arXiv subjects

Zhen Li

Publications and source records attributed to Zhen Li.

At least 415 records · Page 23Linked to original sources

arXiv2020

Multi-Modal Active Learning for Automatic Liver Fibrosis Diagnosis based on Ultrasound Shear Wave Elastography

With the development of radiomics, noninvasive diagnosis like ultrasound (US) imaging plays a very important role in automatic liver fibrosis diagnosis (ALFD). Due to the noisy data, expensive annotations of US images, the application of Artificial Intelligence (AI) assisting approaches encounters a bottleneck. Besides, the use of mono-modal US data limits the further improve of the classification results. In this work, we innovatively propose a multi-modal fusion network with active learning (MMFN-AL) for ALFD to exploit the information of multiple modalities, eliminate the noisy data and reduce the annotation cost. Four image modalities including US and three types of shear wave elastography (SWEs) are exploited. A new dataset containing these modalities from 214 candidates is well-collected and pre-processed, with the labels obtained from the liver biopsy results. Experimental results show that our proposed method outperforms the state-of-the-art performance using less than 30% data, and by using only around 80% data, the proposed fusion network achieves high AUC 89.27% and accuracy 70.59%.

cs.CV↗

arXiv2020

Where to Look and How to Describe: Fashion Image Retrieval with an Attentional Heterogeneous Bilinear Network

Fashion products typically feature in compositions of a variety of styles at different clothing parts. In order to distinguish images of different fashion products, we need to extract both appearance (i.e., "how to describe") and localization (i.e.,"where to look") information, and their interactions. To this end, we propose a biologically inspired framework for image-based fashion product retrieval, which mimics the hypothesized twostream visual processing system of human brain. The proposed attentional heterogeneous bilinear network (AHBN) consists of two branches: a deep CNN branch to extract fine-grained appearance attributes and a fully convolutional branch to extract landmark localization information. A joint channel-wise attention mechanism is further applied to the extracted heterogeneous features to focus on important channels, followed by a compact bilinear pooling layer to model the interaction of the two streams. Our proposed framework achieves satisfactory performance on three image-based fashion product retrieval benchmarks.

cs.CV↗

arXiv2020

Enhanced magnetocaloric effect and magnetic phase diagrams of single-crystal GdCrO$_3$

The crystalline structure, magnetism, and magnetocaloric effect of a GdCrO$_3$ single crystal grown with the laser-diode-heated floating-zone technique have been studied. The GdCrO$_3$ single crystal crystallizes into an orthorhombic structure with the space group $Pmnb$ at room temperature. Upon cooling, under a magnetic field of 0.1 T, it undergoes a magnetic phase transition at $T_{\textrm{N-Cr}} =$ 169.28(2) K with Cr$^{3+}$ ions forming a canted antiferromagnetic (AFM) structure, accompanied by a weak ferromagnetism. Subsequently, a spin reorientation takes place at $T_{\textrm{SR}} =$ 5.18(2) K due to Gd$^{3+}$-Cr$^{3+}$ magnetic couplings. Finally, the long-range AFM order of Gd$^{3+}$ ions establishes at $T_{\textrm{N-Gd}} =$ 2.10(2) K. Taking into account the temperature-(in)dependent components of Cr$^{3+}$ moments, we obtained an ideal model for describing the paramagnetic behavior of Gd$^{3+}$ ions within 30--140 K. We observed a magnetic reversal (positive $\rightarrow$ negative $\rightarrow$ positive) at 50 Oe with a minimum centering around 162 K. In the studied temperature range of 1.8-300 K, there exists a strong competition between magnetic susceptibilities of Gd$^{3+}$ and Cr$^{3+}$ ions, leading to puzzling magnetic phenomena. We have built the magnetic-field-dependent phase diagrams of $T_{\textrm{N-Gd}}$, $T_{\textrm{SR}}$, and $T_{\textrm{N-Cr}}$, shedding light on the nature of the intriguing magnetism. Moreover, we calculated the magnetic entropy change and obtained a maximum value at 6 K and $Δμ_0H$ = 14 T, i.e., -$ΔS_{\textrm{M}} \approx$ 57.5 J/kg K. Among all RCrO$_3$ (R = $4f^n$ rare earths, $n =$ 7-14) compounds, the single-crystal GdCrO$_3$ compound exhibits the highest magnetic entropy change, as well as an enhanced adiabatic temperature, creating a prominent magnetocaloric effect for potential application in magnetic refrigeration.

