Ensuring the privacy of local datasets has emerged as an important concern in decentralized learning. However, the inherent privacy-utility tradeoff remains a fundamental challenge for privacy preserving decentralized algorithms. To address this issue, we introduce Positive-Incentive Noise Generator (PING), a novel mechanism designed to eliminate negative impact of privacy noise on convergence while defending against powerful colluding inference attacks. PING leverages network topologies and lightweight encryption-decryption operations to generate correlated noise. Building upon PING, we propose PP-DPIN, a privacy preserving stochastic algorithm tailored for decentralized learning. By integrating differential privacy and differential information entropy, we provide a comprehensive privacy quantification for PP-DPIN, with at least half nodes achieving arbitrarily strong privacy guarantees. Furthermore, convergence rate of PP-DPIN is established under stochastic convex and nonconvex settings, which characterizes the impact of privacy noise and demonstrates the linear speedup relative to the network size. Experiments on computer vision tasks validate PP-DPIN's superior performance and robustness against attacks compared to state-of-the-art methods.

This article is concerned with the zonotopic set-membership fusion estimation (SMFE) problem for a class of complex networks (CNs). The measurements of the CNs are transmitted to a remote fusion center through a shared communication network. Due to the limited network bandwidth, the transmissions of the measurement information occur intermittently, and the nodes' transmission intervals may exceed their sampling periods. To enhance the utilization of the measurement information, each node of the CN is equipped with a buffer for real-time data storage, so that the fusion center can utilize more measurement information at time instants when the node's transmission interval is larger than its sampling period. The aim of this article is to design SMFE algorithms based on both the parallel fusion scheme and the data-compression fusion scheme, respectively, using the data received at the fusion center. First, by iterating the state equation of the CN, a batch processing method is proposed to process the input data of the fusion center concurrently. Subsequently, by employing the zonotopic set-membership estimation (SME) technique, the desired SMFE algorithms are designed. Moreover, sufficient criteria are established to ensure that the sizes of the output zonotopes of the SMFE algorithms remain uniformly bounded. Finally, two numerical examples are presented to illustrate the effectiveness of the proposed algorithms.

Distributed energy scheduling in residential microgrids faces challenges from renewable uncertainty and privacy constraints. While multiagent reinforcement learning (MARL) enables decentralized coordination, classical MARL methods suffer from exponential growth in the joint state–action space, leading to high computational complexity and low sample efficiency. Moreover, centralized training paradigms often rely on shared representations, raising privacy concerns. In this article, we propose a multiagent distributional reinforcement learning framework with the dynamic hyper policy network (DHPN), which constructs permutation-invariant (PI) representations via entitywise dynamic weight generation and multihead attention, reducing input redundancy and obviating the need for full joint information. To capture stochasticity from renewable variability and demand, we further introduce a distributional value function factorization framework to model return distributions, leveraging quantile regression and mean–shape decomposition. Experiments on a data-driven microgrid demonstrate that the DHPN outperforms strong baselines by achieving substantially lower mean-square error in representational evaluation and higher cumulative rewards, while exhibiting faster convergence, enhanced training stability, and reduced grid fluctuations. These results highlight the effectiveness of combining PI representations with distributional value modeling for scalable privacy-aware multiagent energy management.

Logs of large-scale cloud systems record diverse system events, ranging from routine statuses to critical errors. As the fundamental step of automated log analysis, log parsing is to transform unstructured logs into structured data for easier management and analysis. However, existing syntax-based and deep learning-based parsers struggle with complex real-world logs. Recent parsers based on large language models (LLMs) achieve higher accuracy, but they typically rely on online APIs (e.g., ChatGPT), raising privacy concerns and suffering from network latency. Moreover, with the rise of artificial intelligence for IT operations (AIOps), traditional parsers that focus on syntax-level templates fail to capture the semantics of dynamic log parameters, limiting their usefulness for downstream tasks. These challenges highlight the need for semantic log parsing that goes beyond template extraction to understand parameter semantics.This paper presents SemanticLog, an effective and efficient semantic log parser powered by open-source LLMs. SemanticLog adapts the structure of LLMs to the log parsing task, leveraging their rich knowledge while safeguarding log data privacy. It first extracts informative feature representations from log data, then refines them through fine-grained semantic perception to enable accurate template and parameter extraction together with semantic category prediction. To boost scalability, SemanticLog introduces the EffiParsing tree for faster inference on large-scale logs. Extensive experiments on the LogHub-2.0 dataset show that SemanticLog significantly outperforms the state-of-the-art log parsers in terms of accuracy. Moreover, it also surpasses existing LLM-based parsers in efficiency while showcasing advanced semantic parsing capability. Notably, SemanticLog employs much smaller open-source LLMs compared to existing LLM-based parsers (mainly based on ChatGPT), while maintaining better capability of log data privacy protection.

