arXiv subjects

Kang Li

Publications and source records attributed to Kang Li.

At least 163 records · Page 9Linked to original sources

arXiv2021

Asymptotic expansion in measure and strong ergodicity

In this paper, we introduce and study a notion of asymptotic expansion in measure for measurable actions. This generalises expansion in measure and provides a new perspective on the classical notion of strong ergodicity. Moreover, we obtain structure theorems for asymptotically expanding actions, showing that they admit exhaustions by domains of expansion. As an application, we recover a recent result of Marrakchi, characterising strong ergodicity in terms of local spectral gaps. We also show that homogeneous strongly ergodic actions are always expanding in measure and establish a connection between asymptotic expansion in measure and asymptotic expanders by means of approximating spaces.

math.DS↗

arXiv2021

Dual-Teacher++: Exploiting Intra-domain and Inter-domain Knowledge with Reliable Transfer for Cardiac Segmentation

Annotation scarcity is a long-standing problem in medical image analysis area. To efficiently leverage limited annotations, abundant unlabeled data are additionally exploited in semi-supervised learning, while well-established cross-modality data are investigated in domain adaptation. In this paper, we aim to explore the feasibility of concurrently leveraging both unlabeled data and cross-modality data for annotation-efficient cardiac segmentation. To this end, we propose a cutting-edge semi-supervised domain adaptation framework, namely Dual-Teacher++. Besides directly learning from limited labeled target domain data (e.g., CT) via a student model adopted by previous literature, we design novel dual teacher models, including an inter-domain teacher model to explore cross-modality priors from source domain (e.g., MR) and an intra-domain teacher model to investigate the knowledge beneath unlabeled target domain. In this way, the dual teacher models would transfer acquired inter- and intra-domain knowledge to the student model for further integration and exploitation. Moreover, to encourage reliable dual-domain knowledge transfer, we enhance the inter-domain knowledge transfer on the samples with higher similarity to target domain after appearance alignment, and also strengthen intra-domain knowledge transfer of unlabeled target data with higher prediction confidence. In this way, the student model can obtain reliable dual-domain knowledge and yield improved performance on target domain data. We extensively evaluated the feasibility of our method on the MM-WHS 2017 challenge dataset. The experiments have demonstrated the superiority of our framework over other semi-supervised learning and domain adaptation methods. Moreover, our performance gains could be yielded in bidirections,i.e., adapting from MR to CT, and from CT to MR.

eess.IV↗

arXiv2020

DoFE: Domain-oriented Feature Embedding for Generalizable Fundus Image Segmentation on Unseen Datasets

Deep convolutional neural networks have significantly boosted the performance of fundus image segmentation when test datasets have the same distribution as the training datasets. However, in clinical practice, medical images often exhibit variations in appearance for various reasons, e.g., different scanner vendors and image quality. These distribution discrepancies could lead the deep networks to over-fit on the training datasets and lack generalization ability on the unseen test datasets. To alleviate this issue, we present a novel Domain-oriented Feature Embedding (DoFE) framework to improve the generalization ability of CNNs on unseen target domains by exploring the knowledge from multiple source domains. Our DoFE framework dynamically enriches the image features with additional domain prior knowledge learned from multi-source domains to make the semantic features more discriminative. Specifically, we introduce a Domain Knowledge Pool to learn and memorize the prior information extracted from multi-source domains. Then the original image features are augmented with domain-oriented aggregated features, which are induced from the knowledge pool based on the similarity between the input image and multi-source domain images. We further design a novel domain code prediction branch to infer this similarity and employ an attention-guided mechanism to dynamically combine the aggregated features with the semantic features. We comprehensively evaluate our DoFE framework on two fundus image segmentation tasks, including the optic cup and disc segmentation and vessel segmentation. Our DoFE framework generates satisfying segmentation results on unseen datasets and surpasses other domain generalization and network regularization methods.

cs.CV↗

arXiv2020

CAT-Gen: Improving Robustness in NLP Models via Controlled Adversarial Text Generation

NLP models are shown to suffer from robustness issues, i.e., a model's prediction can be easily changed under small perturbations to the input. In this work, we present a Controlled Adversarial Text Generation (CAT-Gen) model that, given an input text, generates adversarial texts through controllable attributes that are known to be invariant to task labels. For example, in order to attack a model for sentiment classification over product reviews, we can use the product categories as the controllable attribute which would not change the sentiment of the reviews. Experiments on real-world NLP datasets demonstrate that our method can generate more diverse and fluent adversarial texts, compared to many existing adversarial text generation approaches. We further use our generated adversarial examples to improve models through adversarial training, and we demonstrate that our generated attacks are more robust against model re-training and different model architectures.

