arXiv subjects

Chen Chen

Publications and source records attributed to Chen Chen.

At least 865 records · Page 48Linked to original sources

arXiv2021

Cardiac Segmentation on Late Gadolinium Enhancement MRI: A Benchmark Study from Multi-Sequence Cardiac MR Segmentation Challenge

Accurate computing, analysis and modeling of the ventricles and myocardium from medical images are important, especially in the diagnosis and treatment management for patients suffering from myocardial infarction (MI). Late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR) provides an important protocol to visualize MI. However, automated segmentation of LGE CMR is still challenging, due to the indistinguishable boundaries, heterogeneous intensity distribution and complex enhancement patterns of pathological myocardium from LGE CMR. Furthermore, compared with the other sequences LGE CMR images with gold standard labels are particularly limited, which represents another obstacle for developing novel algorithms for automatic segmentation of LGE CMR. This paper presents the selective results from the Multi-Sequence Cardiac MR (MS-CMR) Segmentation challenge, in conjunction with MICCAI 2019. The challenge offered a data set of paired MS-CMR images, including auxiliary CMR sequences as well as LGE CMR, from 45 patients who underwent cardiomyopathy. It was aimed to develop new algorithms, as well as benchmark existing ones for LGE CMR segmentation and compare them objectively. In addition, the paired MS-CMR images could enable algorithms to combine the complementary information from the other sequences for the segmentation of LGE CMR. Nine representative works were selected for evaluation and comparisons, among which three methods are unsupervised methods and the other six are supervised. The results showed that the average performance of the nine methods was comparable to the inter-observer variations. The success of these methods was mainly attributed to the inclusion of the auxiliary sequences from the MS-CMR images, which provide important label information for the training of deep neural networks.

eess.IV↗

arXiv2021

On the Transformer Growth for Progressive BERT Training

Due to the excessive cost of large-scale language model pre-training, considerable efforts have been made to train BERT progressively -- start from an inferior but low-cost model and gradually grow the model to increase the computational complexity. Our objective is to advance the understanding of Transformer growth and discover principles that guide progressive training. First, we find that similar to network architecture search, Transformer growth also favors compound scaling. Specifically, while existing methods only conduct network growth in a single dimension, we observe that it is beneficial to use compound growth operators and balance multiple dimensions (e.g., depth, width, and input length of the model). Moreover, we explore alternative growth operators in each dimension via controlled comparison to give operator selection practical guidance. In light of our analyses, the proposed method speeds up BERT pre-training by 73.6% and 82.2% for the base and large models respectively, while achieving comparable performances

cs.CL↗

arXiv2021

Joint Motion Correction and Super Resolution for Cardiac Segmentation via Latent Optimisation

In cardiac magnetic resonance (CMR) imaging, a 3D high-resolution segmentation of the heart is essential for detailed description of its anatomical structures. However, due to the limit of acquisition duration and respiratory/cardiac motion, stacks of multi-slice 2D images are acquired in clinical routine. The segmentation of these images provides a low-resolution representation of cardiac anatomy, which may contain artefacts caused by motion. Here we propose a novel latent optimisation framework that jointly performs motion correction and super resolution for cardiac image segmentations. Given a low-resolution segmentation as input, the framework accounts for inter-slice motion in cardiac MR imaging and super-resolves the input into a high-resolution segmentation consistent with input. A multi-view loss is incorporated to leverage information from both short-axis view and long-axis view of cardiac imaging. To solve the inverse problem, iterative optimisation is performed in a latent space, which ensures the anatomical plausibility. This alleviates the need of paired low-resolution and high-resolution images for supervised learning. Experiments on two cardiac MR datasets show that the proposed framework achieves high performance, comparable to state-of-the-art super-resolution approaches and with better cross-domain generalisability and anatomical plausibility.

eess.IV↗

arXiv2021

Cooperative Training and Latent Space Data Augmentation for Robust Medical Image Segmentation

