Kun Zhang

Kun Zhang

Auto-generated profile, based on publicly available information

Kun Zhang

Queen’s University Belfast

Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
26 September 2022 in Medical Image Analysis
Verified
Journal Article
Open access
AMAndrew Moyes
Andrew Moyes
RGRichard Gault
Richard Gault
KZKun Zhang
Kun Zhang
JMJi Ming
Ji Ming
DCDanny Crookes
Danny Crookes
JWJing Wang
Jing Wang
Domain shift is a problem commonly encountered when developing automated histopathology pipelines. The performance of machine learning models such as convolutional neural networks within automated histopathology pipelines is often diminished when applying them to novel data domains due to factors arising from differing staining and scanning protocols. The Dual-Channel Auto-Encoder (DCAE) model was previously shown to produce feature representations that are less sensitive to appearance variation introduced by different digital slide scanners. In this work, the Multi-Channel Auto-Encoder (MCAE) model is presented as an extension to DCAE which learns from more than two domains of data. Experimental results show that the MCAE model produces feature representations that are less sensitive to inter-domain variations than the comparative StaNoSA method when tested on a novel synthetic dataset. This was apparent when applying the MCAE, DCAE, and StaNoSA models to three different classification tasks from unseen domains. The results of this experiment show the MCAE model out performs the other models. These results show that the MCAE model is able to generalise better to novel data, including data from unseen domains, than existing approaches by actively learning normalised feature representations.

1 March 2020 in Corrosion
Verified
Journal Article
Closed access
HLHongjuan Li
Hongjuan Li
FNFei Ning
Fei Ning
HDHaiying Dong
Haiying Dong
KZKun Zhang
Kun Zhang
ZLZhanpeng Lu
Zhanpeng Lu
SLSichun Ling
Sichun Ling
The effects of magnetic field on anodic dissolution and passivation of iron in a sulfuric acid solution with chlorides are investigated by electrochemical measurements and surface observations. In the anodic potentiodynamic polarization curve, the potential for the drastic current drop is not significantly affected by the potential sweep rate under 0 T, which moves in the negative direction with increasing potential sweep rate under 0.4 T magnetic field that is parallel to the working electrode surface. The uneven surface produced during the potentiodynamic polarization hinders the transition from active dissolution to passivation. The area fraction of the locally accelerated dissolution increases with prolonging polarization time at high potentials where the surface film precipitation-dissolution process is the rate-determining step for metal dissolution. Pretreatment under potentiostatic polarization at 0.4 T magnetic field produces an uneven surface that would result in unrecoverable electrochemical states after switching from 0.4 T to 0 T, depending on the applied potential. The positive-feedback mechanism for the magnetic field effect and the surface morphological effect is proposed. The results demonstrate the direct magnetohydrodynamic effect and its resultant uneven surface on the anodic behavior of iron.

1 May 2013
Verified
Conference Paper
Closed access
ZKZhang Kun
Zhang Kun
FMFei Minrui
Fei Minrui
WJWu Jianguo
Wu Jianguo
ZPZhang Peijian
Zhang Peijian
According to the analysis of the structure and technology, and then obtaining the energy consumption calculation formula, it comes to conclusion that the hoop standard granulator is a multivariate system with large time delay, which is also uncertain and nonlinear. Aiming at the problem that the hoop standard granulator cannot be controlled well by traditional methods, a technique was proposed, based on controlling the hoop standard granulator automatically by the expert system and SVM method. For good prediction performance, SVM with genetic algorithm is chosen to build the prediction model of output temperature and the main motor current, which is set as soft sense in the expert system. The expert system could be intelligent to solve the decoupling problem, by controlling feature identification, information processing machine, knowledge base, control rule and inference machine. In order to verify the design, based on reliable real-site data some experiments are conducted under the situation of disturbance. Finally the effect of expert control system and SVM is validated by simulation.

1 April 2014 in Key Engineering Materials
Verified
Journal Article
Closed access
PDF file
KZKun Zhang
Kun Zhang
MFMin Rui Fei
Min Rui Fei
Features analysis is an important task which can significantly affect the performance of automatic bacteria colony picking. This paper presents a novel approach for adaptive colony segmentation by classifying the detected peaks of intensity histograms of images. The relevance and importance of these features can be determined in an improved support vector machine classifier using unascertained least square estimation. Experimental results show that the proposed unascertained support vector machine (USVM) has better recognition accuracy than the other state of the art techniques, and its training process takes less time than most of the traditional approaches presented in this paper.

1 January 2014
Verified
Book Chapter
Closed access
PDF file
KZKun Zhang
Kun Zhang
MFMinrui Fei
Minrui Fei
JWJianguo Wu
Jianguo Wu
PZPeijian Zhang
Peijian Zhang

1 January 2014 in The Scientific World Journal
Verified
Journal Article
Open access
PDF file
KZKun Zhang
Kun Zhang
MFMinrui Fei
Minrui Fei
XLXin Li
Xin Li
HZHuiyu Zhou
Huiyu Zhou
Features analysis is an important task which can significantly affect the performance of automatic bacteria colony picking. Unstructured environments also affect the automatic colony screening. This paper presents a novel approach for adaptive colony segmentation in unstructured environments by treating the detected peaks of intensity histograms as a morphological feature of images. In order to avoid disturbing peaks, an entropy based mean shift filter is introduced to smooth images as a preprocessing step. The relevance and importance of these features can be determined in an improved support vector machine classifier using unascertained least square estimation. Experimental results show that the proposed unascertained least square support vector machine (ULSSVM) has better recognition accuracy than the other state-of-the-art techniques, and its training process takes less time than most of the traditional approaches presented in this paper.