This article is concerned with a distributed algorithm for seeking the Nash equilibrium in noncooperative games with partial-decision information, which simultaneously addresses the protection of individual privacy and ensures fast algorithmic convergence. First, a differential privacy mechanism is used in the fully distributed consensus-based projected pseudo-gradient algorithm to obfuscate shared messages over the communication network and quantify the algorithm's privacy level. To achieve fast convergence, a novel relaxed inertial method is designed, consisting of two steps with independently designed parameters: 1) a relaxation step and 2) an inertia step. The adaptive inertia coefficient in the inertia step is designed based on the iteration error of the players' estimated decisions and a decaying sequence, with the only requirement being the non-negativity of its internal parameters. Compared to existing approaches, our algorithm exhibits high flexibility in parameter selection. Furthermore, we analyze the algorithm's convergence and differential privacy under both linearly decaying and fixed stepsizes within a unified framework, providing sufficient conditions that are independent of the number of players. Finally, numerical simulations validate the algorithm's potential, demonstrating significant improvements in convergence rate, accuracy, and privacy level.

This article studies distributed Nash equilibrium seeking of a group of players based on an event-triggered mechanism under switching interaction topologies. First, a dynamic event-triggered function is proposed, which relies on the gradient of the player's payoff function and the triggering state information of its neighbors. It enables a rapid response to system changes and is able to effectively eliminate the Zeno behavior. Next, to maintain the strong connectivity in the directed graph for the links switched by the event-triggered mechanism at all times, a switching-triggered communication mechanism is designed in which triggered conditions are enhanced to include the instants of each player's link changes. Furthermore, by restricting switching frequency, this article establishes sufficient conditions with parameter ranges for exponential convergence to the Nash equilibrium. Finally, simulation results show the validity of the proposed strategy.

The storage and transmission of large-scale data are recognized as significant challenges that incur substantial costs due to the massive volume of information involved. Lossless compression offers a potential solution to this problem. In this work, we explore the online lossless state compression (LSC) in finite-valued networks by effectively combining the methods of finite state automata and reinforcement learning. To achieve online lossless compression, the concept ofk-step compressed walks (state sequences) is introduced. By constructing a finite state automaton to identify all insertable walks, the recoverability conditions of online lossless compression are presented. For any given compressed walkw’c, one algorithm is proposed to find a minimal recoverablekmin-step compressed walk, thereby improving the compression ratio. Furthermore, it can be found that if a compressed walk keeps recoverability, adding more states to it would not break this property. Additionally, a model-free reinforcement learning framework based on Q-learning is developed to obtain globally minimal recoverable compressed walks. Finally, a numerical example is given to demonstrate the efficiency of the proposed online lossless compression scheme, achieving a compression ratio of 44.4%.

This paper proposes an anime audio retrieval method based on audio separation and feature recognition techniques, aiming to help users conveniently locate their desired audio segments and enhance the overall user experience. Additionally, by establishing an audio fingerprint database and a corresponding copyright information management system, it becomes possible to track and manage the audio content within anime, effectively preventing piracy and unauthorized use, thereby improving the management and protection of audio resources. Traditional methods for anime audio feature recognition suffer from issues like low efficiency and subjective factors. In contrast, the proposed approach overcomes these limitations by automatically separating and extracting audio fingerprints from different audio sources within anime and creating an anime audio fingerprint database for fast retrieval. The paper utilizes an improved audio separation model based on the efficient channel attention mechanism to separate the anime audio. Subsequently, feature recognition is performed on the separated anime audio, employing a contrastive learning-based audio fingerprint retrieval method for anime audio fingerprinting. Experimental results demonstrate that the proposed algorithm effectively alleviates the issue of poor audio separation performance in anime audio, while also improving retrieval efficiency and accuracy, meeting the demands for anime audio content retrieval.

