arXiv subjects

Yu Zhou

Publications and source records attributed to Yu Zhou.

At least 361 records · Page 20Linked to original sources

arXiv2021

Rail-5k: a Real-World Dataset for Rail Surface Defects Detection

This paper presents the Rail-5k dataset for benchmarking the performance of visual algorithms in a real-world application scenario, namely the rail surface defects detection task. We collected over 5k high-quality images from railways across China, and annotated 1100 images with the help from railway experts to identify the most common 13 types of rail defects. The dataset can be used for two settings both with unique challenges, the first is the fully-supervised setting using the 1k+ labeled images for training, fine-grained nature and long-tailed distribution of defect classes makes it hard for visual algorithms to tackle. The second is the semi-supervised learning setting facilitated by the 4k unlabeled images, these 4k images are uncurated containing possible image corruptions and domain shift with the labeled images, which can not be easily tackle by previous semi-supervised learning methods. We believe our dataset could be a valuable benchmark for evaluating robustness and reliability of visual algorithms.

cs.CV↗

arXiv2021

All-fiber photonic lantern multimode optical receiver with coherent adaptive optics beam combining

A multimode optical receiver for free space optical communications (FSOC) based on a photonic lantern and adaptive optics coherent beam combining (CBC) of the lantern's single-mode outputs is proposed and demonstrated for the first time. The use of optical coherent combining in fiber serves to increase the signal to noise ratio compared to similar receivers based on electrically combined signals, and represents an all-fiber approach to low-order adaptive optics. This optical receiver is demonstrated using a photonic lantern with three outputs, fibre couplers and active phase locking, and further investigated under atmospheric conditions with and without turbulence.

physics.optics↗

arXiv2021

Exploring Instance Relations for Unsupervised Feature Embedding

Despite the great progress achieved in unsupervised feature embedding, existing contrastive learning methods typically pursue view-invariant representations through attracting positive sample pairs and repelling negative sample pairs in the embedding space, while neglecting to systematically explore instance relations. In this paper, we explore instance relations including intra-instance multi-view relation and inter-instance interpolation relation for unsupervised feature embedding. Specifically, we embed intra-instance multi-view relation by aligning the distribution of the distance between an instance's different augmented samples and negative samples. We explore inter-instance interpolation relation by transferring the ratio of information for image sample interpolation from pixel space to feature embedding space. The proposed approach, referred to as EIR, is simple-yet-effective and can be easily inserted into existing view-invariant contrastive learning based methods. Experiments conducted on public benchmarks for image classification and retrieval report state-of-the-art or comparable performance.

cs.CV↗

arXiv2021

Photon statistics of superbunching pseudothermal light

Superbunching pseudothermal light has important applications in studying the second- and higher-order interference of light in quantum optics. Unlike the photon statistics of thermal or pseudothermal light is well understood, the photon statistics of superbunching pseudothermal light has not been studied yet. In this paper, we will employ single-photon detectors to measure the photon statistics of superbunching pseudothermal light and calculate the degree of second-order coherence. It is found that the larger the value of the degree of second-order coherence of superbunching pseudothermal light is, the more the measured photon distribution deviates from the one of thermal or pseudothermal light in the tail part. The results are helpful to understand the physics of two-photon superbunching with classical light. It is suggested that superbunching pseudothermal light can be employed to generate non-Rayleigh temporal speckles.

quant-ph↗

arXiv2021

Boundary-induced and scene-aggregated network for monocular depth prediction

Monocular depth prediction is an important task in scene understanding. It aims to predict the dense depth of a single RGB image. With the development of deep learning, the performance of this task has made great improvements. However, two issues remain unresolved: (1) The deep feature encodes the wrong farthest region in a scene, which leads to a distorted 3D structure of the predicted depth; (2) The low-level features are insufficient utilized, which makes it even harder to estimate the depth near the edge with sudden depth change. To tackle these two issues, we propose the Boundary-induced and Scene-aggregated network (BS-Net). In this network, the Depth Correlation Encoder (DCE) is first designed to obtain the contextual correlations between the regions in an image, and perceive the farthest region by considering the correlations. Meanwhile, the Bottom-Up Boundary Fusion (BUBF) module is designed to extract accurate boundary that indicates depth change. Finally, the Stripe Refinement module (SRM) is designed to refine the dense depth induced by the boundary cue, which improves the boundary accuracy of the predicted depth. Several experimental results on the NYUD v2 dataset and \xff{the iBims-1 dataset} illustrate the state-of-the-art performance of the proposed approach. And the SUN-RGBD dataset is employed to evaluate the generalization of our method. Code is available at https://github.com/XuefengBUPT/BS-Net.