Deep learning-based segmentation methods are vulnerable to unforeseen data distribution shifts during deployment, e.g. change of image appearances or contrasts caused by different scanners, unexpected imaging artifacts etc. In this paper, we present a cooperative framework for training image segmentation models and a latent space augmentation method for generating hard examples. Both contributions improve model generalization and robustness with limited data. The cooperative training framework consists of a fast-thinking network (FTN) and a slow-thinking network (STN). The FTN learns decoupled image features and shape features for image reconstruction and segmentation tasks. The STN learns shape priors for segmentation correction and refinement. The two networks are trained in a cooperative manner. The latent space augmentation generates challenging examples for training by masking the decoupled latent space in both channel-wise and spatial-wise manners. We performed extensive experiments on public cardiac imaging datasets. Using only 10 subjects from a single site for training, we demonstrated improved cross-site segmentation performance and increased robustness against various unforeseen imaging artifacts compared to strong baseline methods. Particularly, cooperative training with latent space data augmentation yields 15% improvement in terms of average Dice score when compared to a standard training method.

cs.CV↗

arXiv2021

Efficient Top-k Vulnerable Nodes Detection in Uncertain Graphs

Uncertain graphs have been widely used to model complex linked data in many real-world applications, such as guaranteed-loan networks and power grids, where a node or edge may be associated with a probability. In these networks, a node usually has a certain chance of default or breakdown due to self-factors or the influence from upstream nodes. For regulatory authorities and companies, it is critical to efficiently identify the vulnerable nodes, i.e., nodes with high default risks, such that they could pay more attention to these nodes for the purpose of risk management. In this paper, we propose and investigate the problem of top-$k$ vulnerable nodes detection in uncertain graphs. We formally define the problem and prove its hardness. To identify the $k$ most vulnerable nodes, a sampling-based approach is proposed. Rigorous theoretical analysis is conducted to bound the quality of returned results. Novel optimization techniques and a bottom-$k$ sketch based approach are further developed in order to scale for large networks. In the experiments, we demonstrate the performance of proposed techniques on 3 real financial networks and 5 benchmark networks. The evaluation results show that the proposed methods can achieve up to 2 orders of magnitudes speedup compared with the baseline approach. Moreover, to further verify the advantages of our model in real-life scenarios, we integrate the proposed techniques with our current loan risk control system, which is deployed in the collaborated bank, for more evaluation. Particularly, we show that our proposed new model has superior performance on real-life guaranteed-loan network data, which can better predict the default risks of enterprises compared to the state-of-the-art techniques.

cs.CE↗

arXiv2021

TransBTS: Multimodal Brain Tumor Segmentation Using Transformer

Transformer, which can benefit from global (long-range) information modeling using self-attention mechanisms, has been successful in natural language processing and 2D image classification recently. However, both local and global features are crucial for dense prediction tasks, especially for 3D medical image segmentation. In this paper, we for the first time exploit Transformer in 3D CNN for MRI Brain Tumor Segmentation and propose a novel network named TransBTS based on the encoder-decoder structure. To capture the local 3D context information, the encoder first utilizes 3D CNN to extract the volumetric spatial feature maps. Meanwhile, the feature maps are reformed elaborately for tokens that are fed into Transformer for global feature modeling. The decoder leverages the features embedded by Transformer and performs progressive upsampling to predict the detailed segmentation map. Extensive experimental results on both BraTS 2019 and 2020 datasets show that TransBTS achieves comparable or higher results than previous state-of-the-art 3D methods for brain tumor segmentation on 3D MRI scans. The source code is available at https://github.com/Wenxuan-1119/TransBTS

cs.CV↗

arXiv2021

Deep Learning-Aided OFDM-Based Generalized Optical Quadrature Spatial Modulation

In this paper, we propose an orthogonal frequency division multiplexing (OFDM)-based generalized optical quadrature spatial modulation (GOQSM) technique for multiple-input multiple-output optical wireless communication (MIMO-OWC) systems. Considering the error propagation and noise amplification effects when applying maximum likelihood and maximum ratio combining (ML-MRC)-based detection, we further propose a deep neural network (DNN)-aided detection for OFDM-based GOQSM systems. The proposed DNN-aided detection scheme performs the GOQSM detection in a joint manner, which can efficiently eliminate the adverse effects of both error propagation and noise amplification. The obtained simulation results successfully verify the superiority of the deep learning-aided OFDM-based GOQSM technique for high-speed MIMO-OWC systems.

eess.SP↗

arXiv2021

Optimal Power Flow for DC Networks with Robust Feasibility and Stability Guarantees