cond-mat.str-el↗

arXiv2020

Weak cosmic censorship conjecture in higher-dimensional black holes with nonlinear electrodynamic sources

The new version of the gedanken experiments proposed by Sorce and Wald are designed to test the validity of the weak cosmic censorship conjecture (WCCC) by overspinning or overcharging the Kerr-Newman black hole in Einstein-Maxwell gravity. Following their setup, in this paper, we investigate the WCCC in the higher-dimensional charged black hole with a nonlinear electrodynamics source. We derive the first and seconder order perturbation inequalities of the charged collision matter based on the Iyer-Wald formalism as well as the null energy conditions, and show that they share the similar form as that in Einstein-Maxwell gravity. As a result, we find that the static higher-dimensional nonlinear electrodynamics (HDNL) black holes cannot be overcharged after considering these two inequalities. Our result might indicate the validity of WCCC for more general HNDL and related systems.

gr-qc↗

arXiv2020

UXNet: Searching Multi-level Feature Aggregation for 3D Medical Image Segmentation

Aggregating multi-level feature representation plays a critical role in achieving robust volumetric medical image segmentation, which is important for the auxiliary diagnosis and treatment. Unlike the recent neural architecture search (NAS) methods that typically searched the optimal operators in each network layer, but missed a good strategy to search for feature aggregations, this paper proposes a novel NAS method for 3D medical image segmentation, named UXNet, which searches both the scale-wise feature aggregation strategies as well as the block-wise operators in the encoder-decoder network. UXNet has several appealing benefits. (1) It significantly improves flexibility of the classical UNet architecture, which only aggregates feature representations of encoder and decoder in equivalent resolution. (2) A continuous relaxation of UXNet is carefully designed, enabling its searching scheme performed in an efficient differentiable manner. (3) Extensive experiments demonstrate the effectiveness of UXNet compared with recent NAS methods for medical image segmentation. The architecture discovered by UXNet outperforms existing state-of-the-art models in terms of Dice on several public 3D medical image segmentation benchmarks, especially for the boundary locations and tiny tissues. The searching computational complexity of UXNet is cheap, enabling to search a network with the best performance less than 1.5 days on two TitanXP GPUs.

cs.CV↗

arXiv2020

Diagnosing Concept Drift with Visual Analytics

Concept drift is a phenomenon in which the distribution of a data stream changes over time in unforeseen ways, causing prediction models built on historical data to become inaccurate. While a variety of automated methods have been developed to identify when concept drift occurs, there is limited support for analysts who need to understand and correct their models when drift is detected. In this paper, we present a visual analytics method, DriftVis, to support model builders and analysts in the identification and correction of concept drift in streaming data. DriftVis combines a distribution-based drift detection method with a streaming scatterplot to support the analysis of drift caused by the distribution changes of data streams and to explore the impact of these changes on the model's accuracy. A quantitative experiment and two case studies on weather prediction and text classification have been conducted to demonstrate our proposed tool and illustrate how visual analytics can be used to support the detection, examination, and correction of concept drift.

cs.LG↗

arXiv2020

Attention-Guided Lightweight Network for Real-Time Segmentation of Robotic Surgical Instruments

The real-time segmentation of surgical instruments plays a crucial role in robot-assisted surgery. However, it is still a challenging task to implement deep learning models to do real-time segmentation for surgical instruments due to their high computational costs and slow inference speed. In this paper, we propose an attention-guided lightweight network (LWANet), which can segment surgical instruments in real-time. LWANet adopts encoder-decoder architecture, where the encoder is the lightweight network MobileNetV2, and the decoder consists of depthwise separable convolution, attention fusion block, and transposed convolution. Depthwise separable convolution is used as the basic unit to construct the decoder, which can reduce the model size and computational costs. Attention fusion block captures global contexts and encodes semantic dependencies between channels to emphasize target regions, contributing to locating the surgical instrument. Transposed convolution is performed to upsample feature maps for acquiring refined edges. LWANet can segment surgical instruments in real-time while takes little computational costs. Based on 960*544 inputs, its inference speed can reach 39 fps with only 3.39 GFLOPs. Also, it has a small model size and the number of parameters is only 2.06 M. The proposed network is evaluated on two datasets. It achieves state-of-the-art performance 94.10% mean IOU on Cata7 and obtains a new record on EndoVis 2017 with a 4.10% increase on mean IOU.