This brief devotes to investigating the long-run behavior estimation of temporal Boolean networks (TBNs) with multiple data losses, especially the asymptotical stability. The information transmission is modeled by Bernoulli variables, based on which an augmented system is constructed to facilitate the analysis. A theorem guarantees that the asymptotical stability of the original system can be converted to that of the augmented system. Subsequently, one necessary and sufficient condition is obtained for asymptotical stability. Furthermore, an auxiliary system is derived to study the synchronization issue of the ideal TBNs with normal data transmission and TBNs with multiple data losses, as well as an effective criterion for verifying synchronization. Finally, numerical examples are given to illustrate the validity of the theoretical results.

This article addresses distributed optimization problems, in which a group of agents cooperatively minimize the sum of their private objective functions via information exchanging. Building on alternating direction method of multipliers (ADMM), we propose a privacy-preserving and communication-efficient decentralized quadratically approximated ADMM algorithm, termed PC-DQM, for solving such type of problems under the scenario of limited communication. In PC-DQM, an event-triggered mechanism is designed to schedule the communication instants for reducing communication cost. Simultaneously, for privacy preservation, a Hessian matrix with perturbed noise is introduced to quadratically approximate the objective function, which results in a closed form of primal vector update and then avoids solving a subproblem at each iteration with possible high computation cost. In addition, the triggered scheme is also utilized to schedule the update of Hessian, which can also reduce computation cost. We theoretically show that PC-DQM can protect privacy but without losing accuracy. In addition, we rigorously prove that PC-DQM converges linearly to the exact optimal solution for strongly convex and smooth objective functions. Finally, numerical simulation is presented to illustrate the effectiveness and efficiency of our algorithm.

This article investigates the resilient output synchronization problem of a class of linear heterogeneous multiagent systems subjected to denial-of-service (DoS) attacks. Two types of control mechanisms, namely, event- and self-triggered control mechanisms, are presented so as to cut down unnecessary information transmission. Both of these two mechanisms are distributed, and thus, only local information of each agent and its neighboring agents is adopted for the event condition design. The DoS attacks are considered to be aperiodic, and the quantitative relationship between the attributes of the DoS attacks and the synchronization is also revealed. It is shown that the output synchronization can be achieved exponentially in the presence of DoS attacks under the proposed control mechanisms. The validness of the provided mechanisms is certified by a simulation example.

This paper is concerned with the generalized Nash equilibrium (GNE) seeking problem of noncooperative games in dynamic environments, where the cost function and coupled constraint of each agent are time-varying. In this case, each agent is required to make a decision before obtaining its cost function and a local inequality constraint. The purpose of the addressed problem is to establish a new distributed primary-dual and mirror descent online algorithm with inertia that is capable of seeking the GNE via time-varying communication graphs and has the potential of achieving a low average regret. Then, two information transmission modes and two estimate update strategies are discussed, respectively. Finally, a simulation example is presented to illustrate the effectiveness of the algorithms, and to further compare their performances.

In this article, a periodic self-triggered impulsive (PSTI) control scheme is proposed to achieve synchronization of neural networks (NNs). Two kinds of impulsive gains with constant and random values are considered, and the corresponding synchronization criteria are obtained based on tools from impulsive control, event-driven control theory, and stability analysis. The designed triggering protocol is simpler, easier to implement, and more flexible compared with some previously reported algorithms as the protocol combines the advantages of the periodic sampling and event-driven control. In addition, the chaotic synchronization of NNs via the presented PSTI sampling is further applied to encrypt images. Several examples are also utilized to illustrate the validity of the presented synchronization algorithm of NNs based on PSTI control and its potential applications in image processing.

Multiagent systems (MASs) provide an effective means for coordinating spatially distributed and networked agents (or nodes, subsystems) such that the desired cooperative tasks can be accomplished with promising reliability, manipulability, scalability, and efficiency. One key issue in the study of MASs is the design of distributed cooperative control protocol and algorithm that depend on only local and real-time information exchanges among interacting agents over networks. However, network-enabled information sharing and increasing connectivity in practical MASs present several attack factors for malicious adversaries, thereby rendering secure control of MASs fundamentally significant. This article provides a brief survey of systems and control technologies that have been available for addressing different secure control problems of MASs in the face of various malicious attacks. First, attacks on MASs are classified based on different configuration layers. Then, the existing attack models and strategies on communication layer and agent layer are systematically examined, respectively. Furthermore, some typical secure control techniques for MASs that have been employed to handle these attacks are surveyed. Finally, several challenging issues are envisioned for potential future research.