cs.CL↗

arXiv2020

Towards Cross-modality Medical Image Segmentation with Online Mutual Knowledge Distillation

The success of deep convolutional neural networks is partially attributed to the massive amount of annotated training data. However, in practice, medical data annotations are usually expensive and time-consuming to be obtained. Considering multi-modality data with the same anatomic structures are widely available in clinic routine, in this paper, we aim to exploit the prior knowledge (e.g., shape priors) learned from one modality (aka., assistant modality) to improve the segmentation performance on another modality (aka., target modality) to make up annotation scarcity. To alleviate the learning difficulties caused by modality-specific appearance discrepancy, we first present an Image Alignment Module (IAM) to narrow the appearance gap between assistant and target modality data.We then propose a novel Mutual Knowledge Distillation (MKD) scheme to thoroughly exploit the modality-shared knowledge to facilitate the target-modality segmentation. To be specific, we formulate our framework as an integration of two individual segmentors. Each segmentor not only explicitly extracts one modality knowledge from corresponding annotations, but also implicitly explores another modality knowledge from its counterpart in mutual-guided manner. The ensemble of two segmentors would further integrate the knowledge from both modalities and generate reliable segmentation results on target modality. Experimental results on the public multi-class cardiac segmentation data, i.e., MMWHS 2017, show that our method achieves large improvements on CT segmentation by utilizing additional MRI data and outperforms other state-of-the-art multi-modality learning methods.

eess.IV↗

arXiv2020

Integrated and Spectrally Selective Thermal Emitters Enabled by Layered Metamaterials

Nanophotonic engineering of light-matter interaction at subwavelength scale allows thermal radiation that is fundamentally different from that of traditional thermal emitters and provides exciting opportunities for various thermal-photonic applications. We propose a new kind of integrated and electrically controlled thermal emitter that exploits layered metamaterials with lithography-free and dielectric/metallic nanolayers. We demonstrate both theoretically and experimentally that the proposed concept can create a strong photonic bandgap in the visible regime and allow small impedance mismatch at the infrared wavelengths, which gives rise to optical features of significantly enhanced emissivity at the broad infrared wavelengths of 1.4-14 um as well as effectively suppressed emissivity in the visible region. The electrically driven metamaterial devices are optically and thermally stable at temperature up to ~800 K with electro-optical conversion efficiency reaching ~30%. We believe that the proposed high efficiency thermal emitters will pave the way towards integrated infrared light source platforms for various thermal-photonic applications and particularly provide a novel alternative for cost-effective, compact, low glare, and energy-efficient infrared heating.

physics.optics↗

arXiv2020

Coarse Baum-Connes conjecture and rigidity for Roe algebras

In this paper, we connect the rigidity problem and the coarse Baum-Connes conjecture for Roe algebras. In particular, we show that if $X$ and $Y$ are two uniformly locally finite metric spaces such that their Roe algebras are $*$-isomorphic, then $X$ and $Y$ are coarsely equivalent provided either $X$ or $Y$ satisfies the coarse Baum-Connes conjecture with coefficients. It is well-known that coarse embeddability into a Hilbert space implies the coarse Baum-Connes conjecture with coefficients. On the other hand, we provide a new example of a finitely generated group satisfying the coarse Baum-Connes conjecture with coefficients but which does not coarsely embed into a Hilbert space.

math.OA↗

arXiv2020

Dual-Teacher: Integrating Intra-domain and Inter-domain Teachers for Annotation-efficient Cardiac Segmentation

Medical image annotations are prohibitively time-consuming and expensive to obtain. To alleviate annotation scarcity, many approaches have been developed to efficiently utilize extra information, e.g.,semi-supervised learning further exploring plentiful unlabeled data, domain adaptation including multi-modality learning and unsupervised domain adaptation resorting to the prior knowledge from additional modality. In this paper, we aim to investigate the feasibility of simultaneously leveraging abundant unlabeled data and well-established cross-modality data for annotation-efficient medical image segmentation. To this end, we propose a novel semi-supervised domain adaptation approach, namely Dual-Teacher, where the student model not only learns from labeled target data (e.g., CT), but also explores unlabeled target data and labeled source data (e.g., MR) by two teacher models. Specifically, the student model learns the knowledge of unlabeled target data from intra-domain teacher by encouraging prediction consistency, as well as the shape priors embedded in labeled source data from inter-domain teacher via knowledge distillation. Consequently, the student model can effectively exploit the information from all three data resources and comprehensively integrate them to achieve improved performance. We conduct extensive experiments on MM-WHS 2017 dataset and demonstrate that our approach is able to concurrently utilize unlabeled data and cross-modality data with superior performance, outperforming semi-supervised learning and domain adaptation methods with a large margin.

cs.CV↗

arXiv2020

Weakly Supervised Deep Nuclei Segmentation Using Partial Points Annotation in Histopathology Images