cs.CV↗

arXiv2020

Ultrasound Liver Fibrosis Diagnosis using Multi-indicator guided Deep Neural Networks

Accurate analysis of the fibrosis stage plays very important roles in follow-up of patients with chronic hepatitis B infection. In this paper, a deep learning framework is presented for automatically liver fibrosis prediction. On contrary of previous works, our approach can take use of the information provided by multiple ultrasound images. An indicator-guided learning mechanism is further proposed to ease the training of the proposed model. This follows the workflow of clinical diagnosis and make the prediction procedure interpretable. To support the training, a dataset is well-collected which contains the ultrasound videos/images, indicators and labels of 229 patients. As demonstrated in the experimental results, our proposed model shows its effectiveness by achieving the state-of-the-art performance, specifically, the accuracy is 65.6%(20% higher than previous best).

eess.IV↗

arXiv2020

Super-Necking Crystal Growth and Structural and Magnetic Properties of SrTb$_2$O$_4$ Single Crystals

We report on single-crystal growths of the SrTb$_2$O$_4$ compound by a super-necking technique with a laser-floating-zone furnace and study the stoichiometry, growth mode, and structural and magnetic properties by scanning electronic microscopy, neutron Laue, X-ray powder diffraction, and the physical property measurement system. We optimized the growth parameters, mainly the growth speed, atmosphere, and the addition of a Tb$_4$O$_7$ raw material. Neutron Laue diffraction displays the characteristic feature of a single crystal. Our study reveals an atomic ratio of Sr:Tb $ = 0.97(2){:}2.00(1)$ and a possible layer by layer crystal growth mode. Our X-ray powder diffraction study determines the crystal structure, lattice constants and atomic positions. The paramagnetic (PM) Curie--Weiss (CW) temperature $θ_{\texttt{CW}} =$ 5.00(4) K, and the effective PM moment $M^{\texttt{eff}}_{\texttt{mea}} =$ 10.97(1) $μ_\texttt{B}$ per Tb$^{3+}$ ion. The data of magnetization versus temperature can be divided into three regimes, showing a coexistence of antiferromagnetic and ferromagnetic interactions. This probably leads to the magnetic frustration in the SrTb$_2$O$_4$ compound. The magnetization at 2 K and 14 T originates from both the Tb1 and Tb2 sites and is strongly frustrated with an expected saturation field at $\sim$41.5 T, displaying an intricate phase diagram with three ranges.

cond-mat.mtrl-sci↗

arXiv2020

IllumiNet: Transferring Illumination from Planar Surfaces to Virtual Objects in Augmented Reality

This paper presents an illumination estimation method for virtual objects in real environment by learning. While previous works tackled this problem by reconstructing high dynamic range (HDR) environment maps or the corresponding spherical harmonics, we do not seek to recover the lighting environment of the entire scene. Given a single RGB image, our method directly infers the relit virtual object by transferring the illumination features extracted from planar surfaces in the scene to the desired geometries. Compared to previous works, our approach is more robust as it works in both indoor and outdoor environments with spatially-varying illumination. Experiments and evaluation results show that our approach outperforms the state-of-the-art quantitatively and qualitatively, achieving realistic augmented experience.

cs.CV↗

arXiv2020

A Redistribution Tool for Long-Term Archive of Astronomical Observation Data

Astronomical observation data require long-term preservation, and the rapid accumulation of observation data makes it necessary to consider the cost of long-term archive storage. In addition to low-speed disk-based online storage, optical disk or tape-based offline storage can be used to save costs. However, for astronomical research that requires historical data (particularly time-domain astronomy), the performance and energy consumption of data-accessing techniques cause problems because the requested data (which are organized according to observation time) may be located across multiple storage devices. In this study, we design and develop a tool referred to as AstroLayout to redistribute the observation data using spatial aggregation. The core algorithm uses graph partitioning to generate an optimized data placement according to the original observation data statistics and the target storage system. For the given observation data, AstroLayout can copy the long-term archive in the target storage system in accordance with this placement. An efficiency evaluation shows that AstroLayout can reduce the number of devices activated when responding to data-access requests in time-domain astronomy research. In addition to improving the performance of data-accessing techniques, AstroLayout can also reduce the storage systems power consumption. For enhanced adaptability, it supports storage systems of any media, including optical disks, tapes, and hard disks.