With high penetrations of renewable generation and variable loads, there is significant uncertainty associated with power flows in DC networks such that stability and operational constraint satisfaction are of concern. Most existing DC network optimal power flow (DN-OPF) formulations assume exact knowledge of loading conditions and do not provide stability guarantees. In contrast, this paper studies a DN-OPF formulation which considers both stability and operational constraint satisfaction under uncertainty. The need to account for a range of uncertainty realizations in this paper's robust optimization formulation results in a challenging semi-infinite program (SIP). The proposed solution algorithm reformulates this SIP into a computationally tractable problem by constructing a tight convex inner approximation of the stability set using sufficient conditions for the existence of a feasible and stable power flow solution. Optimal generator setpoints are obtained by optimizing over the proposed convex stability set. The validity and value of the proposed algorithm are demonstrated through various DC networks adapted from IEEE test cases.

math.OC↗

arXiv2021

Energy Aligning for Biased Models

Training on class-imbalanced data usually results in biased models that tend to predict samples into the majority classes, which is a common and notorious problem. From the perspective of energy-based model, we demonstrate that the free energies of categories are aligned with the label distribution theoretically, thus the energies of different classes are expected to be close to each other when aiming for ``balanced'' performance. However, we discover a severe energy-bias phenomenon in the models trained on class-imbalanced dataset. To eliminate the bias, we propose a simple and effective method named Energy Aligning by merely adding the calculated shift scalars onto the output logits during inference, which does not require to (i) modify the network architectures, (ii) intervene the standard learning paradigm, (iii) perform two-stage training. The proposed algorithm is evaluated on two class imbalance-related tasks under various settings: class incremental learning and long-tailed recognition. Experimental results show that energy aligning can effectively alleviate class imbalance issue and outperform state-of-the-art methods on several benchmarks.

cs.LG↗

arXiv2021

Fairness-Aware Unsupervised Feature Selection

Feature selection is a prevalent data preprocessing paradigm for various learning tasks. Due to the expensive cost of acquiring supervision information, unsupervised feature selection sparks great interests recently. However, existing unsupervised feature selection algorithms do not have fairness considerations and suffer from a high risk of amplifying discrimination by selecting features that are over associated with protected attributes such as gender, race, and ethnicity. In this paper, we make an initial investigation of the fairness-aware unsupervised feature selection problem and develop a principled framework, which leverages kernel alignment to find a subset of high-quality features that can best preserve the information in the original feature space while being minimally correlated with protected attributes. Specifically, different from the mainstream in-processing debiasing methods, our proposed framework can be regarded as a model-agnostic debiasing strategy that eliminates biases and discrimination before downstream learning algorithms are involved. Experimental results on multiple real-world datasets demonstrate that our framework achieves a good trade-off between utility maximization and fairness promotion.

cs.LG↗

arXiv2021

SGPT: Semantic Graphs based Pre-training for Aspect-based Sentiment Analysis

Previous studies show effective of pre-trained language models for sentiment analysis. However, most of these studies ignore the importance of sentimental information for pre-trained models.Therefore, we fully investigate the sentimental information for pre-trained models and enhance pre-trained language models with semantic graphs for sentiment analysis.In particular, we introduce Semantic Graphs based Pre-training(SGPT) using semantic graphs to obtain synonym knowledge for aspect-sentiment pairs and similar aspect/sentiment terms.We then optimize the pre-trained language model with the semantic graphs.Empirical studies on several downstream tasks show that proposed model outperforms strong pre-trained baselines. The results also show the effectiveness of proposed semantic graphs for pre-trained model.

cs.CL↗

arXiv2021

BDANet: Multiscale Convolutional Neural Network with Cross-directional Attention for Building Damage Assessment from Satellite Images

