Search

Your search keyword '"Zhang, Hongjing"' showing total 204 results
Start Over Author "Zhang, Hongjing"
204 results on '"Zhang, Hongjing"'

8. Deep Fair Discriminative Clustering

Author

Zhang, Hongjing and Davidson, Ian

Subjects
Computer Science - Machine Learning, Computer Science - Computers and Society, Statistics - Machine Learning
Abstract

Deep clustering has the potential to learn a strong representation and hence better clustering performance compared to traditional clustering methods such as $k$-means and spectral clustering. However, this strong representation learning ability may make the clustering unfair by discovering surrogates for protected information which we empirically show in our experiments. In this work, we study a general notion of group-level fairness for both binary and multi-state protected status variables (PSVs). We begin by formulating the group-level fairness problem as an integer linear programming formulation whose totally unimodular constraint matrix means it can be efficiently solved via linear programming. We then show how to inject this solver into a discriminative deep clustering backbone and hence propose a refinement learning algorithm to combine the clustering goal with the fairness objective to learn fair clusters adaptively. Experimental results on real-world datasets demonstrate that our model consistently outperforms state-of-the-art fair clustering algorithms. Our framework shows promising results for novel clustering tasks including flexible fairness constraints, multi-state PSVs and predictive clustering.

Published
2021

9. Deep Descriptive Clustering

Author

Zhang, Hongjing and Davidson, Ian

Subjects
Computer Science - Machine Learning, Computer Science - Artificial Intelligence
Abstract

Recent work on explainable clustering allows describing clusters when the features are interpretable. However, much modern machine learning focuses on complex data such as images, text, and graphs where deep learning is used but the raw features of data are not interpretable. This paper explores a novel setting for performing clustering on complex data while simultaneously generating explanations using interpretable tags. We propose deep descriptive clustering that performs sub-symbolic representation learning on complex data while generating explanations based on symbolic data. We form good clusters by maximizing the mutual information between empirical distribution on the inputs and the induced clustering labels for clustering objectives. We generate explanations by solving an integer linear programming that generates concise and orthogonal descriptions for each cluster. Finally, we allow the explanation to inform better clustering by proposing a novel pairwise loss with self-generated constraints to maximize the clustering and explanation module's consistency. Experimental results on public data demonstrate that our model outperforms competitive baselines in clustering performance while offering high-quality cluster-level explanations., Comment: Paper accepted at IJCAI 2021

Published
2021

10. A Framework for Deep Constrained Clustering

Author

Zhang, Hongjing, Zhan, Tianyang, Basu, Sugato, and Davidson, Ian

Subjects
Computer Science - Machine Learning
Abstract

The area of constrained clustering has been extensively explored by researchers and used by practitioners. Constrained clustering formulations exist for popular algorithms such as k-means, mixture models, and spectral clustering but have several limitations. A fundamental strength of deep learning is its flexibility, and here we explore a deep learning framework for constrained clustering and in particular explore how it can extend the field of constrained clustering. We show that our framework can not only handle standard together/apart constraints (without the well documented negative effects reported earlier) generated from labeled side information but more complex constraints generated from new types of side information such as continuous values and high-level domain knowledge. Furthermore, we propose an efficient training paradigm that is generally applicable to these four types of constraints. We validate the effectiveness of our approach by empirical results on both image and text datasets. We also study the robustness of our framework when learning with noisy constraints and show how different components of our framework contribute to the final performance. Our source code is available at $\href{https://github.com/blueocean92/deep_constrained_clustering}{\text{URL}}$., Comment: Data Mining and Knowledge Discovery, 2021. arXiv admin note: substantial text overlap with arXiv:1901.10061

Published
2021

11. Towards Fair Deep Anomaly Detection

Author

Zhang, Hongjing and Davidson, Ian

Subjects
Computer Science - Machine Learning, Computer Science - Artificial Intelligence, Statistics - Machine Learning
Abstract

