Jifu Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Jifu Zhang has a total of 112 co-authors. Here are the top co-authors by number of shared publications:
- Lihua Hu
School of Computer Science and Technology, Taiyuan University of Science and Technology, Taiyuan, China
10 shared publications - Sulan Zhang
Taiyuan University of Science and Technology
10 shared publications - Xiao Qin
Auburn University
9 shared publications Xiao Qin
Auburn University
8 shared publications- Yaling Xun
School of Computer Science and Technology, Taiyuan University of Science and Technology, Taiyuan, China
6 shared publications - Jianghui Cai
School of Computer Science and Technology, Taiyuan University of Science and Technology (TYUST), Taiyuan, China
5 shared publications
Latest Publications
- 21 October 2026
- Published by Wiley in Concurrency and Computation: Practice and Experience
- Vol. 38 (21) , e71003
Adaptive Multi‐Axis Partitioning and Data Pruning for Parallel Outlier Detection in Large‐Scale High‐Dimensional Data
With the explosive growth of large‐scale, high‐dimensional data in fields such as finance, IoT, and bioinformatics, parallel outlier detection has become essential for efficiently identifying outliers while enabling scalable processing. However, most existing approaches suffer from high interdependency between computational modules, which leads to inefficient resource utilization and escalated I/O overhead. Additionally, their load balancing strategies fail to account for the distribution characteristics of data attributes, hindering the detection of outliers in sparsely distributed attribute subsets. In this paper, we propose a novel parallel outlier detection framework, named HODAP‐DP, which integrates adaptive multi‐axis recursive partitioning and local I/O performance optimization strategies. First, an adaptive multi‐axis recursive partitioning strategy is designed by exploiting the statistical properties of data attribute distributions, so that load balancing across partitions is ensured and outliers within sparse subsets can be effectively detected. Second, we introduce a local I/O performance optimization strategy leveraging data density locality and data pruning in memory pipelining to reduce cross‐node communication and I/O overhead caused by interdependent computation, thereby enhancing the utilization of computational and memory resources. Third, a novel parallel outlier detection algorithm, which employs Chebyshev Coulomb resultant force as the similarity measures, is proposed by integrating recursive multi‐axis partitioning and local I/O performance optimization. In the end, experimental results on UCI, TIGER, and TB‐scale datasets validate the algorithm's superior performance, scalability, and extensibility.
- 21 October 2026
- Published by Wiley in Concurrency and Computation: Practice and Experience
- Vol. 38 (21) , e71003
Adaptive Multi‐Axis Partitioning and Data Pruning for Parallel Outlier Detection in Large‐Scale High‐Dimensional Data
With the explosive growth of large‐scale, high‐dimensional data in fields such as finance, IoT, and bioinformatics, parallel outlier detection has become essential for efficiently identifying outliers while enabling scalable processing. However, most existing approaches suffer from high interdependency between computational modules, which leads to inefficient resource utilization and escalated I/O overhead. Additionally, their load balancing strategies fail to account for the distribution characteristics of data attributes, hindering the detection of outliers in sparsely distributed attribute subsets. In this paper, we propose a novel parallel outlier detection framework, named HODAP‐DP, which integrates adaptive multi‐axis recursive partitioning and local I/O performance optimization strategies. First, an adaptive multi‐axis recursive partitioning strategy is designed by exploiting the statistical properties of data attribute distributions, so that load balancing across partitions is ensured and outliers within sparse subsets can be effectively detected. Second, we introduce a local I/O performance optimization strategy leveraging data density locality and data pruning in memory pipelining to reduce cross‐node communication and I/O overhead caused by interdependent computation, thereby enhancing the utilization of computational and memory resources. Third, a novel parallel outlier detection algorithm, which employs Chebyshev Coulomb resultant force as the similarity measures, is proposed by integrating recursive multi‐axis partitioning and local I/O performance optimization. In the end, experimental results on UCI, TIGER, and TB‐scale datasets validate the algorithm's superior performance, scalability, and extensibility.