Fast and effective responses are required when a natural disaster (e.g., earthquake, hurricane, etc.) strikes. Building damage assessment from satellite imagery is critical before relief effort is deployed. With a pair of pre- and post-disaster satellite images, building damage assessment aims at predicting the extent of damage to buildings. With the powerful ability of feature representation, deep neural networks have been successfully applied to building damage assessment. Most existing works simply concatenate pre- and post-disaster images as input of a deep neural network without considering their correlations. In this paper, we propose a novel two-stage convolutional neural network for Building Damage Assessment, called BDANet. In the first stage, a U-Net is used to extract the locations of buildings. Then the network weights from the first stage are shared in the second stage for building damage assessment. In the second stage, a two-branch multi-scale U-Net is employed as backbone, where pre- and post-disaster images are fed into the network separately. A cross-directional attention module is proposed to explore the correlations between pre- and post-disaster images. Moreover, CutMix data augmentation is exploited to tackle the challenge of difficult classes. The proposed method achieves state-of-the-art performance on a large-scale dataset -- xBD. The code is available at https://github.com/ShaneShen/BDANet-Building-Damage-Assessment.

cs.CV↗

arXiv2021

Learning Normal Dynamics in Videos with Meta Prototype Network

Frame reconstruction (current or future frame) based on Auto-Encoder (AE) is a popular method for video anomaly detection. With models trained on the normal data, the reconstruction errors of anomalous scenes are usually much larger than those of normal ones. Previous methods introduced the memory bank into AE, for encoding diverse normal patterns across the training videos. However, they are memory-consuming and cannot cope with unseen new scenarios in the testing data. In this work, we propose a dynamic prototype unit (DPU) to encode the normal dynamics as prototypes in real time, free from extra memory cost. In addition, we introduce meta-learning to our DPU to form a novel few-shot normalcy learner, namely Meta-Prototype Unit (MPU). It enables the fast adaption capability on new scenes by only consuming a few iterations of update. Extensive experiments are conducted on various benchmarks. The superior performance over the state-of-the-art demonstrates the effectiveness of our method.

cs.CV↗

arXiv2021

Balanced Knowledge Distillation for Long-tailed Learning

Deep models trained on long-tailed datasets exhibit unsatisfactory performance on tail classes. Existing methods usually modify the classification loss to increase the learning focus on tail classes, which unexpectedly sacrifice the performance on head classes. In fact, this scheme leads to a contradiction between the two goals of long-tailed learning, i.e., learning generalizable representations and facilitating learning for tail classes. In this work, we explore knowledge distillation in long-tailed scenarios and propose a novel distillation framework, named Balanced Knowledge Distillation (BKD), to disentangle the contradiction between the two goals and achieve both simultaneously. Specifically, given a vanilla teacher model, we train the student model by minimizing the combination of an instance-balanced classification loss and a class-balanced distillation loss. The former benefits from the sample diversity and learns generalizable representation, while the latter considers the class priors and facilitates learning mainly for tail classes. The student model trained with BKD obtains significant performance gain even compared with its teacher model. We conduct extensive experiments on several long-tailed benchmark datasets and demonstrate that the proposed BKD is an effective knowledge distillation framework in long-tailed scenarios, as well as a new state-of-the-art method for long-tailed learning. Code is available at https://github.com/EricZsy/BalancedKnowledgeDistillation .

cs.CV↗

arXiv2021

VIGOR: Cross-View Image Geo-localization beyond One-to-one Retrieval

Cross-view image geo-localization aims to determine the locations of street-view query images by matching with GPS-tagged reference images from aerial view. Recent works have achieved surprisingly high retrieval accuracy on city-scale datasets. However, these results rely on the assumption that there exists a reference image exactly centered at the location of any query image, which is not applicable for practical scenarios. In this paper, we redefine this problem with a more realistic assumption that the query image can be arbitrary in the area of interest and the reference images are captured before the queries emerge. This assumption breaks the one-to-one retrieval setting of existing datasets as the queries and reference images are not perfectly aligned pairs, and there may be multiple reference images covering one query location. To bridge the gap between this realistic setting and existing datasets, we propose a new large-scale benchmark -- VIGOR -- for cross-View Image Geo-localization beyond One-to-one Retrieval. We benchmark existing state-of-the-art methods and propose a novel end-to-end framework to localize the query in a coarse-to-fine manner. Apart from the image-level retrieval accuracy, we also evaluate the localization accuracy in terms of the actual distance (meters) using the raw GPS data. Extensive experiments are conducted under different application scenarios to validate the effectiveness of the proposed method. The results indicate that cross-view geo-localization in this realistic setting is still challenging, fostering new research in this direction. Our dataset and code will be released at \url{https://github.com/Jeff-Zilence/VIGOR}