cs.CV↗

arXiv2021

Manipulating Berry curvature of SrRuO3 thin films via epitaxial strain

Berry curvature plays a crucial role in exotic electronic states of quantum materials, such as intrinsic anomalous Hall effect. As Berry curvature is highly sensitive to subtle changes of electronic band structures, it can be finely tuned via external stimulus. Here, we demonstrate in SrRuO3 thin films that both the magnitude and sign of anomalous Hall resistivity can be effectively controlled with epitaxial strain. Our first-principles calculations reveal that epitaxial strain induces an additional crystal field splitting and changes the order of Ru d orbital energies, which alters the Berry curvature and leads to the sign and magnitude change of anomalous Hall conductivity. Furthermore, we show that the rotation of Ru magnetic moment in real space of tensile strained sample can result in an exotic nonmonotonic change of anomalous Hall resistivity with the sweeping of magnetic field, resembling the topological Hall effect observed in non-coplanar spin systems. These findings not only deepen our understanding of anomalous Hall effect in SrRuO3 systems, but also provide an effective tuning knob to manipulate Berry curvature and related physical properties in a wide range of quantum materials.

cond-mat.mtrl-sci↗

arXiv2021

A Physics-based and Data-driven Linear Three-Phase Power Flow Model for Distribution Power Systems

Distribution power systems (DPSs) are mostly unbalanced, and their loads may have notable static voltage characteristics (ZIP loads). Hence, despite abundant papers on linear single-phase power flow models, it is still necessary to study linear three-phase distribution power flow models. To this end, this paper proposes a physics-based and data-driven linear three-phase power flow model for DPSs. We first formulate how to amalgamate data-driven techniques into a physics-based power flow model to obtain our linear model. This amalgamation makes our linear model independent of the assumptions commonly used in the literature (e.g., nodal voltages are nearly 1.0 p.u.) and thus have a relatively high accuracy generally - even when those assumptions become invalid. We then reveal how to apply our model to the DPSs with ZIP loads. We also show that with the Huber penalty function employed, the adverse impact of bad data on our model's accuracy is significantly reduced, rendering our model robust against poor data quality. Case studies have demonstrated that our model generally has 2 to over 10-fold smaller average errors than other linear power flow models, enjoys a satisfying accuracy against bad data, and facilitates a faster solution to DPS analysis and optimization problems.

eess.SY↗

arXiv2021

Simple and efficient way to generate superbunching pseudothermal light

By modulating the intensity of laser light before the rotating groundglass, the well-known pseudothermal light source can be modified into superbunching pseudothermal light source, in which the degree of second-order coherence of the scattered light is larger than 2. With the modulated intensities following binary distribution, we experimentally observed the degree of second- and third-order coherence equaling 20.45 and 227.07, which is much larger than the value of thermal or pseudothermal light, 2 and 6, respectively. Numerical simulation predicts that the degree of second-order coherence can be further improved by tuning the parameters of binary distribution. It is also predicted that the quality of temporal ghost imaging can be improved with this superbunching pseudothermal light. This simple and efficient superbunching pseudothermal light source provides an interesting alternative to study the second- and higher-order interference of light in these scenarios where thermal or pseudothermal light source were employed.

physics.optics↗

arXiv2021

Boosting API Recommendation with Implicit Feedback

Developers often need to use appropriate APIs to program efficiently, but it is usually a difficult task to identify the exact one they need from a vast of candidates. To ease the burden, a multitude of API recommendation approaches have been proposed. However, most of the currently available API recommenders do not support the effective integration of users' feedback into the recommendation loop. In this paper, we propose a framework, BRAID (Boosting RecommendAtion with Implicit FeeDback), which leverages learning-to-rank and active learning techniques to boost recommendation performance. By exploiting users' feedback information, we train a learning-to-rank model to re-rank the recommendation results. In addition, we speed up the feedback learning process with active learning. Existing query-based API recommendation approaches can be plugged into BRAID. We select three state-of-the-art API recommendation approaches as baselines to demonstrate the performance enhancement of BRAID measured by Hit@k (Top-k), MAP, and MRR. Empirical experiments show that, with acceptable overheads, the recommendation performance improves steadily and substantially with the increasing percentage of feedback data, comparing with the baselines.

cs.SE↗

arXiv2021

Quick Annotator: an open-source digital pathology based rapid image annotation tool