Anomaly detection aims to find instances that are considered unusual and is a fundamental problem of data science. Recently, deep anomaly detection methods were shown to achieve superior results particularly in complex data such as images. Our work focuses on deep one-class classification for anomaly detection which learns a mapping only from the normal samples. However, the non-linear transformation performed by deep learning can potentially find patterns associated with social bias. The challenge with adding fairness to deep anomaly detection is to ensure both making fair and correct anomaly predictions simultaneously. In this paper, we propose a new architecture for the fair anomaly detection approach (Deep Fair SVDD) and train it using an adversarial network to de-correlate the relationships between the sensitive attributes and the learned representations. This differs from how fairness is typically added namely as a regularizer or a constraint. Further, we propose two effective fairness measures and empirically demonstrate that existing deep anomaly detection methods are unfair. We show that our proposed approach can remove the unfairness largely with minimal loss on the anomaly detection performance. Lastly, we conduct an in-depth analysis to show the strength and limitations of our proposed model, including parameter analysis, feature visualization, and run-time analysis., Comment: Accepted for publication at the ACM Conference on Fairness, Accountability, and Transparency 2021 (ACM FAccT'21)

Published
2020

13. A Graph-Based Approach for Active Learning in Regression

Author

Zhang, Hongjing, Ravi, S. S., and Davidson, Ian

Subjects
Computer Science - Machine Learning, Statistics - Machine Learning
Abstract

Active learning aims to reduce labeling efforts by selectively asking humans to annotate the most important data points from an unlabeled pool and is an example of human-machine interaction. Though active learning has been extensively researched for classification and ranking problems, it is relatively understudied for regression problems. Most existing active learning for regression methods use the regression function learned at each active learning iteration to select the next informative point to query. This introduces several challenges such as handling noisy labels, parameter uncertainty and overcoming initially biased training data. Instead, we propose a feature-focused approach that formulates both sequential and batch-mode active regression as a novel bipartite graph optimization problem. We conduct experiments on both noise-free and noisy settings. Our experimental results on benchmark data sets demonstrate the effectiveness of our proposed approach., Comment: SDM 2020 camera-ready. 9 pages, 4 figures, links to supplementary material available at https://sdm2020.s3-us-west-1.amazonaws.com/supplementary.pdf

Published
2020

15. A Framework for Deep Constrained Clustering -- Algorithms and Advances

Author

Zhang, Hongjing, Basu, Sugato, and Davidson, Ian

Subjects
Computer Science - Machine Learning, Statistics - Machine Learning
Abstract

The area of constrained clustering has been extensively explored by researchers and used by practitioners. Constrained clustering formulations exist for popular algorithms such as k-means, mixture models, and spectral clustering but have several limitations. A fundamental strength of deep learning is its flexibility, and here we explore a deep learning framework for constrained clustering and in particular explore how it can extend the field of constrained clustering. We show that our framework can not only handle standard together/apart constraints (without the well documented negative effects reported earlier) generated from labeled side information but more complex constraints generated from new types of side information such as continuous values and high-level domain knowledge., Comment: Updated for ECML/PKDD 2019

Published
2019

19. Secret to Decision: Analysis of the Influencing Factors of the Company’s Top Decisions

Author

Zhang Hongjing

Subjects
Social Sciences
Abstract

The purpose of this paper is to discuss the importance of decision-making by the top management of the company. Taking Apple as an example, this paper deeply analyzes the process of how to make the right decision in the face of crisis and its positive impact on the company, which further proves the positive impact of successful decision-making on the enterprise. The connotation of the concept of comprehensive high-level decision-making is expounded from many aspects. These elements are interrelated and influence each other, which together constitute a comprehensive system of high-level decision-making. Firstly, the paper expounds the importance of high-level decision-making in modern enterprise management, and through the case analysis of Apple’s high-level decision-making process, it proves that successful decision-making plays a key role in the company’s transformation, innovation and sustainable development. At the same time, the paper comprehensively analyzes the factors that affect the high-level decision-making and the methods and steps for the high-level leaders to make decisions. Finally, the paper summarizes the importance and practical significance of how the company’s top management makes decisions. Senior leaders need to have clear decision-making ideas, scientific decision-making methods and good decision-making environment in order to formulate reasonable development strategies and cope with market changes. Through effective decision-making, enterprises can enhance their competitiveness, achieve sustainable development and create long-term value. Therefore, senior leaders should attach importance to the decision-making process and constantly improve their decision-making ability and level.

