Correlations among variables in complex ecosystems such as weather systems and financial markets result in large amounts of dynamic and co-evolving time sequences. The benefits of discovering and predicting intricate patterns (aka regimes) in time sequences are multifold, including better understanding of the ecosystem dynamics, optimizing model selection, and improving interpretability of results. Despite recent advancements, existing methods primarily emphasize predictive accuracy, which might overshadow the need to comprehend the structural dynamics within the series. Additionally, these methods often encounter one or more of the following limitations: (1) difficulty in identifying regimes within domain-dependent segmentations; (2) inability to integrate nonlinear relationships across time series; (3) lack of an effective method to encapsulate the temporal behaviors. To tackle these challenges, we introduce a twin learning regime-switch model to simultaneously learn domain-agnostic segmentation and regime switch in a principled way. Specifically, we devise a kernel-based method that determines the duration of regime and captures dynamic switches through potent representations, accounting for the non-linear interactions between series. With this model, it is feasible to automatically achieve the two subtasks of identifying the optimal regimes and determining the most suitable segmentation. Experimental results on synthetic and real-world datasets indicate that our method is capable of revealing the structures that underpin the behavior of co-evolving ecosystems, which display different dynamics. These structures can be leveraged to better define regimes with superior predictive capabilities compared to widely used traditional models and state-of-the-art neural network models. • We propose a novel regime-switch model to identify domain-agnostic regimes in time series. • We develop a kernel representation learning approach to capturing dynamic regime switches. • This representation allows effectively modeling nonlinear interactions and co-evolving patterns in time series. • Our model transforms heavy sets of time series into a lighter and meaningful structure, enabling a deeper understanding of structural dynamics. • The extensive experimental results showcase superior predictive performance compared to traditional and state-of-the-art methods.

In the contemporary of deep learning, where models often grapple with the challenge of simultaneously achieving robustness against adversarial attacks and strong generalization capabilities, this study introduces an innovative Local Feature Masking (LFM) strategy aimed at fortifying the performance of Convolutional Neural Networks (CNNs) on both fronts. During the training phase, we strategically incorporate random feature masking in the shallow layers of CNNs, effectively alleviating overfitting issues, thereby enhancing the model’s generalization ability and bolstering its resilience to adversarial attacks. LFM compels the network to adapt by leveraging remaining features to compensate for the absence of certain semantic features, nurturing a more elastic feature learning mechanism. The efficacy of LFM is substantiated through a series of quantitative and qualitative assessments, collectively showcasing a consistent and significant improvement in CNN’s generalization ability and resistance against adversarial attacks—a phenomenon not observed in current and prior methodologies. The seamless integration of LFM into established CNN frameworks underscores its potential to advance both generalization and adversarial robustness within the deep learning paradigm. Through comprehensive experiments, including robust person re-identification baseline generalization experiments and adversarial attack experiments, we demonstrate the substantial enhancements offered by LFM in addressing the aforementioned challenges. This contribution represents a noteworthy stride in advancing robust neural network architectures.

We investigate the problem of discovering and forecasting regular regime switches in a financial ecosystem comprising multiple time series. Such regime switches, indicative of varying market behaviors across distinct time intervals, are pivotal for a nuanced understanding of market dynamics, which in turn allows informed model selection for forecasting and enhanced interpretability of predictive outcomes. Despite strides in this domain, prevailing methodologies often falter due to: (1) an inability to effectively model the temporal behaviors inherent in financial series; and (2) neglecting the interdependencies among series when discovering regimes. In this paper, we propose RHINE, a Regime-switcHIng model with Nonlinear rEpresentation. RHINE stands out with its kernel-based representation, adept at capturing the dynamic shifts in market regimes. This representation encapsulates the nonlinear interplay across multiple financial time series. By leveraging the kernel representation, we introduce an eigengap thresholding measure, designed to automatically discern the optimal number of financial market regimes, enhancing the model's adaptability to market fluctuations. Empirical assessments on both synthetic and real-world stock market datasets underscore RHINE's prowess. The findings illuminate that the inherent structures governing financial market behaviors are dynamic, and harnessing these dynamics via RHINE leads to a regime-based model that outperforms both conventional and state-of-the-art neural network models in predictive capabilities.

Electroencephalography (EEG) is a unique and valuable ancillary examination, which is essential for the diagnosis and analysis of various neurological diseases. Deep learning methods have been demonstrated to be very promising for challenging EEG screening tasks. However, most of them mainly focus on improving the representation ability by increasing the network depth and width, and cannot take full advantage of the rich hierarchical feature information. More importantly, the importance of utilizing the spatial information between features in a network model is ignored. To this end, a novel multi-level feature fusion capsule network (MFF-CapsNet) is proposed for accurate and efficient EEG pathology detection. Firstly, we devise a new lightweight feature fusion module as a basic unit of feature extraction, through concatenating different level feature maps densely. And then, the innovative CapsNet structure enables it to well capture the significant spatial relationship among different features. Especially, compared with the original CapsNet, we design two particular layers in this structure to reduce the number of parameters, memory, as well as computation, so that the probability of overfitting can be lessened. The experimental outcomes show that MFF-CapsNet can effectively differentiate EEG recordings as pathological or healthy, and its performance is better than the current advanced methodologies, which is the first to meet the requirement of clinical application.