Nuclei segmentation is a fundamental task in histopathology image analysis. Typically, such segmentation tasks require significant effort to manually generate accurate pixel-wise annotations for fully supervised training. To alleviate such tedious and manual effort, in this paper we propose a novel weakly supervised segmentation framework based on partial points annotation, i.e., only a small portion of nuclei locations in each image are labeled. The framework consists of two learning stages. In the first stage, we design a semi-supervised strategy to learn a detection model from partially labeled nuclei locations. Specifically, an extended Gaussian mask is designed to train an initial model with partially labeled data. Then, selftraining with background propagation is proposed to make use of the unlabeled regions to boost nuclei detection and suppress false positives. In the second stage, a segmentation model is trained from the detected nuclei locations in a weakly-supervised fashion. Two types of coarse labels with complementary information are derived from the detected points and are then utilized to train a deep neural network. The fully-connected conditional random field loss is utilized in training to further refine the model without introducing extra computational complexity during inference. The proposed method is extensively evaluated on two nuclei segmentation datasets. The experimental results demonstrate that our method can achieve competitive performance compared to the fully supervised counterpart and the state-of-the-art methods while requiring significantly less annotation effort.

cs.CV↗

arXiv2020

Higher Kazhdan projections, $\ell_2$-Betti numbers and Baum-Connes conjectures

We introduce higher-dimensional analogs of Kazhdan projections in matrix algebras over group $C^*$-algebras and Roe algebras. These projections are constructed in the framework of cohomology with coefficients in unitary representations and in certain cases give rise to non-trivial $K$-theory classes. We apply the higher Kazhdan projections to establish a relation between $\ell_2$-Betti numbers of a group and surjectivity of different Baum-Connes type assembly maps.

math.OA↗

arXiv2020

Performance evaluation of on-chip wavelength conversion based on InP/In$_{1-x}$Ga$_x$As$_y$P$_{1-y}$ semiconductor waveguide platforms

We propose and design the high confinement InP/In1-xGaxAsyP1-y semiconductor waveguides and report the results of effective wavelength conversion based on this platform. Efficient confinement and mode field area fluctuation at different wavelength is analyzed to achieve the high nonlinear coefficient. The numerical results show that nearly zero phase-mismatch condition can be satisfied through dispersion tailoring of InP/In1-xGaxAsyP1-y waveguides, and the wavelength conversion ranging over 40 nm with the maximum conversion efficiency -26.3 dB is achieved for fixing pump power 100 mW. Meanwhile, the influences of the doping parameter y and pumping wavelength on the bandwidth and conversion efficiency are also discussed and optimized. It is indicated the excellent optical properties of the InP/In1-xGaxAsyP1-y waveguides and pave the way towards direct integration telecom band devices on stand semiconductor platforms.

physics.optics↗

arXiv2020

Quasi-local Algebras and Asymptotic Expanders

In this paper, we study the relation between the uniform Roe algebra and the uniform quasi-local algebra associated to a metric space of bounded geometry. In the process, we introduce a weakening of the notion of expanders, called asymptotic expanders. We show that being a sequence of asymptotic expanders is a coarse property under certain connectedness condition, and it implies non-uniformly local amenability. Moreover, we also analyse some $C^*$-algebraic properties of uniform quasi-local algebras. In particular, we show that a uniform quasi-local algebra is nuclear if and only if the underlying metric space has Property A.

math.OA↗

arXiv2020

Strict comparison for $C^*$-algebras arising from Almost finite groupoids

In this paper we show that for an almost finite minimal ample groupoid $G$, its reduced $\mathrm{C}^*$-algebra $C_r^*(G)$ has real rank zero and strict comparison even though $C_r^*(G)$ may not be nuclear in general. Moreover, if we further assume $G$ being also second countable and non-elementary, then its Cuntz semigroup ${\rm Cu}(C_r^*(G))$ is almost divisible and ${\rm Cu}(C_r^*(G))$ and ${\rm Cu}(C_r^*(G)\otimes \mathcal{Z})$ are canonically order-isomorphic, where $\mathcal{Z}$ denotes the Jiang-Su algebra.

