arXiv subjects

Zhong-Yuan Zhang

Publications and source records attributed to Zhong-Yuan Zhang.

At least 19 recordsLinked to original sources

arXiv2026

CellRFT: Reinforcement Fine-Tuning for Single-Cell Perturbation Modeling

Predicting cellular responses to perturbations supports the study of gene function, disease mechanisms, and therapeutic strategies. Despite advances in single-cell perturbation modeling, existing models typically optimize surrogate losses that do not directly reflect the biological criteria used for evaluation, so better data fitting need not yield better biological predictions. To address this mismatch, we introduce \textbf{CellRFT}, a reinforcement fine-tuning framework that uses biological evaluation as direct training feedback. CellRFT uses policy-gradient optimization to learn from non-differentiable evaluations of generated cell populations and integrates multiple biological rewards through hierarchical reward aggregation. Comprehensive experiments demonstrate CellRFT's applicability across different pretrained models and effectiveness in improving perturbation prediction, reveal that optimizing one biological criterion can help or hinder others, and show that complementary rewards can improve criteria beyond those directly optimized, offering a way to probe how biological metrics shape model behavior, with the potential to inform evaluation design. Code will be made available.

cs.LG↗

arXiv2025

OmniFC: Rethinking Federated Clustering via Lossless and Secure Distance Reconstruction

Federated clustering (FC) aims to discover global cluster structures across decentralized clients without sharing raw data, making privacy preservation a fundamental requirement. There are two critical challenges: (1) privacy leakage during collaboration, and (2) robustness degradation due to aggregation of proxy information from non-independent and identically distributed (Non-IID) local data, leading to inaccurate or inconsistent global clustering. Existing solutions typically rely on model-specific local proxies, which are sensitive to data heterogeneity and inherit inductive biases from their centralized counterparts, thus limiting robustness and generality. We propose Omni Federated Clustering (OmniFC), a unified and model-agnostic framework. Leveraging Lagrange coded computing, our method enables clients to share only encoded data, allowing exact reconstruction of the global distance matrix--a fundamental representation of sample relationships--without leaking private information, even under client collusion. This construction is naturally resilient to Non-IID data distributions. This approach decouples FC from model-specific proxies, providing a unified extension mechanism applicable to diverse centralized clustering methods. Theoretical analysis confirms both reconstruction fidelity and privacy guarantees, while comprehensive experiments demonstrate OmniFC's superior robustness, effectiveness, and generality across various benchmarks compared to state-of-the-art methods. Code will be released.

cs.LG↗

arXiv2025

CCFC: Bridging Federated Clustering and Contrastive Learning

Federated clustering, an essential extension of centralized clustering for federated scenarios, enables multiple data-holding clients to collaboratively group data while keeping their data locally. In centralized scenarios, clustering driven by representation learning has made significant advancements in handling high-dimensional complex data. However, the combination of federated clustering and representation learning remains underexplored. To bridge this, we first tailor a cluster-contrastive model for learning clustering-friendly representations. Then, we harness this model as the foundation for proposing a new federated clustering method, named cluster-contrastive federated clustering (CCFC). Benefiting from representation learning, the clustering performance of CCFC even double those of the best baseline methods in some cases. Compared to the most related baseline, the benefit results in substantial NMI score improvements of up to 0.4155 on the most conspicuous case. Moreover, CCFC also shows superior performance in handling device failures from a practical viewpoint.

cs.LG↗

arXiv2024

CCFC++: Enhancing Federated Clustering through Feature Decorrelation

In federated clustering, multiple data-holding clients collaboratively group data without exchanging raw data. This field has seen notable advancements through its marriage with contrastive learning, exemplified by Cluster-Contrastive Federated Clustering (CCFC). However, CCFC suffers from heterogeneous data across clients, leading to poor and unrobust performance. Our study conducts both empirical and theoretical analyses to understand the impact of heterogeneous data on CCFC. Findings indicate that increased data heterogeneity exacerbates dimensional collapse in CCFC, evidenced by increased correlations across multiple dimensions of the learned representations. To address this, we introduce a decorrelation regularizer to CCFC. Benefiting from the regularizer, the improved method effectively mitigates the detrimental effects of data heterogeneity, and achieves superior performance, as evidenced by a marked increase in NMI scores, with the gain reaching as high as 0.32 in the most pronounced case.

cs.LG↗

arXiv2023

Federated clustering with GAN-based data synthesis

Federated clustering (FC) is an extension of centralized clustering in federated settings. The key here is how to construct a global similarity measure without sharing private data, since the local similarity may be insufficient to group local data correctly and the similarity of samples across clients cannot be directly measured due to privacy constraints. Obviously, the most straightforward way to analyze FC is to employ the methods extended from centralized ones, such as K-means (KM) and fuzzy c-means (FCM). However, they are vulnerable to non independent-and-identically-distributed (non-IID) data among clients. To handle this, we propose a new federated clustering framework, named synthetic data aided federated clustering (SDA-FC). It trains generative adversarial network locally in each client and uploads the generated synthetic data to the server, where KM or FCM is performed on the synthetic data. The synthetic data can make the model immune to the non-IID problem and enable us to capture the global similarity characteristics more effectively without sharing private data. Comprehensive experiments reveals the advantages of SDA-FC, including superior performance in addressing the non-IID problem and the device failures.