Published
2024
Full Text
View/download PDF

20. A Framework for Deep Constrained Clustering - Algorithms and Advances

Author

Zhang, Hongjing, Basu, Sugato, Davidson, Ian, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Woeginger, Gerhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Brefeld, Ulf, editor, Fromont, Elisa, editor, Hotho, Andreas, editor, Knobbe, Arno, editor, Maathuis, Marloes, editor, and Robardet, Céline, editor

Published
2020
Full Text
View/download PDF

22. Identification of Biomarkers Associated with Oxidative Stress and Immune Cells in Acute Pancreatitis

Author

Xu,Yuan, Song,Jie, Gao,Jie, Zhang,Hongjing, Xu,Yuan, Song,Jie, Gao,Jie, and Zhang,Hongjing

Abstract

Yuan Xu,1 Jie Song,1 Jie Gao,1 Hongjing Zhang2 1Department of Gastroenterology, Dongying People’s Hospital(Dongying Hospital of Shandong Provincial Hospital Group), Dongying, Shandong, People’s Republic of China; 2Community Health Service Center in Hekou District, Dongying, Shandong, People’s Republic of ChinaCorrespondence: Yuan Xu, Department of Gastroenterology, Dongying People’s Hospital (Dongying Hospital of Shandong Provincial Hospital Group), No. 317 Nanyi Road, Dongying, Shandong, 257091, People’s Republic of China, Email jry0919@sina.comPurpose: Oxidative stress promotes disease progression by stimulating the humoral and cellular immune responses. However, the molecular mechanisms underlying oxidative stress and immune responses in acute pancreatitis (AP) have not been extensively studied.Patients and Methods: We analyzed the GSE194331 dataset and oxidative stress-related genes (OSRGs). We identified differentially expressed immune cell-associated OSRGs (DE-ICA-OSRGs) by overlapping key module genes from weighted gene co-expression network analysis, OSRGs, and DEGs between AP and normal samples. Functional enrichment analysis was performed to investigate the functions of DE-ICA-OSRGs. We then filtered diagnostic genes using receiver operating characteristic curves and investigated their molecular mechanisms using single-gene set enrichment analysis (GSEA). We also explored the correlation between diagnostic genes and differential immune cells. Finally, we constructed a transcription factor-microRNA-messenger RNA (TF-miRNA-mRNA) network of biomarkers.Results: In this study, three DE-ICA-OSRGs (ARG1, NME8 and VNN1) were filtered by overlapping key module genes, OSRGs and DEGs. Functional enrichment results revealed that DE-ICA-OSRGs were involved in the cellular response to reactive oxygen species and arginine biosynthesis. Latterly, a total of two diagnostic genes (ARG1 and VNN1) were derived and their expression was higher in the AP group than in

Published
2024

23. "Point by point" source: The Chinese pine plantations in North China by evidence from mtDNA.

Author

Zhou, Biao, Zhang, Zijie, Zhang, Hongjing, Li, Yupeng, Ma, Yanguang, Zhang, Shubin, Niu, Shihui, and Li, Yue

Subjects
PLANTATIONS, POPULATION of China, PINE, GENETIC variation, MITOCHONDRIAL DNA, GERMPLASM
Abstract

The geographical variation and domestication of tree species are an important part of the theory of forest introduction, and the tracing of the germplasm is the theoretical basis for the establishment of high‐quality plantations. Chinese pine (Pinus tabuliformis Carr.) is an important native timber tree species widely distributed in northern China, but it is unclear exactly where germplasm of the main Chinese pine plantation populations originated. Here, using two mtDNA markers, we analyzed 796 individuals representing 35 populations (matR marker), and 873 individuals representing 38 populations (nad5‐1 marker) of the major natural and artificial populations in northern China, respectively (Shanxi, Hebei and Liaoning provinces). The results confirmed that the core position of natural SX* populations ("*" means natural population) in the Chinese pine populations of northern China, the genetic diversity of HB and LN plantations was higher than that of natural SX* populations, and there was a large difference in genetic background within the groups of SX* and LN, HB showed the opposite. More importantly, we completed the "point by point" tracing of the HB and LN plantings. The results indicated that almost all HB populations originated from SX* (GDS*, ZTS*, GCS*, and THS*), which resulted in homogeneity of the genetic background of HB populations. Most of germplasm of the LN plantations originated from LN* (ZJS* and WF*), and the other part originated from GDS* (SX*), resulting in the large differences in the genetic background within the LN group. Our results provided a reliable theoretical basis for the scientific allocation, management, and utilization of Chinese pine populations in northern China, and for promoting the high‐quality establishment of Chinese pine plantations. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

24. Fabrication of asymmetric wettability silk fabrics with plasma-induced grafting polymerization of octamethylcyclotetrasiloxane.

