minimizing the cost. We conduct extensive experimentation with multiple jobs and datasets. The experimental results show that our proposed approaches significantly outperform baseline approaches in terms of training time (up to 8.67 times faster) and accuracy (up to 44.6% higher).

cs.DC↗

arXiv2021

Multi-Modal Multi-Instance Learning for Retinal Disease Recognition

This paper attacks an emerging challenge of multi-modal retinal disease recognition. Given a multi-modal case consisting of a color fundus photo (CFP) and an array of OCT B-scan images acquired during an eye examination, we aim to build a deep neural network that recognizes multiple vision-threatening diseases for the given case. As the diagnostic efficacy of CFP and OCT is disease-dependent, the network's ability of being both selective and interpretable is important. Moreover, as both data acquisition and manual labeling are extremely expensive in the medical domain, the network has to be relatively lightweight for learning from a limited set of labeled multi-modal samples. Prior art on retinal disease recognition focuses either on a single disease or on a single modality, leaving multi-modal fusion largely underexplored. We propose in this paper Multi-Modal Multi-Instance Learning (MM-MIL) for selectively fusing CFP and OCT modalities. Its lightweight architecture (as compared to current multi-head attention modules) makes it suited for learning from relatively small-sized datasets. For an effective use of MM-MIL, we propose to generate a pseudo sequence of CFPs by over sampling a given CFP. The benefits of this tactic include well balancing instances across modalities, increasing the resolution of the CFP input, and finding out regions of the CFP most relevant with respect to the final diagnosis. Extensive experiments on a real-world dataset consisting of 1,206 multi-modal cases from 1,193 eyes of 836 subjects demonstrate the viability of the proposed model.

cs.CV↗

arXiv2021

Defect Extremal Surface for Reflected Entropy

Defect extremal surface is defined by extremizing the Ryu-Takayanagi formula corrected by the quantum defect theory. This is interesting when the AdS bulk contains a defect brane (or string). We introduce a defect extremal surface formula for reflected entropy, which is a mixed state generalization of entanglement entropy measure. Based on a decomposition procedure of an AdS bulk with a brane, we demonstrate the equivalence between defect extremal surface formula and island formula for reflected entropy in AdS$_3$/BCFT$_2$. We also compute the evolution of reflected entropy in evaporating black hole model and find that defect extremal surface formula agrees with island formula.

hep-th↗

arXiv2021

Virtual synchronous generator of PV generation without energy storage for frequency support in autonomous microgrid

In autonomous microgrids frequency regulation (FR) is a critical issue, especially with a high level of penetration of the photovoltaic (PV) generation. In this study, a novel virtual synchronous generator (VSG) control for PV generation was introduced to provide frequency support without energy storage. PV generation reserve a part of the active power in accordance with the pre-defined power versus voltage curve. Based on the similarities of the synchronous generator power-angle characteristic curve and the PV array characteristic curve, PV voltage Vpv can be analogized to the power angle δ. An emulated governor (droop control) and the swing equation control is designed and applied to the DC-DC converter. PV voltage deviation is subsequently generated and the pre-defined power versus voltage curve is modified to provide the primary frequency and inertia support. A simulation model of an autonomous microgrid with PV, storage, and diesel generator was built. The feasibility and effectiveness of the proposed VSG strategy are examined under different operating conditions.

eess.SY↗