Image based biomarker discovery typically requires an accurate segmentation of histologic structures (e.g., cell nuclei, tubules, epithelial regions) in digital pathology Whole Slide Images (WSI). Unfortunately, annotating each structure of interest is laborious and often intractable even in moderately sized cohorts. Here, we present an open-source tool, Quick Annotator (QA), designed to improve annotation efficiency of histologic structures by orders of magnitude. While the user annotates regions of interest (ROI) via an intuitive web interface, a deep learning (DL) model is concurrently optimized using these annotations and applied to the ROI. The user iteratively reviews DL results to either (a) accept accurately annotated regions, or (b) correct erroneously segmented structures to improve subsequent model suggestions, before transitioning to other ROIs. We demonstrate the effectiveness of QA over comparable manual efforts via three use cases. These include annotating (a) 337,386 nuclei in 5 pancreatic WSIs, (b) 5,692 tubules in 10 colorectal WSIs, and (c) 14,187 regions of epithelium in 10 breast WSIs. Efficiency gains in terms of annotations per second of 102x, 9x, and 39x were respectively witnessed while retaining f-scores >.95, suggesting QA may be a valuable tool for efficiently fully annotating WSIs employed in downstream biomarker studies.

eess.IV↗

arXiv2020

Classification and Generation of Light Sources Using Gamma Fitting

In general, the typical approach to discriminate antibunching, bunching or superbunching categories make use of calculating the second-order coherence function ${g^{(2)}}(τ)$ of light. Although the classical light sources correspond to the specific degree of second-order coherence ${g^{(2)}}(0)$, it does not alone constitute a distinguishable metric to characterize and determine light sources. Here we propose a new mechanism to directly classify and generate antibunching, bunching or superbunching categories of light, as well as the classical light sources such as thermal and coherent light, by Gamma fitting according to only one characteristic parameter $α$ or $β$. Experimental verification of beams from four-wave mixing process is in agreement with the presented mechanism, and the in fluence of temperature $T$ and laser detuning $Δ$ on the measured results are investigated. The proposal demonstrates the potential of classifying and identifying light with different nature, and the most importantly, provides a convenient and simple method to generate light sources meeting various application requirements according to the presented rules. Most notably, the bunching and superbunching are distinguishable in super-Poissonian statistics using our mechanism.

physics.optics↗

arXiv2020

Two-photon superbunching effect of broadband chaotic stationary light at femtosecond timescale based on cascaded Michelson interferometer

It is challenging for observing superbunching effect with true chaotic light, here we propose and demonstrate a method to achieve superbunching effect of the degree of second-order coherence is 2.42 with broadband stationary chaotic light based on a cascaded Michelson interferometer (CMI), exceeding the theoretical upper limit of 2 for the two-photon bunching effect of chaotic light. The superbunching correlation peak is measured with an ultrafast two-photon absorption detector which the full width at half maximum reaches about 95 fs. Two-photon superbunching theory in a CMI is developed to interpret the effect and is in agreement with experimental results. The theory also predicts that the degree of second-order coherence can be much greater than $2$ if chaotic light propagates $N$ times in a CMI. Finally, a new type of weak signals detection setup which employs broadband chaotic light circulating in a CMI is proposed. Theoretically, it can increase the detection sensitivity of weak signals 79 times after the chaotic light circulating 100 times in the CMI.

quant-ph↗

arXiv2020

Gaussian Constrained Attention Network for Scene Text Recognition

Scene text recognition has been a hot topic in computer vision. Recent methods adopt the attention mechanism for sequence prediction which achieve convincing results. However, we argue that the existing attention mechanism faces the problem of attention diffusion, in which the model may not focus on a certain character area. In this paper, we propose Gaussian Constrained Attention Network to deal with this problem. It is a 2D attention-based method integrated with a novel Gaussian Constrained Refinement Module, which predicts an additional Gaussian mask to refine the attention weights. Different from adopting an additional supervision on the attention weights simply, our proposed method introduces an explicit refinement. In this way, the attention weights will be more concentrated and the attention-based recognition network achieves better performance. The proposed Gaussian Constrained Refinement Module is flexible and can be applied to existing attention-based methods directly. The experiments on several benchmark datasets demonstrate the effectiveness of our proposed method. Our code has been available at https://github.com/Pay20Y/GCAN.

cs.CV↗

arXiv2020

Study on Leveraging Wind Farm Reactive Power Potential for Uncertain Power System Reactive Power Optimization

This paper suggests leveraging reactive power potential (RPP) embedded in wind farms to improve power system operational safety and optimality. First, three typical RPP provision approaches are analyzed and a two-stage robust linear optimization based RPP evaluation method is proposed. This approach yields an RPP range that ensures the security of wind farm operations under any realization of uncertainty regarding the wind farm. Simplified DistFlow equations are employed here for a compromise between computational accuracy and cost. Next, an uncertain RPP-involved reactive power optimization problem is introduced, through which system operators ensure system-wide security and optimality regarding the base case and against any possible deviation caused by uncertain lumped loads and renewable generation. Steady-state models of automatic generation control and local voltage control are also captured in this uncertain reactive power optimization, which is then transformed through Soyster's method into a deterministic optimization problem that is readily solvable. Case studies have conceptually validated that even with notable uncertainty, wind farms are still a competent reactive power resource providing considerable RPP. Also, simulation confirms positive and notable improvement of leveraging wind-farm RPP on system-wide operational security and optimality, especially for power systems with high wind penetration.