Author

Chen, Xu, Xing, Yajie, Mu, Qingyuan, Zhang, Hongjing, Zhou, Ke, Yu, Xiaotian, Hao, Haitao, and Li, Yongqiang

Abstract

Most superhydrophobic surfaces contain fluorine, which is very harmful to the environment. So, it is necessary to change the situation. This paper explains that a simple and fluorine-free plasma technology for superhydrophobic modification of silk fabrics by using one-sided protection method and low-temperature plasma induced grafting polymerization of octamethyl-cyclotetrasiloxane (D4) to form the superhydrophobic films. Treated silk and pristine silk fabrics were measured using field emission scanning electron microscopy(FESEM), Fourier-transform infrared spectroscopy(FTIR), X-ray photoelectron spectroscopy(XPS), and water contact angle measurement. The test results showed that the polymer films could be successful coated on the silk fibers surface via the low-temperature plasma. In addition, it has good durability, thermal stability, and a water contact angle of 152°. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

25. Identification of Biomarkers Associated with Oxidative Stress and Immune Cells in Acute Pancreatitis.

Author

Xu, Yuan, Song, Jie, Gao, Jie, and Zhang, Hongjing

Subjects
RECEIVER operating characteristic curves, GENE regulatory networks, KILLER cells, IMMUNOLOGIC memory, REACTIVE oxygen species, GENE ontology
Abstract

Purpose: Oxidative stress promotes disease progression by stimulating the humoral and cellular immune responses. However, the molecular mechanisms underlying oxidative stress and immune responses in acute pancreatitis (AP) have not been extensively studied. Patients and Methods: We analyzed the GSE194331 dataset and oxidative stress-related genes (OSRGs). We identified differentially expressed immune cell-associated OSRGs (DE-ICA-OSRGs) by overlapping key module genes from weighted gene co-expression network analysis, OSRGs, and DEGs between AP and normal samples. Functional enrichment analysis was performed to investigate the functions of DE-ICA-OSRGs. We then filtered diagnostic genes using receiver operating characteristic curves and investigated their molecular mechanisms using single-gene set enrichment analysis (GSEA). We also explored the correlation between diagnostic genes and differential immune cells. Finally, we constructed a transcription factor-microRNA-messenger RNA (TF-miRNA-mRNA) network of biomarkers. Results: In this study, three DE-ICA-OSRGs (ARG1, NME8 and VNN1) were filtered by overlapping key module genes, OSRGs and DEGs. Functional enrichment results revealed that DE-ICA-OSRGs were involved in the cellular response to reactive oxygen species and arginine biosynthesis. Latterly, a total of two diagnostic genes (ARG1 and VNN1) were derived and their expression was higher in the AP group than in the normal group. The single-gene GSEA enrichment results revealed that diagnostic genes were mainly enriched in macroautophagy and Toll-like receptor signaling pathways. Correlation analysis revealed that CD8 T cells, resting memory T CD4 cells, and resting NK cells were negatively correlated with ARG1, and neutrophils were positively correlated with ARG1, which was consistent with that of VNN1. The TF-miRNA-mRNA regulatory network included 11 miRNAs, 2 mRNAs, 10 transcription factors (TFs), and 26 pairs of regulatory relationships, like NFKB1-has-miR-2909-VNN1. Conclusion: In this study, two immune cell oxidative stress-related AP diagnostic genes (ARG1 and VNN1) were screened to offer a new reference for the diagnosis of patients with AP. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

28. Modernizing Deep Unsupervised Learning with Human Experience

Author

Zhang, Hongjing

Subjects
Computer science, Anomaly Detection, Constrained Clustering, Explainable AI, Fair Machine Learning, Representation Learning
Abstract

