Medical images exhibit latent anatomical groupings, such as organs, tissues, and pathological regions, that standard Vision Transformers (ViTs) fail to exploit. While recent work like SBM-Transformer attempts to incorporate such structures through stochastic binary masking, they suffer from non-differentiability, training instability, and the inability to model complex community structure. We present DCMM-Transformer, a novel ViT architecture for medical image analysis that incorporates a Degree-Corrected Mixed-Membership (DCMM) model as an additive bias in self-attention. Unlike prior approaches that rely on multiplicative masking and binary sampling, our method introduces community structure and degree heterogeneity in a fully differentiable and interpretable manner. Comprehensive experiments across diverse medical imaging datasets, including brain, chest, breast, and ocular modalities, demonstrate the superior performance and generalizability of the proposed approach. Furthermore, the learned group structure and structured attention modulation substantially enhance interpretability by yielding attention maps that are anatomically meaningful and semantically coherent.

The emergence of networked electrical machines has increased susceptibility to anomalies, including cyber-attack and physical faults, potentially leading to significant operational disruptions. In this article, we propose an online adaptive anomaly detection algorithm, adaptive enveloped singular spectrum transformation (AdaESST), which aims to identify hard-to-detect anomalies effectively. AdaESST first extracts informative components of signals by embedding the waveform data into subspaces using singular value decomposition, and then calculates anomalous score based on the subspace distance between two subsequence time series. AdaESST outperforms traditional detection methods by its capacity to adjust to new operational scenarios, thereby offering persistent protection in dynamic industrial environments. Throughout all numerical experiments simulating real-world industrial conditions, AdaESST exhibits high detection accuracy in monitoring motor and point of common coupling (PCC) currents, demonstrating its capability to safeguard against sophisticated anomalies. The detection accuracy for PCC currents is on par with that for motor currents. In essence, AdaESST has the potential to reduce the requirements for sensors, thereby lowering maintenance costs while maintaining high data integrity and security. The work contributes to enhancing the security of networked electrical machines, presenting a resilient and cost-efficient strategy in the face of emerging anomalies.

Knowledge distillation (KD) has been widely used for model compression and deployment acceleration. Nonetheless, the statistical insight of the remarkable performance of KD remains elusive, and methods for evaluating the uncertainty of the distilled model/student model are lacking. To address these issues, we establish a close connection between KD and a Bayesian model. In particular, we develop an innovative method named Bayesian Knowledge Distillation (BKD) to provide a transparent interpretation of the working mechanism of KD, and a suite of Bayesian inference tools for the uncertainty quantification of the student model. In BKD, the regularization imposed by the teacher model in KD is formulated as a teacher-informed prior for the student model’s parameters. Consequently, we establish the equivalence between minimizing the KD loss and estimating the posterior mode in BKD. Efficient Bayesian inference algorithms are developed based on the stochastic gradient Langevin Monte Carlo and examined with extensive experiments on uncertainty ranking and credible intervals construction for predicted class probabilities.

Multimodal integration combines information from different sources or modalities to gain a more comprehensive understanding of a phenomenon. The challenges in multi-omics data analysis lie in the complexity, high dimensionality, and heterogeneity of the data, which demands sophisticated computational tools and visualization methods for proper interpretation and visualization of multi-omics data. In this paper, we propose a novel method, termed Orthogonal Multimodality Integration and Clustering (OMIC), for analyzing CITE-seq. Our approach enables researchers to integrate multiple sources of information while accounting for the dependence among them. We demonstrate the effectiveness of our approach using CITE-seq data sets for cell clustering. Our results show that our approach outperforms existing methods in terms of accuracy, computational efficiency, and interpretability. We conclude that our proposed OMIC method provides a powerful tool for multimodal data analysis that greatly improves the feasibility and reliability of integrated data.

Rapid and accurate detection and localization of electronic disturbances simultaneously are important for preventing its potential damages and determining potential remedies. The existing anomaly detection methods are severely limited by the low accuracy, expensive computational cost, and the need for highly trained personnel. There is an urgent need for a scalable online algorithm for the in-field analysis of large-scale power electronics networks. In this article, we propose a fast and accurate algorithm for anomaly detection and localization of power electronics networks: the stratified colored-node graph (CONGO). This algorithm hierarchically models the change of correlated waveforms and then correlated sensors using the CONGO. By aggregating the change of each sensor with its neighbors’ inputs, we can spontaneously identify and localize the anomaly that cannot be detected by data collected from a single sensor. As our proposed method only focuses on the changes within a short time frame, it is highly computational efficient and only needs small data storage. Thus, our method is ideal for online and reliable anomaly detection and localization of large-scale power electronic networks. Compared to the existing anomaly detection methods, our method is entirely data driven without training data, highly accurate and reliable for wide-spectrum anomalies detection, and more importantly, capable of both detection and localization. Thus, it is ideal for the in-field deployment for large-scale power electronic networks. As illustrated by a distributed energy resources (DERs) power grid with 37-node, our method can effectively detect and localize various cyber and physical attacks.