math.OA↗

arXiv2020

Automatic Health Problem Detection from Gait Videos Using Deep Neural Networks

The aim of this study is developing an automatic system for detection of gait-related health problems using Deep Neural Networks (DNNs). The proposed system takes a video of patients as the input and estimates their 3D body pose using a DNN based method. Our code is publicly available at https://github.com/rmehrizi/multi-view-pose-estimation. The resulting 3D body pose time series are then analyzed in a classifier, which classifies input gait videos into four different groups including Healthy, with Parkinsons disease, Post Stroke patient, and with orthopedic problems. The proposed system removes the requirement of complex and heavy equipment and large laboratory space, and makes the system practical for home use. Moreover, it does not need domain knowledge for feature engineering since it is capable of extracting semantic and high level features from the input data. The experimental results showed the classification accuracy of 56% to 96% for different groups. Furthermore, only 1 out of 25 healthy subjects were misclassified (False positive), and only 1 out of 70 patients were classified as a healthy subject (False negative). This study presents a starting point toward a powerful tool for automatic classification of gait disorders and can be used as a basis for future applications of Deep Learning in clinical gait analysis. Since the system uses digital cameras as the only required equipment, it can be employed in domestic environment of patients and elderly people for consistent gait monitoring and early detection of gait alterations.

cs.CV↗

arXiv2019

Incorporating demand response of electric vehicles in scheduling of isolated microgrids with renewables using a bi-level programming approach

In this work, a novel optimal scheduling approach is proposed for isolated microgrids (MGs) with renewable generations by incorporating demand response of electric vehicles (EVs). First, a bi-level programming-based MG scheduling model is proposed under real-time pricing environments, where the upper- and lower- levels seek to minimize the MG net operating cost and the EV charging cost. Second, a hybrid solution algorithm called JAYA-interior point method is put forward to solve the model. And finally, the simulation results demonstrate that incorporating demand response of electric vehicles is able to guide EV users to actively participate in MG scheduling and achieve the peak load shaving, which offers a fundamental way to balance the interests between MG and EV users.

eess.SP↗

arXiv2019

Collaborative Multi-agent Learning for MR Knee Articular Cartilage Segmentation

The 3D morphology and quantitative assessment of knee articular cartilages (i.e., femoral, tibial, and patellar cartilage) in magnetic resonance (MR) imaging is of great importance for knee radiographic osteoarthritis (OA) diagnostic decision making. However, effective and efficient delineation of all the knee articular cartilages in large-sized and high-resolution 3D MR knee data is still an open challenge. In this paper, we propose a novel framework to solve the MR knee cartilage segmentation task. The key contribution is the adversarial learning based collaborative multi-agent segmentation network. In the proposed network, we use three parallel segmentation agents to label cartilages in their respective region of interest (ROI), and then fuse the three cartilages by a novel ROI-fusion layer. The collaborative learning is driven by an adversarial sub-network. The ROI-fusion layer not only fuses the individual cartilages from multiple agents, but also backpropagates the training loss from the adversarial sub-network to each agent to enable joint learning of shape and spatial constraints. Extensive evaluations are conducted on a dataset including hundreds of MR knee volumes with diverse populations, and the proposed method shows superior performance.

eess.IV↗

arXiv2019

Boundary and Entropy-driven Adversarial Learning for Fundus Image Segmentation

Accurate segmentation of the optic disc (OD) and cup (OC)in fundus images from different datasets is critical for glaucoma disease screening. The cross-domain discrepancy (domain shift) hinders the generalization of deep neural networks to work on different domain datasets.In this work, we present an unsupervised domain adaptation framework,called Boundary and Entropy-driven Adversarial Learning (BEAL), to improve the OD and OC segmentation performance, especially on the ambiguous boundary regions. In particular, our proposed BEAL frame-work utilizes the adversarial learning to encourage the boundary prediction and mask probability entropy map (uncertainty map) of the target domain to be similar to the source ones, generating more accurate boundaries and suppressing the high uncertainty predictions of OD and OC segmentation. We evaluate the proposed BEAL framework on two public retinal fundus image datasets (Drishti-GS and RIM-ONE-r3), and the experiment results demonstrate that our method outperforms the state-of-the-art unsupervised domain adaptation methods. Codes will be available at https://github.com/EmmaW8/BEAL.

cs.CV↗

arXiv2019

Structure and $K$-theory of $\ell^p$ uniform Roe algebras

In this paper, we characterize when the $\ell^p$ uniform Roe algebra of a metric space with bounded geometry is (stably) finite and when it is properly infinite in standard form for $p\in [1,\infty)$. Moreover, we show that the $\ell^p$ uniform Roe algebra is a (non-sequential) spatial $L^p$ AF algebra in the sense of Phillips and Viola if and only if the underlying metric space has asymptotic dimension zero. We also consider the ordered $K_0$ groups of $\ell^p$ uniform Roe algebras for metric spaces with low asymptotic dimension, showing that (1) the ordered $K_0$ group is trivial when the metric space is non-amenable and has asymptotic dimension at most one, and (2) when the metric space is a countable locally finite group, the (ordered) $K_0$ group is a complete invariant for the (bijective) coarse equivalence class of the underlying locally finite group. It happens that in both cases the ordered $K_0$ group does not depend on $p\in [1,\infty)$.

math.FA↗