cs.LG↗

arXiv2023

Privacy-Preserving Federated Deep Clustering based on GAN

Federated clustering (FC) is an essential extension of centralized clustering designed for the federated setting, wherein the challenge lies in constructing a global similarity measure without the need to share private data. Conventional approaches to FC typically adopt extensions of centralized methods, like K-means and fuzzy c-means. However, these methods are susceptible to non-independent-and-identically-distributed (non-IID) data among clients, leading to suboptimal performance, particularly with high-dimensional data. In this paper, we present a novel approach to address these limitations by proposing a Privacy-Preserving Federated Deep Clustering based on Generative Adversarial Networks (GANs). Each client trains a local generative adversarial network (GAN) locally and uploads the synthetic data to the server. The server applies a deep clustering network on the synthetic data to establish $k$ cluster centroids, which are then downloaded to the clients for cluster assignment. Theoretical analysis demonstrates that the GAN-generated samples, shared among clients, inherently uphold certain privacy guarantees, safeguarding the confidentiality of individual data. Furthermore, extensive experimental evaluations showcase the effectiveness and utility of our proposed method in achieving accurate and privacy-preserving federated clustering.

cs.LG↗

arXiv2022

Selective clustering ensemble based on kappa and F-score

Clustering ensemble has an impressive performance in improving the accuracy and robustness of partition results and has received much attention in recent years. Selective clustering ensemble (SCE) can further improve the ensemble performance by selecting base partitions or clusters in according to diversity and stability. However, there is a conflict between diversity and stability, and how to make the trade-off between the two is challenging. The key here is how to evaluate the quality of the base partitions and clusters. In this paper, we propose a new evaluation method for partitions and clusters using kappa and F-score, leading to a new SCE method, which uses kappa to select informative base partitions and uses F-score to weight clusters based on stability. The effectiveness and efficiency of the proposed method is empirically validated over real datasets.

cs.LG↗

arXiv2020

Comment on "Improved mutual information measure for clustering, classification, and community detection"

A recent article proposed reduced mutual information for evaluation of clustering, classification and community detection. The motivation is that the standard normalized mutual information (NMI) may give counter-intuitive answers under certain conditions and particularly when the number of clusters differs between the two divisions under consideration. The motivation makes sense. However, the examples given in the article are not accurate, and this comment discusses why. In addition, this comment also empirically demonstrates that the reduced mutual information cannot handle the difficulties of NMI and even brings more. The necessity of Kappa is also empirically validated in this comment.

cs.SI↗

arXiv2019

Evaluation of Community Detection Methods

Community structures are critical towards understanding not only the network topology but also how the network functions. However, how to evaluate the quality of detected community structures is still challenging and remains unsolved. The most widely used metric, normalized mutual information (NMI), was proved to have finite size effect, and its improved form relative normalized mutual information (rNMI) has reverse finite size effect. Corrected normalized mutual information (cNMI) was thus proposed and has neither finite size effect nor reverse finite size effect. However, in this paper we show that cNMI violates the so-called proportionality assumption. In addition, NMI-type metrics have the problem of ignoring importance of small communities. Finally, they cannot be used to evaluate a single community of interest. In this paper, we map the computed community labels to the ground-truth ones through integer linear programming, then use kappa index and F-score to evaluate the detected community structures. Experimental results demonstrate the advantages of our method.

cs.SI↗

arXiv2018

Joint Nonnegative Matrix Factorization for Community Structures Detection in Signed Networks

Community structures detection in signed network is very important for understanding not only the topology structures of signed networks, but also the functions of them, such as information diffusion, epidemic spreading, etc. In this paper, we develop a joint nonnegative matrix factorization model to detect community structures. In addition, we propose modified partition density to evaluate the quality of community structures. We use it to determine the appropriate number of communities. The effectiveness of our approach is demonstrated based on both synthetic and real-world networks.

cs.SI↗

arXiv2018

On approximate equivalence of modularity, D and non-negative matrix factorization

Community structures detection is one of the fundamental problems in complex network analysis towards understanding the topology structures of the network and the functions of it. Nonnegative matrix factorization (NMF) is a widely used method for community detection, and modularity Q and modularity density D are criteria to evaluate the quality of community structures. In this paper, we establish the connections between Q, D and NMF for the first time. Q maximization can be approximately reformulated under the framework of NMF with Frobenius norm, especially when $n$ is large, and D maximization can also be reformulated under the framework of NMF. Q minimization can be reformulated under the framework of NMF with Kullback-Leibler divergence. We propose new methods for community structures detection based on the above findings, and the experimental results on synthetic networks demonstrate their effectiveness.