Sufficient dimension reduction is used pervasively as a supervised dimension reduction approach. Most existing sufficient dimension reduction methods are developed for data with a continuous response and may have an unsatisfactory performance for the categorical response, especially for the binary-response. To address this issue, we propose a novel estimation method of sufficient dimension reduction subspace (SDR subspace) using optimal transport. The proposed method, named principal optimal transport direction (POTD), estimates the basis of the SDR subspace using the principal directions of the optimal transport coupling between the data respecting different response categories. The proposed method also reveals the relationship among three seemingly irrelevant topics, i.e., sufficient dimension reduction, support vector machine, and optimal transport. We study the asymptotic properties of POTD and show that in the cases when the class labels contain no error, POTD estimates the SDR subspace exclusively. Empirical studies show POTD outperforms most of the state-of-the-art linear dimension reduction methods.

Testing the hypothesis of parallelism is a fundamental statistical problem arising from many applied sciences. In this paper, we develop a nonparametric parallelism test for inferring whether the trends are parallel in treatment and control groups. In particular, the proposed nonparametric parallelism test is a Wald type test based on a smoothing spline ANOVA (SSANOVA) model which can characterize the complex patterns of the data. We derive that the asymptotic null distribution of the test statistic is a Chi-square distribution, unveiling a new version of Wilks phenomenon. Notably, we establish the minimax sharp lower bound of the distinguishable rate for the nonparametric parallelism test by using the information theory, and further prove that the proposed test is minimax optimal. Simulation studies are conducted to investigate the empirical performance of the proposed test. DNA methylation and neuroimaging studies are presented to illustrate potential applications of the test. The software is available at https://github.com/BioAlgs/Parallelism.

Estimating the dependence structure of multidimensional time series data in real-time is challenging. With large volumes of streaming data, the problem becomes more difficult when the multidimensional data are collected asynchronously across distributed nodes, which motivates us to sample representative data points from streams. We propose a leverage score sampling (LSS) method for efficient online inference of the streaming vector autoregressive (VAR) model. We define the leverage score for the streaming VAR model so that the LSS method selects informative data points in real-time with statistical guarantees of parameter estimation efficiency. Moreover, our LSS method can be directly deployed in an asynchronous decentralized environment, e.g., a sensor network without a fusion center, and produce asynchronous consensus online parameter estimation over time. By exploiting the temporal dependence structure of the VAR model, the LSS method selects samples independently on each dimension and thus is able to update the estimation asynchronously. We illustrate the effectiveness of the LSS method in synthetic, gas sensor and seismic datasets.

This paper studies the estimation of large-scale optimal transport maps (OTM), which is a well known challenging problem owing to the curse of dimensionality. Existing literature approximates the large-scale OTM by a series of one-dimensional OTM problems through iterative random projection. Such methods, however, suffer from slow or none convergence in practice due to the nature of randomly selected projection directions. Instead, we propose an estimation method of large-scale OTM by combining the idea of projection pursuit regression and sufficient dimension reduction. The proposed method, named projection pursuit Monge map (PPMM), adaptively selects the most informative'' projection direction in each iteration. We theoretically show the proposed dimension reduction method can consistently estimate the mostinformative'' projection direction in each iteration. Furthermore, the PPMM algorithm weakly convergences to the target large-scale OTM in a reasonable number of steps. Empirically, PPMM is computationally easy and converges fast. We assess its finite sample performance through the applications of Wasserstein distance estimation and generative models.

BACKGROUND: Accurate identification of differentially expressed (DE) genes in time course RNA-Seq data is crucial for understanding the dynamics of transcriptional regulatory network. However, most of the available methods treat gene expressions at different time points as replicates and test the significance of the mean expression difference between treatments or conditions irrespective of time. They thus fail to identify many DE genes with different profiles across time. In this article, we propose a negative binomial mixed-effect model (NBMM) to identify DE genes in time course RNA-Seq data. In the NBMM, mean gene expression is characterized by a fixed effect, and time dependency is described by random effects. The NBMM is very flexible and can be fitted to both unreplicated and replicated time course RNA-Seq data via a penalized likelihood method. By comparing gene expression profiles over time, we further classify the DE genes into two subtypes to enhance the understanding of expression dynamics. A significance test for detecting DE genes is derived using a Kullback-Leibler distance ratio. Additionally, a significance test for gene sets is developed using a gene set score. RESULTS: Simulation analysis shows that the NBMM outperforms currently available methods for detecting DE genes and gene sets. Moreover, our real data analysis of fruit fly developmental time course RNA-Seq data demonstrates the NBMM identifies biologically relevant genes which are well justified by gene ontology analysis. CONCLUSIONS: The proposed method is powerful and efficient to detect biologically relevant DE genes and gene sets in time course RNA-Seq data.

MOTIVATION: Identification of transcription factor binding motifs (TFBMs) is a crucial first step towards the understanding of regulatory circuitries controlling the expression of genes. In this paper, we propose a novel procedure called regularized sliced inverse regression (RSIR) for identifying TFBMs. RSIR follows a recent trend to combine information contained in both gene expression measurements and genes' promoter sequences. Compared with existing methods, RSIR is efficient in computation, very stable for data with high dimensionality and high collinearity, and improves motif detection sensitivities and specificities by avoiding inappropriate model specification. RESULTS: We compare RSIR with SIR and stepwise regression based on simulated data and find that RSIR has a lower false positive rate. We also demonstrate an excellent performance of RSIR by applying it to the yeast amino acid starvation data and cell cycle data. AVAILABILITY: Matlab programs are available upon request from the authors.