arXiv subjects

Yun Zheng

Publications and source records attributed to Yun Zheng.

60 records · Page 4Linked to original sources

arXiv2021

Large-Scale Visual Search with Binary Distributed Graph at Alibaba

Graph-based approximate nearest neighbor search has attracted more and more attentions due to its online search advantages. Numbers of methods studying the enhancement of speed and recall have been put forward. However, few of them focus on the efficiency and scale of offline graph-construction. For a deployed visual search system with several billions of online images in total, building a billion-scale offline graph in hours is essential, which is almost unachievable by most existing methods. In this paper, we propose a novel algorithm called Binary Distributed Graph to solve this problem. Specifically, we combine binary codes with graph structure to speedup online and offline procedures, and achieve comparable performance with the ones in real-value based scenarios by recalling more binary candidates. Furthermore, the graph-construction is optimized to completely distributed implementation, which significantly accelerates the offline process and gets rid of the limitation of memory and disk within a single machine. Experimental comparisons on Alibaba Commodity Data Set (more than three billion images) show that the proposed method outperforms the state-of-the-art with respect to the online/offline trade-off.

cs.IR↗

arXiv2020

Weakly Supervised Learning with Side Information for Noisy Labeled Images

In many real-world datasets, like WebVision, the performance of DNN based classifier is often limited by the noisy labeled data. To tackle this problem, some image related side information, such as captions and tags, often reveal underlying relationships across images. In this paper, we present an efficient weakly supervised learning by using a Side Information Network (SINet), which aims to effectively carry out a large scale classification with severely noisy labels. The proposed SINet consists of a visual prototype module and a noise weighting module. The visual prototype module is designed to generate a compact representation for each category by introducing the side information. The noise weighting module aims to estimate the correctness of each noisy image and produce a confidence score for image ranking during the training procedure. The propsed SINet can largely alleviate the negative impact of noisy image labels, and is beneficial to train a high performance CNN based classifier. Besides, we released a fine-grained product dataset called AliProducts, which contains more than 2.5 million noisy web images crawled from the internet by using queries generated from 50,000 fine-grained semantic classes. Extensive experiments on several popular benchmarks (i.e. Webvision, ImageNet and Clothing-1M) and our proposed AliProducts achieve state-of-the-art performance. The SINet has won the first place in the classification task on WebVision Challenge 2019, and outperformed other competitors by a large margin.

cs.CV↗

arXiv2020

Intrinsic radiation background of LaBr$_3$(Ce) detector via coincidence measurements and simulations

The LaBr$_3$(Ce) detector has attracted much attention in recent years for its superior characteristics to other scintillating materials in terms of resolution and efficiency. However, it has relatively high intrinsic radiation background due to the naturally occurring radioisotope in lanthanum, actinium and their daughter nuclei. This limits its applications in low counting rate experiments. In this paper, we identified the radioactive isotopes in the $\phi3''\times 3''$ Saint-Gobain B380 detector by a coincidence measurement using a Clover detector in a low-background shielding system. Moreover, we carried out a Geant4 simulation to the experimental spectra to evaluate the activities of the main internal radiation components. % by means of the $α$, $γ$ and $β$ decays from the $^{138}$La and $^{227}$Ac. The activity of radiation background of B380 is determined to be 1.480 (69) Bq/cm$^3$, the main sources of which include $^{138}$La of 1.425 (59) Bq/cm$^3$, $^{211}$Bi of 0.0136 (15) Bq/cm$^3$, $^{219}$Rn of 0.0125 (17) Bq/cm$^3$, $^{223}$Ra of 0.0127 (14) Bq/cm$^3$, and $^{227}$Th of 0.0158 (22) Bq/cm$^3$.

physics.ins-det↗

arXiv2019

The $ΔI$=2 bands in $^{109}$In: possible antimagnetic rotation

The high-spin structure of $^{109}$In was investigated with the $^{100}$Mo($^{14}$N, 5$n$)$^{109}$In fusion-evaporation reaction at CIAE, Beijing. Eleven new $γ$-rays of $^{109}$In were identified, by which the bandheads of the $ΔI$=2 rotational bands were confirmed. The configurations were assigned with the help of the systematic discussion. Furthermore, the rotational bands are compared with the tilted-axis cranking calculations based on a relativistic mean-field approach. The rotational bands involving the $1p1h$ excitation to the $π$$d_{5/2}$ and $π$$g_{7/2}$ orbitals are suggested as candidates for antimagnetic rotation based on the theoretical results.

nucl-ex↗

arXiv2016

Mini-Orange Spectrometer at CIAE

A Mini-Orange spectrometer used for in-beam measurements of internal conversion electrons, which consists of a Si(Li) detector and different sets of SmO$_5$ permanent magnets for filtering and transporting the conversion electrons to the Si(Li) detector, has been developed at China Institute of Atomic Energy. The working principle and configuration of the Mini-Orange spectrometer are described. The performance of the setup is illustrated by measured singles conversion electron spectra using the Mini-Orange spectrometer.

physics.ins-det↗

arXiv2015

High-spin level structure of the neutron-rich nucleus 91Y

High-spin level structure of the neutron-rich nucleus 91Y has been reinvestigated via the 82Se(13C, p3n)91Y reaction. A newly constructed level scheme including several key levels clarifies the uncertainties in the earlier studies. These levels are characterized by the breaking of the Z=38 and N=56 subshell closures, which involves in the spin-isospin dependent central force and tensor force.

nucl-ex↗