Filters








156 Hits in 1.6 sec

The Features of Acting by Yang Mingkun

LIU Liu
2020 Management Studies  
Yang Mingkun 楊明坤 (1949-) is proud of his performance techniques in Yangzhou storytelling (Yangzhou pinghua 揚州評話).  ...  This paper aims to explore the acting features of Yang Mingkun in terms of kou, shou, shēn, bu, and shén.  ...  Today, Yang Mingkun is also regarded as a master of mimicry and humour.  ... 
doi:10.17265/1539-8072/2020.10.001 fatcat:nc2np6wxprch7hc5wa66ccn5z4

Dynamic latency speech recognition with asynchronous revision [article]

Mingkun Huang, Meng Cai, Jun Zhang, Yang Zhang, Yongbin You, Yi He, Zejun Ma
2020 arXiv   pre-print
In this work we propose an inference technique, asynchronous revision, to unify streaming and non-streaming speech recognition models. Specifically, we achieve dynamic latency with only one model by using arbitrary right context during inference. The model is composed of a stack of convolutional layers for audio encoding. In inference stage, the history states of encoder and decoder can be asynchronously revised to trade off between the latency and the accuracy of the model. To alleviate
more » ... g and inference mismatch, we propose a training technique, segment cropping, which randomly splits input utterances into several segments with forward connections. This allows us to have dynamic latency speech recognition results with large improvements in accuracy. Experiments show that our dynamic latency model with asynchronous revision gives 8\%-14\% relative improvements over the streaming models.
arXiv:2011.01570v1 fatcat:khgungukbfapfiun7mkxg5acge

Improving RNN transducer with normalized jointer network [article]

Mingkun Huang, Jun Zhang, Meng Cai, Yang Zhang, Jiali Yao, Yongbin You, Yi He, Zejun Ma
2020 arXiv   pre-print
Recurrent neural transducer (RNN-T) is a promising end-to-end (E2E) model in automatic speech recognition (ASR). It has shown superior performance compared to traditional hybrid ASR systems. However, training RNN-T from scratch is still challenging. We observe a huge gradient variance during RNN-T training and suspect it hurts the performance. In this work, we analyze the cause of the huge gradient variance in RNN-T training and proposed a new normalized jointer network to overcome it. We also
more » ... ropose to enhance the RNN-T network with a modified conformer encoder network and transformer-XL predictor networks to achieve the best performance. Experiments are conducted on the open 170-hour AISHELL-1 and industrial-level 30000-hour mandarin speech dataset. On the AISHELL-1 dataset, our RNN-T system gets state-of-the-art results on AISHELL-1's streaming and non-streaming benchmark with CER 6.15% and 5.37% respectively. We further compare our RNN-T system with our well trained commercial hybrid system on 30000-hour-industry audio data and get 9% relative improvement without pre-training or external language model.
arXiv:2011.01576v1 fatcat:3xn72mrnz5azte7cqnfh32o2xe

Scene Text Retrieval via Joint Text Detection and Similarity Learning [article]

Hao Wang, Xiang Bai, Mingkun Yang, Shenggao Zhu, Jing Wang, Wenyu Liu
2021 arXiv   pre-print
Scene text retrieval aims to localize and search all text instances from an image gallery, which are the same or similar to a given query text. Such a task is usually realized by matching a query text to the recognized words, outputted by an end-to-end scene text spotter. In this paper, we address this problem by directly learning a cross-modal similarity between a query text and each text instance from natural images. Specifically, we establish an end-to-end trainable network, jointly
more » ... g the procedures of scene text detection and cross-modal similarity learning. In this way, scene text retrieval can be simply performed by ranking the detected text instances with the learned similarity. Experiments on three benchmark datasets demonstrate our method consistently outperforms the state-of-the-art scene text spotting/retrieval approaches. In particular, the proposed framework of joint detection and similarity learning achieves significantly better performance than separated methods. Code is available at: https://github.com/lanfeng4659/STR-TDSL.
arXiv:2104.01552v1 fatcat:zprdsfjlo5dvddzfyzcotzelnu

Symmetry-constrained Rectification Network for Scene Text Recognition [article]

MingKun Yang, Yushuo Guan, Minghui Liao, Xin He, Kaigui Bian, Song Bai, Cong Yao, Xiang Bai
2019 arXiv   pre-print
Yang et al. [52] add character-level supervision to guide the attention learning on the 2D feature maps. Liu et al.  ... 
arXiv:1908.01957v1 fatcat:a6lu6xyy7velnpzi7a2j3d6dmy

Deep learning for scene text detection and recognition

Xiang BAI, Mingkun YANG, Baoguang SHI, Minghui LIAO
2018 Scientia Sinica Informationis  
doi:10.1360/n112018-00003 fatcat:pr722ie3rjbl3dleqfydt6dmpq

AutoSTR: Efficient Backbone Search for Scene Text Recognition [article]