cs.SI↗

arXiv2017

On Equivalence of Likelihood Maximization of Stochastic Block Model and Constrained Nonnegative Matrix Factorization

Community structures detection in complex network is important for understanding not only the topological structures of the network, but also the functions of it. Stochastic block model and nonnegative matrix factorization are two widely used methods for community detection, which are proposed from different perspectives. In this paper, the relations between them are studied. The logarithm of likelihood function for stochastic block model can be reformulated under the framework of nonnegative matrix factorization. Besides the model equivalence, the algorithms employed by the two methods are different. Preliminary numerical experiments are carried out to compare the behaviors of the algorithms.

cs.SI↗

arXiv2017

Community Detection in Complex Networks using Link Prediction

Community detection and link prediction are both of great significance in network analysis, which provide very valuable insights into topological structures of the network from different perspectives. In this paper, we propose a novel community detection algorithm with inclusion of link prediction, motivated by the question whether link prediction can be devoted to improving the accuracy of community partition. For link prediction, we propose two novel indices to compute the similarity between each pair of nodes, one of which aims to add missing links, and the other tries to remove spurious edges. Extensive experiments are conducted on benchmark data sets, and the results of our proposed algorithm are compared with two classes of baseline. In conclusion, our proposed algorithm is competitive, revealing that link prediction does improve the precision of community detection.

cs.SI↗

arXiv2016

Effectiveness of Rapid Rail Transit System in Beijing

The effectiveness of rapid rail transit system is analyzed using tools of complex network for the first time. We evaluated the effectiveness of the system in Beijing quantitatively from different perspectives, including descriptive statistics analysis, bridging property, centrality property, ability of connecting different part of the system and ability of disease spreading. The results showed that the public transport of Beijing does benefit from the rapid rail transit lines, but there is still room to improve. The paper concluded with some policy suggestions regarding how to promote the system. This study offered significant insight that can help understand the public transportation better. The methodology can be easily applied to analyze other urban public systems, such as electricity grid, water system, to develop more livable cities.

cs.SI↗

arXiv2016

Community Detection in Complex Networks Using Density-based Clustering Algorithm

Like clustering analysis, community detection aims at assigning nodes in a network into different communities. Fdp is a recently proposed density-based clustering algorithm which does not need the number of clusters as prior input and the result is insensitive to its parameter. However, Fdp cannot be directly applied to community detection due to its inability to recognize the community centers in the network. To solve the problem, a new community detection method (named IsoFdp) is proposed in this paper. First, we use Isomap technique to map the network data into a low dimensional manifold which can reveal diverse pair-wised similarity. Then Fdp is applied to detect the communities in networks. An improved partition density function is proposed to select the proper number of communities automatically. We test our method on both synthetic and real-world networks, and the results demonstrate the effectiveness of our algorithm over the state-of-the-art methods.

cs.SI↗

arXiv2016

Link community detection through global optimization and the inverse resolution limit of partition density

We investigate the possibility of global optimization-based overlapping community detection, using link community framework. We first show that partition density, the original quality function used in link community detection method, is not suitable as a quality function for global optimization because it prefers breaking communities into triangles except in highly limited conditions. We analytically derive those conditions and confirm it with computational results on direct optimization of various synthetic and real-world networks. To overcome this limitation, we propose alternative approaches combining the weighted line graph transformation and existing quality functions for node-based communities. We suggest a new line graph weighting scheme, a normalized Jaccard index. Computational results show that community detection using the weighted line graphs generated with the normalized Jaccard index leads to a more accurate community structure.

physics.soc-ph↗

arXiv2015

Community detection in bipartite networks using weighted symmetric binary matrix factorization

In this paper we propose weighted symmetric binary matrix factorization (wSBMF) framework to detect overlapping communities in bipartite networks, which describe relationships between two types of nodes. Our method improves performance by recognizing the distinction between two types of missing edges---ones among the nodes in each node type and the others between two node types. Our method can also explicitly assign community membership and distinguish outliers from overlapping nodes, as well as incorporating existing knowledge on the network. We propose a generalized partition density for bipartite networks as a quality function, which identifies the most appropriate number of communities. The experimental results on both synthetic and real-world networks demonstrate the effectiveness of our method.

cs.SI↗

arXiv2013

Enhanced Community Structure Detection in Complex Networks with Partial Background Information

Community structure detection in complex networks is important since it can help better understand the network topology and how the network works. However, there is still not a clear and widely-accepted definition of community structure, and in practice, different models may give very different results of communities, making it hard to explain the results. In this paper, different from the traditional methodologies, we design an enhanced semi-supervised learning framework for community detection, which can effectively incorporate the available prior information to guide the detection process and can make the results more explainable. By logical inference, the prior information is more fully utilized. The experiments on both the synthetic and the real-world networks confirm the effectiveness of the framework.

cs.SI↗