- 26 February 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 339
Adaptive deep multi-view clustering via quality-aware representation learning
- 26 February 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 339
Adaptive deep multi-view clustering via quality-aware representation learning
- 25 February 2026
- Published by Elsevier in Future Generation Computer Systems
- Vol. 181
Dynamic UAV task offloading combining deep reinforcement learning and two-stage stochastic optimization
- 25 February 2026
- Published by Elsevier in Future Generation Computer Systems
- Vol. 181
Dynamic UAV task offloading combining deep reinforcement learning and two-stage stochastic optimization
- 28 January 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 337
Multimodal interpretable image recognition network via language-guided global-local collaboratively alignment
- 28 January 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 337
Multimodal interpretable image recognition network via language-guided global-local collaboratively alignment
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 28, 12255-12266
Hyperbolic Concept Bottleneck Model for Interpretable Fine-Grained Image Classification
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 28, 12255-12266
Hyperbolic Concept Bottleneck Model for Interpretable Fine-Grained Image Classification
- 25 July 2025
- Published by Springer Nature
- p. 75-86
A Relevant Subspace-Based Contextual Outlier Detection Using Chebyshev Coulomb Resultant Force
- 25 July 2025
- Published by Springer Nature
- p. 75-86
A Relevant Subspace-Based Contextual Outlier Detection Using Chebyshev Coulomb Resultant Force
- 25 July 2025
- Published by Springer Nature
- p. 63-74
Similarity Attribute-Based Categorical Attribute Grouping for Outlier Detecting
- 25 July 2025
- Published by Springer Nature
- p. 63-74
Similarity Attribute-Based Categorical Attribute Grouping for Outlier Detecting
- Open Access
- 15 April 2025
- Published by Springer Nature in Complex & Intelligent Systems
- Vol. 11 (6) , 1-17
Attribute grouping-based categorical outlier detection using causal coupling weight
- 15 April 2025
- Published by Springer Nature in Complex & Intelligent Systems
- Vol. 11 (6) , 1-17
Open AccessAttribute grouping-based categorical outlier detection using causal coupling weight
- Open Access
- 8 April 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 19 (3) , 1-25
Similarity Metrics: Chebyshev Coulomb Force and Resultant Force for High-Dimensional Data
The similarity metric has garnered widespread attention thanks to its potential applications in the fields of data mining, machine learning, and so on. Due to the interference of “distance concentration” caused by “Curse of dimensionality,” however, existing similarity metrics are inadequate in high-dimensional data analysis. In this study, we propose two innovative similarity metrics—Chebyshev Coulomb force and Chebyshev Coulomb resultant force—anchored on Chebyshev p-norms. In the initial phase, we eliminate dependency relationships among attributes by applying a metric matrix—and the theoretical analysis reveals that the Chebyshev p-norms is capable of mitigating the effect of “distance concentration” among high-dimensional data objects. Next, we devise two similarity metrics—Chebyshev Coulomb force and Chebyshev Coulomb resultant force—by adopting the metric matrix and Chebyshev p-norms. Chebyshev Coulomb force and Chebyshev Coulomb resultant force, being effective in characterizing the similarity among data objects, quantify the deviation of data objects from their respective dataset centers. Additionally, the two metrics alleviate the interference of “distance concentration.” Importantly, the discrepancy of data objects in attribute dimensions is captured by Chebyshev Coulomb force vector, rendering the similarity metric interpretable. By utilizing the UCI dataset, the experimental validation demonstrates the superiority of our similarity metrics, confirming their efficacy in mitigating the interference of “distance concentration.” Compared with the existing similarity metric approaches, the AUC index of outlier detection shows an average improvement of 8.18%—and the ARI, NMI, and F_score indices of clustering are revamped by averages 6.56%, 6.87%, and 6.01%, respectively.