Hui Zhang, Quanming Yao, Mingkun Yang, Yongchao Xu, Xiang Bai
2020 arXiv   pre-print
., Yang, Y.: DARTS: differentiable architecture search. In: International Conference on Learning Representations (2019) 26.  ...  Shi, B., Yang, M., Wang, X., Lyu, P., Yao, C., Bai, X.: Aster: An attentional scene text recognizer with flexible rectification.  ... 
arXiv:2003.06567v2 fatcat:xomhulw5tbesrcc27cen3xlmde

All You Need Is Boundary: Toward Arbitrary-Shaped Text Spotting [article]

Hao Wang, Pu Lu, Hui Zhang, Mingkun Yang, Xiang Bai, Yongchao Xu, Mengchao He, Yongpan Wang, Wenyu Liu
2019 arXiv   pre-print
Recently, end-to-end text spotting that aims to detect and recognize text from cluttered images simultaneously has received particularly growing interest in computer vision. Different from the existing approaches that formulate text detection as bounding box extraction or instance segmentation, we localize a set of points on the boundary of each text instance. With the representation of such boundary points, we establish a simple yet effective scheme for end-to-end text spotting, which can read
more » ... the text of arbitrary shapes. Experiments on three challenging datasets, including ICDAR2015, TotalText and COCO-Text demonstrate that the proposed method consistently surpasses the state-of-the-art in both scene text detection and end-to-end text recognition tasks.
arXiv:1911.09550v1 fatcat:nhvu6c5x5resdphe2k63i47l4a

A Monitoring and Control System for Stormwater Management of Urban Green Infrastructure

Mingkun Xie, Ruijun Wang, Jing Yang, Yuning Cheng
2021 Water  
Urban green infrastructure (UGI) can be used as a sustainable stormwater management approach. UGI can bring numerous ecological benefits to cities, including increased urban resilience, increased availability of water resources, and optimization of habitats. This paper used empirical research methods to describe an Internet of things (IoT)-based UGI monitoring and control system for stormwater management (MCSSWM). Using a Xuzhou-based practical project in China as a case study, we introduce the
more » ... construction process, method, and monitoring results of the system. The results showed that the MCSSWM could be beneficial for UGI ecological performance evaluation and management.
doi:10.3390/w13111438 fatcat:67qpqhbnavhf5mt4o3tp7ondfa

Acetylome Analysis Reveals Diverse Functions of Lysine Acetylation inMycobacterium tuberculosis

Fengying Liu, Mingkun Yang, Xude Wang, Shanshan Yang, Jing Gu, Jie Zhou, Xian-En Zhang, Jiaoyu Deng, Feng Ge
2014 Molecular & Cellular Proteomics  
The lysine acetylation of proteins is a reversible posttranslational modification that plays a critical regulatory role in both eukaryotes and prokaryotes. Mycobacterium tuberculosis is a facultative intracellular pathogen and the causative agent of tuberculosis. Increasing evidence shows that lysine acetylation may play an important role in the pathogenesis of M. tuberculosis. However, only a few acetylated proteins of M. tuberculosis are known, presenting a major obstacle to understanding the
more » ... functional roles of reversible lysine acetylation in this pathogen. We performed a global acetylome analysis of M. tuberculosis H37Ra by combining protein/peptide prefractionation, antibody enrichment, and LC-MS/MS. In total, we identified 226 acetylation sites in 137 proteins of M. tuberculosis H37Ra. The identified acetylated proteins were functionally categorized into an interaction map and shown to be involved in various biological processes. Consistent with previous reports, a large proportion of the acetylation sites were present on proteins involved in glycolysis/gluconeogenesis, the citrate cycle, and fatty acid metabolism. A NAD ؉ -dependent deacetylase (MRA_ 1161) deletion mutant of M. tuberculosis H37Ra was constructed and its characterization showed a different colony morphology, reduced biofilm formation, and increased tolerance of heat stress. Interestingly, lysine acetylation was found, for the first time, to block the immunogenicity of a peptide derived from a known immunogen, HspX, suggesting that lysine acetylation plays a regulatory role in immunogenicity. Our data provide the first global survey of lysine acetylation in M. tuberculosis. The dataset should be an important resource for the functional analysis of lysine acetylation in M. tuberculosis and facilitate the clarification of the entire metabolic networks of this life-threatening pathogen. Molecular & Cellular
doi:10.1074/mcp.m114.041962 pmid:25180227 pmcid:PMC4256489 fatcat:xks4i3lwrfhodl76qkkwa4g5um

Real-Time Pedestrian Tracking Terminal Based on Adaptive Zero Velocity Update

Ran Wei, Hongda Xu, Mingkun Yang, Xinguo Yu, Zhuoling Xiao, Bo Yan
2021 Sensors  
In the field of pedestrian dead reckoning (PDR), the zero velocity update (ZUPT) method with an inertial measurement unit (IMU) is a mature technology to calibrate dead reckoning. However, due to the complex walking modes of different individuals, it is essential and challenging to determine the ZUPT conditions, which has a direct and significant influence on the tracking accuracy. In this research, we adopted an adaptive zero velocity update (AZUPT) method based on convolution neural networks
more » ... o classify the ZUPT conditions. The AZUPT model was robust regardless of the different motion types of various individuals. AZUPT was then implemented on the Zynq-7000 SoC platform to work in real time to validate its computational efficiency and performance superiority. Extensive real-world experiments were conducted by 60 different individuals in three different scenarios. It was demonstrated that the proposed system could work equally well in different environments, making it portable for PDR to be widely performed in various real-world situations.
doi:10.3390/s21113808 pmid:34072810 fatcat:h37uu7elwbdtjcgh22jqi6d4qi