BACKGROUND: Inferring gene regulatory networks (GRNs) from gene expression data remains a challenge in system biology. In past decade, numerous methods have been developed for the inference of GRNs. It remains a challenge due to the fact that the data is noisy and high dimensional, and there exists a large number of potential interactions. RESULTS: We present a novel method, namely priori-fused boosting network inference method (PFBNet), to infer GRNs from time-series expression data by using the non-linear model of Boosting and the prior information (e.g., the knockout data) fusion scheme. Specifically, PFBNet first calculates the confidences of the regulation relationships using the boosting-based model, where the information about the accumulation impact of the gene expressions at previous time points is taken into account. Then, a newly defined strategy is applied to fuse the information from the prior data by elevating the confidences of the regulation relationships from the corresponding regulators. CONCLUSIONS: The experiments on the benchmark datasets from DREAM challenge as well as the E.coli datasets show that PFBNet achieves significantly better performance than other state-of-the-art methods (Jump3, GEINE3-lag, HiDi, iRafNet and BiXGBoost).

The linguistic intuitionistic fuzzy sets (LIFSs) and linguistic Pythagorean fuzzy sets (LPFSs) are two linguistic orthopair fuzzy sets whose membership grades are pairs of linguistic terms from the predefined linguistic term sets (LTSs). One linguistic term indicates the membership degree (MD), while the other one gives the nonmembership degree (NMD). In each LIFS, the sum of the subscripts of MD and NMD is less than the cardinality of LTS. In the LPFSs, the sum of the squares of the subscripts of MD and NMD is less than the square of the cardinality of LTS. In this paper, we propose a general form of these two linguistic orthopair fuzzy sets, which can be named linguistic q-rung orthopair fuzzy sets. We devise the operational laws, based on which, the linguistic q-rung orthopair fuzzy weighted averaging (LqROFWA) operator and linguistic q-rung orthopair fuzzy weighted geometric (LqROFWG) operator are developed to aggregate the linguistic q-rung orthopair fuzzy numbers (LqROFNs). Then, the novel interactional operational laws that consider the interactions between the MD and NMD from different LqROFNs are given. The partitioned geometric Heronian mean (PGHM) operator can effectively solve the decision-making problems in which the attributes grouped into the same clusters have interrelationships and the attributes belonging to different clusters have no interrelationship. Based on these novel operational laws and PGHM operator, the linguistic q-rung orthopair fuzzy interactional PGHM (LqROFIPGHM) operator and linguistic q-rung orthopair fuzzy interactional weighted PGHM (LqROFIWPGHM) operator are proposed and their properties are discussed. Based on the LqROFIWPGHM operator, an efficient multiattribute group decision-making model is given to deal with the linguistic q-rung orthopair fuzzy information. Finally, the superiorities of the interactional operational laws and LqROFIWPGHM operator are tested using some illustrative examples.

Recently, studies have been performed for speech emotion recognition. However, little research focused on the emotion of the elderly, especially the lonely elderly. In this paper, we propose a six layer Wavelet Packet Coefficients Model for speech emotion recognition of the Chinese elderly. Six lay er Wavelet Packet Coefficients, Mel Frequency Cepstrum Coefficient and the Fourier Parameter features are extracted from speech emotion database of Chinese elderly, respectively. Experimental results show that the six layer wavelet packet coefficients features are effective for recognizing emotions from speech. In particularly, when combining these three features, the recognition rates of the elderly can be improved.

Cluster validation, which is the process of evaluating the quality of clustering results, plays an important role for practical machine learning systems. Categorical sequences, such as biological sequences in computational biology, have become common in real-world applications. Different from previous studies, which mainly focused on attribute-value data, in this paper, we work on the cluster validation problem for categorical sequences. The evaluation of sequences clustering is currently difficult due to the lack of an internal validation criterion defined with regard to the structural features hidden in sequences. To solve this problem, in this paper, a novel cluster validity index (CVI) is proposed as a function of clustering, with the intracluster structural compactness and intercluster structural separation linearly combined to measure the quality of sequence clusters. A partition-based algorithm for robust clustering of categorical sequences is also proposed, which provides the new measure with high-quality clustering results by the deterministic initialization and the elimination of noise clusters using an information theoretic method. The new clustering algorithm and the CVI are then assembled within the common model selection procedure to determine the number of clusters in categorical sequence sets. A case study on commonly used protein sequences and the experimental results on some real-world sequence sets from different domains are given to demonstrate the performance of the proposed method.

Survival prediction is crucial to healthcare research, but is confined primarily to specific types of data involving only the present measurements. This paper considers the more general class of healthcare data found in practice, which includes a wealth of intermittently varying historical measurements in addition to the present measurements. Making survival predictions on such data bristles with challenges to the existing prediction models. For this reason, we propose a new semi-proportional hazards model using locally time-varying coefficients, and a novel complete-data model learning criterion for coefficient optimization. Experiments on the healthcare data demonstrate the effectiveness and generalizability of our model and its promise in practical applications.