astro-ph.IM↗

arXiv2020

A Robust Attentional Framework for License Plate Recognition in the Wild

Recognizing car license plates in natural scene images is an important yet still challenging task in realistic applications. Many existing approaches perform well for license plates collected under constrained conditions, eg, shooting in frontal and horizontal view-angles and under good lighting conditions. However, their performance drops significantly in an unconstrained environment that features rotation, distortion, occlusion, blurring, shading or extreme dark or bright conditions. In this work, we propose a robust framework for license plate recognition in the wild. It is composed of a tailored CycleGAN model for license plate image generation and an elaborate designed image-to-sequence network for plate recognition. On one hand, the CycleGAN based plate generation engine alleviates the exhausting human annotation work. Massive amount of training data can be obtained with a more balanced character distribution and various shooting conditions, which helps to boost the recognition accuracy to a large extent. On the other hand, the 2D attentional based license plate recognizer with an Xception-based CNN encoder is capable of recognizing license plates with different patterns under various scenarios accurately and robustly. Without using any heuristics rule or post-processing, our method achieves the state-of-the-art performance on four public datasets, which demonstrates the generality and robustness of our framework. Moreover, we released a new license plate dataset, named "CLPD", with 1200 images from all 31 provinces in mainland China. The dataset can be available from: https://github.com/wangpengnorman/CLPD_dataset.

cs.CV↗

arXiv2020

BARNet: Bilinear Attention Network with Adaptive Receptive Fields for Surgical Instrument Segmentation

Surgical instrument segmentation is extremely important for computer-assisted surgery. Different from common object segmentation, it is more challenging due to the large illumination and scale variation caused by the special surgical scenes. In this paper, we propose a novel bilinear attention network with adaptive receptive field to solve these two challenges. For the illumination variation, the bilinear attention module can capture second-order statistics to encode global contexts and semantic dependencies between local pixels. With them, semantic features in challenging areas can be inferred from their neighbors and the distinction of various semantics can be boosted. For the scale variation, our adaptive receptive field module aggregates multi-scale features and automatically fuses them with different weights. Specifically, it encodes the semantic relationship between channels to emphasize feature maps with appropriate scales, changing the receptive field of subsequent convolutions. The proposed network achieves the best performance 97.47% mean IOU on Cata7 and comes first place on EndoVis 2017 by 10.10% IOU overtaking second-ranking method.

cs.CV↗

arXiv2020

Active- and transfer-learning applied to microscale-macroscale coupling to simulate viscoelastic flows

Active- and transfer-learning are applied to polymer flows for the multiscale discovery of effective constitutive approximations required in viscoelastic flow simulation. The result is macroscopic rheology directly connected to a microstructural model. Micro and macroscale simulations are adaptively coupled by means of Gaussian process regression to run the expensive microscale computations only as necessary. This active-learning guided multiscale method can automatically detect the inaccuracy of the learned constitutive closure and initiate simulations at new sampling points informed by proper acquisition functions, leading to an autonomic microscale-macroscale coupled system. Also, we develop a new dissipative particle dynamics model with the range of interaction cutoff between particles allowed to vary with the local strain-rate invariant, which is able to capture both the shear-thinning viscosity and the normal stress difference functions consistent with rheological experiments for aqueous polyacrylamide solutions. Our numerical experiments demonstrate the effectiveness of using active- and transfer-learning schemes to on-the-fly couple a spectral element solver and a mesoscopic particle-based simulator, and verify that the microscale-macroscale coupled model with effective constitutive closure learned from microscopic dynamics can outperform empirical constitutive models compared to experimental observations. The effective closure learned in a channel simulation is then transferred directly to the flow past a circular cylinder, where the results show that only two additional microscopic simulations are required to achieve a satisfactory constitutive model to once again close the continuum equations. This new paradigm of active- and transfer-learning for multiscale modeling is readily applicable to other microscale-macroscale coupled simulations of complex fluids and other materials.

physics.comp-ph↗

arXiv2020

PointASNL: Robust Point Clouds Processing using Nonlocal Neural Networks with Adaptive Sampling