cs.CV↗

arXiv2021

Extreme High-velocity Outflows from High-redshift BOSS Quasars

It is common to assume that all narrow absorption lines (NALs) at extreme high-velocity shifts form in cosmologically intervening gas or galaxies unrelated to quasars. However, previous detailed studies of individual quasars have shown that some NALs at these large velocity shifts do form in high-speed quasar ejecta. We search for extreme high-velocity NAL outflows (with speeds $\sim$0.1-0.2c) based on relationships with associated absorption lines (AALs) and broad absorption-line (BAL) outflows. We find that high-velocity NALs are strongly correlated with AALs, BALs, and radio loudness, indicating that a significant fraction of high-velocity systems are either ejected from the quasars or form in material swept up by the radio jets (and are not unrelated intervening gas). We also consider line-locked C IV doublets as another indicator of high-velocity NALs formed in outflows. The fact that line-locked NALs are highly ionized and correlated with BAL outflows and radio-loud quasars implies that physical line locking due to radiative forces is both common and real, which provides indirect evidence that a significant fraction of high-velocity NALs are intrinsic to quasars.

astro-ph.GA↗

arXiv2021

Good Solid-State Electrolytes Have Low, Glass-like Thermal Conductivity

Management of heat during charging and discharging of Li-ion batteries is critical for their safety, reliability, and performance. Understanding the thermal conductivity of the materials comprising batteries is crucial for controlling the temperature and temperature distribution in batteries. This work provides systemic quantitative measurements of the thermal conductivity of three important classes of solid electrolytes (oxides, sulfides, and halides) over the temperature range 150-350 K. Studies include the oxides Li1.5Al0.5Ge1.5(PO4)3 and Li6.4La3Zr1.4Ta0.6O12, sulfides Li2S-P2S5, Li6PS5Cl, and Na3PS4, and halides Li3InCl6 and Li3YCl6. Thermal conductivities of sulfide and halide solid electrolytes are in the range 0.45-0.70 W m-1 K-1; thermal conductivities of Li6.4La3Zr1.4Ta0.6O12 and Li1.5Al0.5Ge1.5(PO4)3 are 1.4 W m-1 K-1 and 2.2 W m-1 K-1, respectively. For most of the solid electrolytes studied in this work, the thermal conductivity increases with increasing temperature; i.e., the thermal conductivity has a glass-like temperature dependence. The measured room-temperature thermal conductivities agree well with the calculated minimum thermal conductivities indicating the phonon mean-free-paths in these solid electrolytes are close to an atomic spacing. We attribute the low, glass-like thermal conductivity of the solid electrolytes investigated to the combination of their complex crystal structures and the atomic-scale disorder induced by the materials processing methods that are typically needed to produce high ionic conductivities.

cond-mat.mtrl-sci↗

arXiv2021

Optimal Scheduling of Integrated Demand Response-Enabled Integrated Energy Systems with Uncertain Renewable Generations: A Stackelberg Game Approach

In order to balance the interests of integrated energy operator (IEO) and users, a novel Stackelberg game-based optimization framework is proposed for the optimal scheduling of integrated demand response (IDR)-enabled integrated energy systems with uncertain renewable generations, where the IEO acts as the leader who pursues the maximization of his profits by setting energy prices, while the users are the follower who adjusts energy consumption plans to minimize their energy costs. Taking into account the inherent uncertainty of renewable generations, the probabilistic spinning reserve is written in the form of a chance constraint; in addition, a district heating network model is built considering the characteristics of time delay and thermal attenuation by fully exploiting its potential, and the flexible thermal comfort requirements of users in IDR are considered by introducing a predicted mean vote (PMV) index. To solve the raised model, sequence operation theory is introduced to convert the chance constraint into its deterministic equivalent form, and thereby, the leader-follower Stackelberg game is tackled into a mixed-integer quadratic programming formulation through Karush-Kuhn-Tucker optimality conditions and is finally solved by the CPLEX optimizer. The results of two case studies demonstrate that the proposed Stackelberg game-based approach manages to achieve the Stackelberg equilibrium between IEO and users by the coordination of renewable generations and IDR. Furthermore, the study on a real integrated energy system in China verifies the applicability of the proposed approach for real-world applications.

eess.SP↗