Chronic obstructive pulmonary disease (COPD) accounts for the highest rate of hospital readmissions and is the third leading cause of death in Canada, the United States and worldwide. Predicting COPD failure provides a prognostic warning of death or readmission, and is crucial to early intervention and decision-making. The aim of this study is to perform COPD failure prediction on longitudinal data. To address the inappropriate estimation of Cox hazard in current approaches, we propose a new representation of hazard to capture the relationship between survival probability and time-varying risk factors in a concise but effective way. To optimize model parameters, we design and maximize a new joint likelihood that comprises two components used to estimate survival status separately for failure and censored patients. A regularized optimization is performed on the joint likelihood to prevent overfitting arising from model learning. Our approach is applied to a real-life COPD data set and outperforms the current state-of-the-art prediction models in terms of the survival AUC, concordance index and Birer score metrics, this reveals that the great promise of our approach for clinical prediction.

BACKGROUND: Inferring the topology of gene regulatory networks (GRNs) from microarray gene expression data has many potential applications, such as identifying candidate drug targets and providing valuable insights into the biological processes. It remains a challenge due to the fact that the data is noisy and high dimensional, and there exists a large number of potential interactions. RESULTS: We introduce an ensemble gene regulatory network inference method PLSNET, which decomposes the GRN inference problem with p genes into p subproblems and solves each of the subproblems by using Partial least squares (PLS) based feature selection algorithm. Then, a statistical technique is used to refine the predictions in our method. The proposed method was evaluated on the DREAM4 and DREAM5 benchmark datasets and achieved higher accuracy than the winners of those competitions and other state-of-the-art GRN inference methods. CONCLUSIONS: Superior accuracy achieved on different benchmark datasets, including both in silico and in vivo networks, shows that PLSNET reaches state-of-the-art performance.

Privacy is an important concern in the society, and it has been a fundamental issue when to analyze and publish data involving human individual's sensitive information. Recently, the slicing method has been popularly used for privacy preservation in data publishing, because of its potential for preserving more data utility than others such as the generalization and bucketization approaches. However, in this paper, we show that the slicing method has disclosure risks for some absolute facts, which would help the adversary to find invalid records in the sliced microdata table, resulting in breach of individual privacy. To increase the privacy of published data in the sliced tables, a new method called value swapping is proposed in this work, aimed at decreasing the attribute disclosure risk for the absolute facts and ensuring the l-diverse slicing. By value swapping, the published table contains no invalid information such that the adversary cannot breach the individual privacy. Experimental results also show that the NEW method is able to keep more data utility than the existing slicing methods in a published microdata table. Copyright © 2016 John Wiley & Sons, Ltd.

The discovery of circle of friends has risen rapidly in recent years. Traditional methods are mainly based on social network analysis which relies heavily on self-report data, such that these methods have isolated successes with limited accuracy, breadth, and depth. In this paper, we propose a new method which combines clustering technique to automatically discover the circle of friends on mobile data. In our method, the circle of friends is modeled as non-overlapping subspace clusters on mobile data with a Vector Space Model (VSM) based representation, for which a new subspace clustering algorithm is proposed to mine the underlying friend-relationship. The experimental studies on real mobile data demonstrate the effectiveness of the new method, and the results show that our clustering algorithm achieves better performance than the existing clustering algorithms.

Naive Bayes (NB for short) is one of the popular methods for supervised classification in a knowledge management system. Currently, in many real-world applications, high-dimensional data pose a major challenge to conventional NB classifiers, due to noisy or redundant features and local relevance of these features to classes. In this paper, an automated feature weighting solution is proposed to result in a NB method effective in dealing with high-dimensional data. We first propose a locally weighted probability model, for Bayesian modeling in high-dimensional spaces, to implement a soft feature selection scheme. Then we propose an optimization algorithm to find the weights in linear time complexity, based on the Logitnormal priori distribution and the Maximum a Posteriori principle. Experimental studies show the effectiveness and suitability of the proposed model for high-dimensional data classification.

Clustering high-dimensional data is a major challenge due to the curse of dimensionality. To solve this problem, projective clustering has been defined as an extension to traditional clustering that attempts to find projected clusters in subsets of the dimensions of a data space. In this paper, a probability model is first proposed to describe projected clusters in high-dimensional data space. Then, we present a model-based algorithm for fuzzy projective clustering that discovers clusters with overlapping boundaries in various projected subspaces. The suitability of the proposal is demonstrated in an empirical study done with synthetic data set and some widely used real-world data set.

Clustering high dimensional data is a big challenge in data mining due to the curse of dimensionality. To solve this problem, projective clustering has been defined as an extension of traditional clustering that seeks to find projected clusters in subsets of dimensions of a data space. In this paper, the problem of modeling projected clusters is first discussed, and an extended Gaussian model is proposed. Second, a general objective criterion used with k-means type projective clustering is presented based on the model. Finally, the expressions to learn model parameters are derived and then used in a new algorithm named FPC to perform fuzzy clustering on high dimensional data. The experimental results on document clustering show the effectiveness of the proposed clustering model.