- 8 April 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 19 (3) , 1-25
Open AccessSimilarity Metrics: Chebyshev Coulomb Force and Resultant Force for High-Dimensional Data
The similarity metric has garnered widespread attention thanks to its potential applications in the fields of data mining, machine learning, and so on. Due to the interference of “distance concentration” caused by “Curse of dimensionality,” however, existing similarity metrics are inadequate in high-dimensional data analysis. In this study, we propose two innovative similarity metrics—Chebyshev Coulomb force and Chebyshev Coulomb resultant force—anchored on Chebyshev p-norms. In the initial phase, we eliminate dependency relationships among attributes by applying a metric matrix—and the theoretical analysis reveals that the Chebyshev p-norms is capable of mitigating the effect of “distance concentration” among high-dimensional data objects. Next, we devise two similarity metrics—Chebyshev Coulomb force and Chebyshev Coulomb resultant force—by adopting the metric matrix and Chebyshev p-norms. Chebyshev Coulomb force and Chebyshev Coulomb resultant force, being effective in characterizing the similarity among data objects, quantify the deviation of data objects from their respective dataset centers. Additionally, the two metrics alleviate the interference of “distance concentration.” Importantly, the discrepancy of data objects in attribute dimensions is captured by Chebyshev Coulomb force vector, rendering the similarity metric interpretable. By utilizing the UCI dataset, the experimental validation demonstrates the superiority of our similarity metrics, confirming their efficacy in mitigating the interference of “distance concentration.” Compared with the existing similarity metric approaches, the AUC index of outlier detection shows an average improvement of 8.18%—and the ARI, NMI, and F_score indices of clustering are revamped by averages 6.56%, 6.87%, and 6.01%, respectively.
- 27 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
LrpcSIDE: Unsupervised Depth Estimation Powered by Left–Right Point Cloud Consistency
As one of the most studied topics in 3D reconstruction, single image-based depth estimation (SIDE) has achieved good performance in some scenes. However, SIDE methods suffer from distortions in the in-planar regions of reconstructed 3D point clouds due to the lack of consideration for directly using geometric information to impose constraints in 3D space, especially when repeated textural and complex structural patterns are abundant. In the work, we proposed an unsupervised SIDE method (LrpcSIDE) by using the left-right point cloud consistency constraint. First, we guided the dense-encoder features toward the desired depth prediction across multiple scales of the encoder-decoder structure and strengthened the perception of the estimation network in both detail and global perspective. Second, we designed a novel training loss function based on the left-right point cloud consistency constraint which enforced a consistency between the disparities produced relative to both the left and right point cloud, thus the distortion in the point cloud plane is curbed. Third, an unsupervised network was devised by using binocular image pairs, eliminating any potential adverse effects caused by unknown scale or pose estimation errors. Forth, the disparity map was transformed into a depth map by leveraging the relationship between the camera baseline distance and focal length, achieving end-to-end single-image depth estimation effectively. In the end, extensive experimental results on the KITTI and ancient architecture datasets verify that our method could accurately recover the 3D structure of scenes with high depth prediction and minimal depth estimation errors, and exhibit good depth estimation performance for application scenes with complex structures and repetitive textures. Source code is available at https://github.com/yaozz1110/LrpcSIDE.
- 27 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-11
LrpcSIDE: Unsupervised Depth Estimation Powered by Left–Right Point Cloud Consistency
As one of the most studied topics in 3D reconstruction, single image-based depth estimation (SIDE) has achieved good performance in some scenes. However, SIDE methods suffer from distortions in the in-planar regions of reconstructed 3D point clouds due to the lack of consideration for directly using geometric information to impose constraints in 3D space, especially when repeated textural and complex structural patterns are abundant. In the work, we proposed an unsupervised SIDE method (LrpcSIDE) by using the left-right point cloud consistency constraint. First, we guided the dense-encoder features toward the desired depth prediction across multiple scales of the encoder-decoder structure and strengthened the perception of the estimation network in both detail and global perspective. Second, we designed a novel training loss function based on the left-right point cloud consistency constraint which enforced a consistency between the disparities produced relative to both the left and right point cloud, thus the distortion in the point cloud plane is curbed. Third, an unsupervised network was devised by using binocular image pairs, eliminating any potential adverse effects caused by unknown scale or pose estimation errors. Forth, the disparity map was transformed into a depth map by leveraging the relationship between the camera baseline distance and focal length, achieving end-to-end single-image depth estimation effectively. In the end, extensive experimental results on the KITTI and ancient architecture datasets verify that our method could accurately recover the 3D structure of scenes with high depth prediction and minimal depth estimation errors, and exhibit good depth estimation performance for application scenes with complex structures and repetitive textures. Source code is available at https://github.com/yaozz1110/LrpcSIDE.