Deep Dive into ZGC: A Modern Garbage Collector in OpenJDK

Albert Mingkun Yang, Tobias Wrigstad
2022 ACM Transactions on Programming Languages and Systems  
In the past, we have ourselves built on ZGC [Yang et al. 2020b,a], and been forced to piece together information from a large number of sources, expressed slightly diferently.  ...  cut-point is picked, subject to the fragmentation limit threshold, so that the trailing multiple pages (too many live objects) are dropped from the EC set; more details on this are covered in HCSGC [Yang  ... 
doi:10.1145/3538532 fatcat:3w67hnpkz5dt7hocp7d5ax3fcy

The Thermal Balance Temperature Field of the Electro-Hydraulic Servo Pump Control System

Guishan Yan, Zhenlin Jin, Mingkun Yang, Bing Yao
2021 Energies  
The electro-hydraulic servo pump control system (EHSPCS) is a high-performance control unit that integrates a permanent magnet synchronous motor (PMSM) and a closed hydraulic system (CHS). The design features of high integration and a high power-weight ratio lead to the poor heat dissipation capacity and high thermal balance temperature of the system. Excessive temperature will seriously affect the system's performance and service life. Therefore, the thermal balance analysis method of the
more » ... S under different loads and different wind speeds was proposed in this paper. Firstly, the PMSM and CHS were taken as research objects to analyze the heating source of the system. The thermal power model was established, and the calculation was performed. The variation rule of the thermal power generated by each component with the load torque under the typical rotating speed was obtained. Secondly, the system's temperature field thermal balance solution model was established. ANSYS and AMESim software were used to calculate the thermal balance points of the PMSM and CHS under different working conditions. The change rule of the thermal balance temperature field of the EHSPCS under different loads was analyzed with the conditions of the natural cooling and forced wind cooling of the PMSM. Finally, an experimental platform was built for experimental research, and the experimental temperature of the system under different loads and different wind speeds was measured. Through comparison and analysis with the simulation results, the correctness and feasibility of the thermal balance theoretical analysis method were verified. The research results will lay the foundation for thermal balance research on the EHSPCS, and have guiding significance for system design, component selection, and load matching.
doi:10.3390/en14051364 fatcat:vr4grvlzijepxohuyibf5bfbe4

Long noncoding RNA HOTAIR interacts with Y-Box Protein-1 (YBX1) to regulate cell proliferation

Siting Li, Qian Xiong, Minghai Chen, Bing Wang, Xue Yang, Mingkun Yang, Qiang Wang, Zongqiang Cui, Feng Ge
2021 Life Science Alliance  
X Yang: investigation and methodology. M Yang: investigation. Q Wang: resources. Z Cui: conceptualization, funding acquisition, and writing-original draft.  ... 
doi:10.26508/lsa.202101139 pmid:34266873 pmcid:PMC8321693 fatcat:hhjohhoxlrbxfk4im3kd2cyova

Genomic perspectives on the emerging SARS-CoV-2 Omicron variant [article]

Wentai Ma, Jing Yang, Haoyi Fu, Chao Su, Caixia Yu, Qihui Wang, Ana Tereza Ribeiro de Vasconcelos, Georgii Bazykin, Yiming Bao, Mingkun Li
2022 bioRxiv   pre-print
A new variant of concern for SARS-CoV-2, Omicron (B.1.1.529), was designated by the World Health Organization on November 26, 2021. This study analyzed the viral genome sequencing data of 108 samples collected from patients infected with Omicron. First, we found that the enrichment efficiency of viral nucleic acids was reduced due to mutations in the region where the primers anneal to. Second, the Omicron variant possesses an excessive number of mutations compared to other variants circulating
more » ... t the same time (62 vs. 45), especially in the Spike gene. Mutations in the Spike gene confer alterations in 32 amino acid residues, which was more than those observed in other SARS-CoV-2 variants. Moreover, a large number of nonsynonymous mutations occur in the codons for the amino acid residues located on the surface of the Spike protein, which could potentially affect the replication, infectivity, and antigenicity of SARS-CoV-2. Third, there are 53 mutations between the Omicron variant and its closest sequences available in public databases. Many of those mutations were rarely observed in the public database and had a low mutation rate. In addition, the linkage disequilibrium between these mutations was low, with a limited number of mutations (6) concurrently observed in the same genome, suggesting that the Omicron variant would be in a different evolutionary branch from the currently prevalent variants. To improve our ability to detect and track the source of new variants rapidly, it is imperative to further strengthen genomic surveillance and data sharing globally in a timely manner.
doi:10.1101/2022.01.05.474231 fatcat:o4ndoovuh5cbna34etkji6vlca
« Previous Showing results 1 — 15 out of 156 results