Deep unsupervised learning has emerged as a promising alternative to supervised approaches. However, supervised learning needs a tremendous amount of information in the form of annotations on specific pre-defined tasks. In contrast, human learning requires much fewer annotations and is flexible. Recent research efforts have been motivated to explore different deep unsupervised learning algorithms to leverage the massive unlabeled data for various applications that move beyond the supervised learning setting. While recent deep unsupervised learning works have shown their success in representation learning, clustering, and anomaly detection, many challenges remain unsolved. For example, how to improve the quality of learned representations used for downstream applications (the quality of learned representations challenge)? How to interpret and understand the deep unsupervised learning model predictions (the explainability challenge)? Is there any risk of bias for deep unsupervised learning applications (the bias and fairness challenge)?To gain insights into the aforementioned challenges, we propose a broad range of novel techniques to address them. Each injects human-level knowledge into deep unsupervised learning. To be specific, this dissertation presents five approaches. The first two address the quality of representation challenge, the third the explainability challenge, and the last two the bias and fairness challenges. Our first formulation introduces a deep constrained clustering framework that enhances clustering performance via various constraints. Our second formulation is a self-supervised representation learning framework that automatically discovers and differentiates different categories. The third formulation simultaneously performs representation learning for clustering and describing the generated clusters with semantic tags associated with the clustered instances. Our fourth formulation proposes a novel deep fair anomaly detection architecture that uses adversarial learning to inject human fairness rules. Finally, our fifth formulation enforces disparate impact rules into deep clustering models via minimal modification learning. These methods are unified in modernizing deep unsupervised learning with different types of human guidance.

Published
2022

30. Preparation of Porous TiO2/PVDF/CA Fibers for Photocatalytic Degradation of Dyes by Centrifugal Spinning.

Author

Li, Yongqiang, Zhang, Hongjing, Li, Ya'nan, Yu, Xiaotian, Zhou, Ke, Chen, Xu, Hao, Haitao, and Huang, Yi

Abstract

A facile method for the preparation of porous micro/nano fibers was demonstrated. The TiO2/PVDF/CA-composite micro/nano fibers were prepared by centrifugal spinning and then CA was removed by a selective dissolution: reaction with the solution of Acetic Acid. The obtained TiO2/PVDF/CA micro/nano fibers showed highly porous surfaces after the extraction of CA. The structure and properties of PVDF/CA micro/nano fibers were characterized by fourier transform infrared (FTIR), and thermogravimetry (TG), and scanning electron microscope (SEM). The results indicated that CA could be introduced into the PVDF solution and PVDF/CA-composite fibers could be fabricated via centrifugal spinning successfully. CA is dissolved and pores are formed on the fibers. The morphology image of the fibers was detected by scanning electron microscope (SEM) and showed that many pores aligned the nanofibers. The photocatalytic degradation properties of the treated composite fibers were studied. The results showed that the composite fibers prepared by centrifugal spinning had good degradation effects for different concentrations and dyes, and the composite fibers had good reusability. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

35. Mental health problems and coping styles of urban and rural high school students in China

Author

Zhang, Hongjing, Chang, Kyle, Zhang, Fan, Greenberger, Ellen, and Chen, Chuansheng

Subjects
Basic Behavioral and Social Science, Clinical Research, Pediatric, Pediatric Research Initiative, Mental Health, Behavioral and Social Science, Rural Health, Mental health, Good Health and Well Being, mental health problems, coping styles, urban areas, rural areas, high school students, China, Human, Male, Female, Adolescence, Coping Style Scale of Middle School Students, Mental Health Inventory for Adolescents, Symptom Checklist-90–Revised, Empirical Study, Quantitative Study, Coping Behavior, Urban Environments, High School Students, Rural Environments, article, 2840:Psychosocial & Personality Development, Psychology, Developmental & Child Psychology
Abstract

Few studies have compared urban and rural adolescents' mental health problems, especially in developing countries. The purpose of this study was to investigate the mental health problems and coping styles of adolescents in urban and rural areas in China. A total of 927 urban and rural high school students in Shandong Province of China were recruited for the study. The results of the study are as follows: (a) students in rural low-socioeconomic status (SES) areas, especially females, had more mental health problems than did those in rural high-SES and urban areas; (b) rural low-SES students were more likely to cope by venting and fantasizing than did the other two groups; (c) regardless of residence and SES, mental health problems were inversely related to the use of problem solving as a coping strategy, but positively related to fantasizing; and (d) urban-rural differences in mental health problems were not entirely accounted for by group differences in coping strategies. © 2011 Wiley Periodicals, Inc.

Published
2011

Catalog

Books, media, physical & digital resources
See catalog results