This paper proposes a distributed hierarchical control framework for energy storage systems (ESSs) in DC microgrid clusters, which achieves voltage regulation and current sharing for ESSs in each microgrid as well as the whole microgrid cluster. The primary control stage adopts a droop controller which only requires local information while the secondary control stage provides correction terms for ESSs within microgrids. The tertiary control stage samples the pinned ESSs in different microgrids with low sampling rate to provide the voltage setpoint, which ensures global current sharing among microgrid cluster. The corresponding multilayered event-triggered consensus algorithm for clusters is proposed to reduce the communication cost generated by operation of the distributed controller. Both the control framework and the consensus algorithm can be extended for satisfying higher dimensional regulation needs. The controller is validated in a DC microgrid cluster through simulation under different scenarios, and the results illustrate the effectiveness of the proposed controller.

This paper develops a fully distributed self-triggered framework for achieving the second-order consensus in multiagent systems. In this framework, a fully distributed self-triggered scheme is proposed to schedule information transmission for each communication channel. Thus, the communication frequency of each channel is significantly reduced. Here, communication over different channels is independent with each other and, thus, a channel-based control protocol is further proposed. The update times of control protocol could be also lowered in this framework. Here, an iterative evaluation method is constructed to design a fully distributed self-triggered scheme without involving any global information, especially the eigenvalue information of the Laplacian matrix. In addition, by introducing several variables, some sufficient conditions for achieving the second-order consensus and average consensus are obtained, respectively, in a distributed fashion, and the Zeno behavior is successfully eliminated. Finally, a simulation example is provided to verify the theoretical analysis.

This paper develops a fully distributed framework to investigate the cooperative behavior of multiagent systems in the presence of distributed denial-of-service (DoS) attacks launched by multiple adversaries. In such an insecure network environment, two kinds of communication schemes, that is, sample-data and event-triggered communication schemes, are discussed. Then, a fully distributed control protocol with strong robustness and high scalability is well designed. This protocol guarantees asymptotic consensus against distributed DoS attacks. In this paper, "fully" emphasizes that the eigenvalue information of the Laplacian matrix is not required in the design of both the control protocol and event conditions. For the event-triggered case, two effective dynamical event-triggered schemes are proposed, which are independent of any global information. Such event-triggered schemes do not exhibit Zeno behavior even in the insecure environment. Finally, a simulation example is provided to verify the effectiveness of theoretical analysis.

In this paper, a controller with multilayer structure is proposed to regulate the thermostatically controlled loads (TCLs), so that power sharing and comfort states consensus can be achieved. Since TCLs have great potential to reduce the fluctuations caused by photovoltaic in the building microgrid community, the control of a cluster of TCLs has practical significance. The multilayer structure can capture the nature of inner and inter communications among the building microgrids. The self-triggered mechanism is adopted to avoid continuous data transmission. The controller is validated through study of two different building microgrid communities with different number of TCLs.

This brief proposes a unified theoretical framework to investigate the finite/fixed-time synchronization of complex networks with stochastic disturbances. By designing a common pinning controller with different ranges of power parameters, both the goals of finite-time and fixed-time synchronization in probability for the network topology containing spanning trees can be achieved. Moveover, with the help of finite-time stochastic stability theory, two types of explicit expressions of finite/fixed (dependent/independent on the initial values) settling times are calculated as well. One numerical example is finally presented to demonstrate the effectiveness of the theoretical analysis.

This paper investigates the leader-following consensus issue with event/self-triggered schemes under an unreliable network environment. First, we characterize network communication and control protocol update in the presence of denial-of-service (DoS) attacks. In this situation, an event-triggered communication scheme is first proposed to effectively schedule information transmission over the network possibly subject to malicious attacks. In this communication framework, synchronous and asynchronous updated strategies of control protocols are constructed to achieve leader-following consensus in the presence of DoS attacks. Moreover, to further reduce the cost induced by event detection, a self-triggered communication scheme is proposed in which the next triggering instant can be determined by computing with the most updated information. Finally, a numerical example is provided to verify the effectiveness of the proposed communication schemes and updated strategies in the unreliable network environment.

This paper is concerned with the finite-horizon consensus problem for a class of discrete time-varying multiagent systems with external disturbances and missing measurements. To improve the communication reliability, redundant channels are introduced and the corresponding protocol is constructed for the information transmission over redundant channels. An event-triggered scheme is adopted to determine whether the information of agents should be transmitted to their neighbors. Subsequently, an observer-type event-triggered control protocol is proposed based on the latest received neighbors' information. The purpose of the addressed problem is to design a time-varying controller based on the observed information to achieve the consensus performance in a finite horizon. By utilizing a constrained recursive Riccati difference equation approach, some sufficient conditions are obtained to guarantee the consensus performance, and the controller parameters are also designed. Finally, a numerical example is provided to demonstrate the desired reliability of redundant channels and the effectiveness of the event-triggered control protocol.