eess.SY↗

arXiv2020

Graded decorated marked surfaces: Calabi-Yau-$\mathbb{X}$ categories of gentle algebras

Let $\mathbf{S}$ be a graded marked surface. We construct a string model for Calabi-Yau-$\mathbb{X}$ category $\mathcal{D}_\mathbb{X}(\mathbf{S}_\bigtriangleup)$, via the graded DMS (=decorated marked surface) $\mathbf{S}_Δ$. We prove an isomorphism between the braid twist group of $\mathbf{S}_\bigtriangleup$ and the spherical twist group of $\mathcal{D}_\mathbb{X}(\mathbf{S}_\bigtriangleup)$, and $\mathbf{q}$-intersection formulas. We also give a topological realization of the Lagrangian immersion $\mathcal{D}_\infty(\mathbf{S})\to\mathcal{D}_\mathbb{X}(\mathbf{S}_\bigtriangleup)$, where $\mathcal{D}_\infty(\mathbf{S})$ is the topological Fukaya category associated to $\mathbf{S}$, that is triangle equivalent to the bounded derived category of some graded gentle algebra. This generalizes previous works of [Qiu, Qiu-Zhou] in the Calabi-Yau-3 case and and also unifies the Calabi-Yau-$\infty$ case $\mathcal{D}_\infty(\mathbf{S})$ (cf. [Haiden-Katzarkov-Kontsevich, Opper-Plamondon-Schroll]).

math.RT↗

arXiv2020

Exploring Relations in Untrimmed Videos for Self-Supervised Learning

Existing video self-supervised learning methods mainly rely on trimmed videos for model training. However, trimmed datasets are manually annotated from untrimmed videos. In this sense, these methods are not really self-supervised. In this paper, we propose a novel self-supervised method, referred to as Exploring Relations in Untrimmed Videos (ERUV), which can be straightforwardly applied to untrimmed videos (real unlabeled) to learn spatio-temporal features. ERUV first generates single-shot videos by shot change detection. Then a designed sampling strategy is used to model relations for video clips. The strategy is saved as our self-supervision signals. Finally, the network learns representations by predicting the category of relations between the video clips. ERUV is able to compare the differences and similarities of videos, which is also an essential procedure for action and video related tasks. We validate our learned models with action recognition and video retrieval tasks with three kinds of 3D CNNs. Experimental results show that ERUV is able to learn richer representations and it outperforms state-of-the-art self-supervised methods with significant margins.

cs.CV↗

arXiv2020

Two-Level Residual Distillation based Triple Network for Incremental Object Detection

Modern object detection methods based on convolutional neural network suffer from severe catastrophic forgetting in learning new classes without original data. Due to time consumption, storage burden and privacy of old data, it is inadvisable to train the model from scratch with both old and new data when new object classes emerge after the model trained. In this paper, we propose a novel incremental object detector based on Faster R-CNN to continuously learn from new object classes without using old data. It is a triple network where an old model and a residual model as assistants for helping the incremental model learning on new classes without forgetting the previous learned knowledge. To better maintain the discrimination of features between old and new classes, the residual model is jointly trained on new classes in the incremental learning procedure. In addition, a corresponding distillation scheme is designed to guide the training process, which consists of a two-level residual distillation loss and a joint classification distillation loss. Extensive experiments on VOC2007 and COCO are conducted, and the results demonstrate that the proposed method can effectively learn to incrementally detect objects of new classes, and the problem of catastrophic forgetting is mitigated in this context.

cs.CV↗

arXiv2020

Progressive Cluster Purification for Unsupervised Feature Learning

In unsupervised feature learning, sample specificity based methods ignore the inter-class information, which deteriorates the discriminative capability of representation models. Clustering based methods are error-prone to explore the complete class boundary information due to the inevitable class inconsistent samples in each cluster. In this work, we propose a novel clustering based method, which, by iteratively excluding class inconsistent samples during progressive cluster formation, alleviates the impact of noise samples in a simple-yet-effective manner. Our approach, referred to as Progressive Cluster Purification (PCP), implements progressive clustering by gradually reducing the number of clusters during training, while the sizes of clusters continuously expand consistently with the growth of model representation capability. With a well-designed cluster purification mechanism, it further purifies clusters by filtering noise samples which facilitate the subsequent feature learning by utilizing the refined clusters as pseudo-labels. Experiments on commonly used benchmarks demonstrate that the proposed PCP improves baseline method with significant margins. Our code will be available at https://github.com/zhangyifei0115/PCP.

cs.CV↗