Raw point clouds data inevitably contains outliers or noise through acquisition from 3D sensors or reconstruction algorithms. In this paper, we present a novel end-to-end network for robust point clouds processing, named PointASNL, which can deal with point clouds with noise effectively. The key component in our approach is the adaptive sampling (AS) module. It first re-weights the neighbors around the initial sampled points from farthest point sampling (FPS), and then adaptively adjusts the sampled points beyond the entire point cloud. Our AS module can not only benefit the feature learning of point clouds, but also ease the biased effect of outliers. To further capture the neighbor and long-range dependencies of the sampled point, we proposed a local-nonlocal (L-NL) module inspired by the nonlocal operation. Such L-NL module enables the learning process insensitive to noise. Extensive experiments verify the robustness and superiority of our approach in point clouds processing tasks regardless of synthesis data, indoor data, and outdoor data with or without noise. Specifically, PointASNL achieves state-of-the-art robust performance for classification and segmentation tasks on all datasets, and significantly outperforms previous methods on real-world outdoor SemanticKITTI dataset with considerate noise. Our code is released through https://github.com/yanx27/PointASNL.

cs.CV↗

arXiv2020

Multimode silicon photonics using on-chip geometrical-optics

On-chip optical interconnect has been widely accepted as a promising technology to realize future large-scale multiprocessors. Mode-division multiplexing (MDM) provides a new degree of freedom for optical interconnects to dramatically increase the link capacity. Present on-chip multimode devices are based on traditional wave-optics. Although large amount of computation and optimization are adopted to support more modes, mode-independent manipulation is still hard to be achieved due to severe mode dispersion. Here, we propose a universal solution to standardize the design of fundamental multimode building blocks, by introducing a geometrical-optics-like concept adopting waveguide width larger than the working wavelength. The proposed solution can tackle a group of modes at the same time with very simple processes, avoiding demultiplexing procedure and ensuring compact footprint. Compare to conventional schemes, it is scalable to larger mode channels without increasing the complexity and whole footprint. As a proof of concept, we demonstrate a set of multimode building blocks including crossing, bend, coupler and switches. Low losses of multimode waveguide crossing and bend are achieved, as well as ultra-low power consumption of the multimode switch is realized since it enables reconfigurable routing for a group of modes simultaneously. Our work promotes the multimode photonics research and makes the MDM technique more practical.

physics.optics↗

arXiv2020

Exemplar Normalization for Learning Deep Representation

Normalization techniques are important in different advanced neural networks and different tasks. This work investigates a novel dynamic learning-to-normalize (L2N) problem by proposing Exemplar Normalization (EN), which is able to learn different normalization methods for different convolutional layers and image samples of a deep network. EN significantly improves flexibility of the recently proposed switchable normalization (SN), which solves a static L2N problem by linearly combining several normalizers in each normalization layer (the combination is the same for all samples). Instead of directly employing a multi-layer perceptron (MLP) to learn data-dependent parameters as conditional batch normalization (cBN) did, the internal architecture of EN is carefully designed to stabilize its optimization, leading to many appealing benefits. (1) EN enables different convolutional layers, image samples, categories, benchmarks, and tasks to use different normalization methods, shedding light on analyzing them in a holistic view. (2) EN is effective for various network architectures and tasks. (3) It could replace any normalization layers in a deep network and still produce stable model training. Extensive experiments demonstrate the effectiveness of EN in a wide spectrum of tasks including image recognition, noisy label learning, and semantic segmentation. For example, by replacing BN in the ordinary ResNet50, improvement produced by EN is 300% more than that of SN on both ImageNet and the noisy WebVision dataset.

cs.CV↗

arXiv2020

Unifying Specialist Image Embedding into Universal Image Embedding

Deep image embedding provides a way to measure the semantic similarity of two images. It plays a central role in many applications such as image search, face verification, and zero-shot learning. It is desirable to have a universal deep embedding model applicable to various domains of images. However, existing methods mainly rely on training specialist embedding models each of which is applicable to images from a single domain. In this paper, we study an important but unexplored task: how to train a single universal image embedding model to match the performance of several specialists on each specialist's domain. Simply fusing the training data from multiple domains cannot solve this problem because some domains become overfitted sooner when trained together using existing methods. Therefore, we propose to distill the knowledge in multiple specialists into a universal embedding to solve this problem. In contrast to existing embedding distillation methods that distill the absolute distances between images, we transform the absolute distances between images into a probabilistic distribution and minimize the KL-divergence between the distributions of the specialists and the universal embedding. Using several public datasets, we validate that our proposed method accomplishes the goal of universal image embedding.

cs.CV↗