This paper studies the dynamics of multiagent systems with a multilayer structure, proposing a novel layered event-triggered scheme (LETS) for consensus. This LETS emphasizes on synchronous information transmission in the same layer but asynchronous message update between different layers, which differs from the existing centralized or distributed event-triggered schemes. Moreover, under the LETS, agents in different layers achieve asymptotical consensus eventually and the Zeno behavior is successfully eliminated. Furthermore, an algorithm is provided to avoid continuous event detection, verified by a numerical example.

This paper investigates the leader-following consensus for multiagent systems with general linear dynamics by means of event-triggered scheme (ETS). We propose three types of schemes, namely, distributed ETS (distributed-ETS), centralized ETS (centralized-ETS), and clustered ETS (clustered-ETS) for different network topologies. All these schemes guarantee that all followers can track the leader eventually. It should be emphasized that all event-triggered protocols in this paper depend on local information and their executions are distributed. Moreover, it is shown that such event-triggered mechanism can significantly reduce the frequency of control's update. Further, positive inner-event time intervals are assured for those cases of distributed-ETS, centralized-ETS, and clustered-ETS. In addition, two methods are proposed to avoid continuous communication between agents for event detection. Finally, numerical examples are provided to illustrate the effectiveness of the ETSs.

This brief investigates the problem of finite-time robust consensus (FTRC) for second-order nonlinear multiagent systems with external disturbances. Based on the global finite-time stability theory of discontinuous homogeneous systems, a novel finite-time convergent discontinuous disturbed observer (DDO) is proposed for the leader-following multiagent systems. The states of the designed DDO are then used to design the control inputs to achieve the FTRC of nonlinear multiagent systems in the presence of bounded disturbances. The simulation results are provided to validate the effectiveness of these theoretical results.

This paper proposes a new event-triggered scheme, in which each agent has various mechanisms to respectively determine when to exchange information with each of its different neighbors. This kind of event-triggered mechanism avoids two constraints in previous mechanisms: (1) simultaneous event detection; (2) synchronous information transmission to all of neighbors. Thus our proposed scheme can be applied into more general situations. In addition, our scheme is able to guarantee the asymptotic consensus and exclude the Zeno behavior. Furthermore, an alternative self-triggered algorithm is presented to thoroughly exclude continuous event detection. Finally, a numerical example is provided to verify the theoretical analysis.

This paper studies the stability and Hopf bifurcation in a class of high-dimension neural network involving the discrete and distributed delays under a new framework. By introducing some virtual neurons to the original system, the impact of distributed delay can be described in a simplified way via an equivalent new model. This paper extends the existing works on neural networks to high-dimension cases, which is much closer to complex and real neural networks. Here, we first analyze the Hopf bifurcation in this special class of high dimensional model with weak delay kernel from two aspects: one is induced by the time delay, the other is induced by a rate parameter, to reveal the roles of discrete and distributed delays on stability and bifurcation. Sufficient conditions for keeping the original system to be stable, and undergoing the Hopf bifurcation are obtained. Besides, this new framework can also apply to deal with the case of the strong delay kernel and corresponding analysis for different dynamical behaviors is provided. Finally, the simulation results are presented to justify the validity of our theoretical analysis.

This paper studies the stability and Hopf bifurcation in a high-dimension neural network involving the discrete and distributed delays. Such model extends the existing models of neural networks from low-dimension to high-dimension. Therefore, our model is much close to large real neural networks. Here, the delay is chosen as the bifurcation parameter and we obtain the sufficient conditions for the system keeping stable and undergoing the Hopf bifurcation. Moreover, the software package DDE-BIFTOOL is introduced to better display the properties of the system and the effect of gain parameters of the system and delay kernel on the onset of the bifurcation. The simulation results further justify the validity of our theoretical analysis.

BACKGROUND: NAC domain transcription factors are important transcriptional regulators involved in plant growth, development and stress responses. Recent studies have revealed several classes of NAC transcriptional factors crucial for controlling secondary cell wall biosynthesis. These transcriptional factors mainly include three classes, SND, NST and VND. Despite progress, most current analysis is carried out in the model plant Arabidopsis. Moreover, many downstream genes regulated by these transcriptional factors are still not clear. METHODS: In order to identify the key homologue genes across species and discover the network controlling cell wall biosynthesis, we carried out comparative genome analysis of NST, VND and SND genes across 19 higher plant species along with computational modelling of genes regulated or co-regulated with these transcriptional factors. RESULTS: The comparative genome analysis revealed that evolutionarily the secondary-wall-associated NAC domain transcription factors first appeared in Selaginella moellendorffii. In fact, among the three groups, only VND genes appeared in S. moellendorffii, which is evolutionarily earlier than the other two groups. The Arabidopsis and rice gene expression analysis showed specific patterns of the secondary cell wall-associated NAC genes (SND, NST and VND). Most of them were preferentially expressed in the stem, especially the second internodes. Furthermore, comprehensive co-regulatory network analysis revealed that the SND and MYB genes were co-regulated, which indicated the coordinative function of these transcriptional factors in modulating cell wall biosynthesis. In addition, the co-regulatory network analysis revealed many novel genes and pathways that could be involved in cell wall biosynthesis and its regulation. The gene ontology analysis also indicated that processes like carbohydrate synthesis, transport and stress response, are coordinately regulated toward cell wall biosynthesis. CONCLUSIONS: Overall, we provided a new insight into the evolution and the gene regulatory network of a subgroup of the NAC gene family controlling cell wall composition through bioinformatics data mining and bench validation. Our work might benefit to elucidate the possible molecular mechanism underlying the regulation network of secondary cell wall biosynthesis.

BACKGROUND: PHB (Prohibitin) gene family is involved in a variety of functions important for different biological processes. PHB genes are ubiquitously present in divergent species from prokaryotes to eukaryotes. Human PHB genes have been found to be associated with various diseases. Recent studies by our group and others have shown diverse function of PHB genes in plants for development, senescence, defence, and others. Despite the importance of the PHB gene family, no comprehensive gene family analysis has been carried to evaluate the relatedness of PHB genes across different species. In order to better guide the gene function analysis and understand the evolution of the PHB gene family, we therefore carried out the comparative genome analysis of the PHB genes across different kingdoms. RESULTS: The relatedness, motif distribution, and intron/exon distribution all indicated that PHB genes is a relatively conserved gene family. The PHB genes can be classified into 5 classes and each class have a very deep evolutionary origin. The PHB genes within the class maintained the same motif patterns during the evolution. With Arabidopsis as the model species, we found that PHB gene intron/exon structure and domains are also conserved during the evolution. Despite being a conserved gene family, various gene duplication events led to the expansion of the PHB genes. Both segmental and tandem gene duplication were involved in Arabidopsis PHB gene family expansion. However, segmental duplication is predominant in Arabidopsis. Moreover, most of the duplicated genes experienced neofunctionalization. The results highlighted that PHB genes might be involved in important functions so that the duplicated genes are under the evolutionary pressure to derive new function. CONCLUSION: PHB gene family is a conserved gene family and accounts for diverse but important biological functions based on the similar molecular mechanisms. The highly diverse biological function indicated that more research needs to be carried out to dissect the PHB gene function. The conserved gene evolution indicated that the study in the model species can be translated to human and mammalian studies.

BACKGROUND: Expression microarray analysis is one of the most popular molecular diagnostic techniques in the post-genomic era. However, this technique faces the fundamental problem of potential cross-hybridization. This is a pervasive problem for both oligonucleotide and cDNA microarrays; it is considered particularly problematic for the latter. No comprehensive multivariate predictive modeling has been performed to understand how multiple variables contribute to (cross-) hybridization. RESULTS: We propose a systematic search strategy using multiple multivariate models [multiple linear regressions, regression trees, and artificial neural network analyses (ANNs)] to select an effective set of predictors for hybridization. We validate this approach on a set of DNA microarrays with cytochrome p450 family genes. The performance of our multiple multivariate models is compared with that of a recently proposed third-order polynomial regression method that uses percent identity as the sole predictor. All multivariate models agree that the 'most contiguous base pairs between probe and target sequences,' rather than percent identity, is the best univariate predictor. The predictive power is improved by inclusion of additional nonlinear effects, in particular target GC content, when regression trees or ANNs are used. CONCLUSION: A systematic multivariate approach is provided to assess the importance of multiple sequence features for hybridization and of relationships among these features. This approach can easily be applied to larger datasets. This will allow future developments of generalized hybridization models that will be able to correct for false-positive cross-hybridization signals in expression experiments.