Global Information Integration Research Articles

Digital Diplomacy and Its Impacts on Uzbekistan’s International Relations

This paper investigates the role of digital technologies in Uzbekistan’s diplomatic activities, emphasizing innovation strategies, project support, personnel training, and the development of digital platforms. It explores the concept of digital diplomacy, focusing on the use of digital media in diplomacy and how countries utilize these media to achieve their foreign policy objectives. The study delves into the opportunities and challenges presented by digital media for diplomatic activities, highlighting the necessity for countries to actively engage in digital diplomacy and keep abreast of emerging trends in this field. The article emphasizes the importance of digital diplomacy not only as a new tool but also as a factor that transforms general diplomatic strategies. By examining Uzbekistan’s efforts in integrating digital technologies into its diplomatic strategies, this research aims to contribute to a deeper understanding of how digitalization is shaping contemporary diplomacy. It underscores the importance of technological advancement and digital literacy for diplomatic personnel to navigate the complex landscape of global information integration. Through a comprehensive analysis, the paper presents evidence supporting the active implementation of digital diplomacy by nations and discusses the implications for future diplomatic practices. The findings reveal that digital diplomacy not only enhances the efficiency and reach of diplomatic efforts but also poses certain challenges that need to be addressed to maximize its potential. Overall, the research provides valuable insights into the dynamic intersection of digital innovation and diplomatic strategy in the context of Uzbekistan’s international relations.

Evolving brain network dynamics in early childhood: Insights from modular graph metrics

Modular dynamic graph theory metrics effectively capture the patterns of dynamic information interaction during human brain development. While existing research has employed modular algorithms to examine the overall impact of dynamic changes in community structure throughout development, there is a notable gap in understanding the cross-community dynamic changes within different functional networks during early childhood and their potential contributions to the efficiency of brain information transmission. This study seeks to address this gap by tracing the trajectories of cross-community structural changes within early childhood functional networks and modeling their contributions to information transmission efficiency. We analyzed 194 functional imaging scans from 83 children aged 2 to 8 years, who participated in passive viewing functional magnetic resonance imaging sessions. Utilizing sliding windows and modular algorithms, we evaluated three spatiotemporal metrics—temporal flexibility, spatiotemporal diversity, and within-community spatiotemporal diversity—and four centrality metrics: within-community degree centrality, eigenvector centrality, between-community degree centrality, and between-community eigenvector centrality. Mixed-effects linear models revealed significant age-related increases in the temporal flexibility of the default mode network (DMN), executive control network (ECN), and salience network (SN), indicating frequent adjustments in community structure within these networks during early childhood. Additionally, the spatiotemporal diversity of the SN also displayed significant age-related increases, highlighting its broad pattern of cross-community dynamic interactions. Conversely, within-community spatiotemporal diversity in the language network exhibited significant age-related decreases, reflecting the network's gradual functional specialization. Furthermore, our findings indicated significant age-related increases in between-community degree centrality across the DMN, ECN, SN, language network, and dorsal attention network, while between-community eigenvector centrality also increased significantly for the DMN, ECN, and SN. However, within-community eigenvector centrality remained stable across all functional networks during early childhood. These results suggest that while centrality of cross-community interactions in early childhood functional networks increases, centrality within communities remains stable. Finally, mediation analysis was conducted to explore the relationships between age, brain dynamic graph metrics, and both global and local efficiency based on community structure. The results indicated that the dynamic graph metrics of the SN primarily mediated the relationship between age and the decrease in global efficiency, while those of the DMN, language network, ECN, dorsal attention network, and SN primarily mediated the relationship between age and the increase in local efficiency. This pattern suggests a developmental trajectory in early childhood from global information integration to local information segregation, with the SN playing a pivotal role in this transformation. This study provides novel insights into the mechanisms by which early childhood brain functional development impacts information transmission efficiency through cross-community adjustments in functional networks.

A statistical analysis method for probability distributions in Erdös-Rényi random networks with preferential cutting-rewiring operation.

The study of specific physiological processes from the perspective of network physiology has gained recent attention. Modeling the global information integration among the separated functionalized modules in structural and functional brain networks is a central problem. In this article, the preferentially cutting-rewiring operation (PCRO) is introduced to approximatively describe the above physiological process, which consists of the cutting procedure and the rewiring procedure with specific preferential constraints. By applying the PCRO on the classical Erdös-Rényi random network (ERRN), three types of isolated nodes are generated, based on which the common leaves (CLs) are formed between the two hubs. This makes the initially homogeneous ERRN experience drastic changes and become heterogeneous. Importantly, a statistical analysis method is proposed to theoretically analyze the statistical properties of an ERRN with a PCRO. Specifically, the probability distributions of these three types of isolated nodes are derived, based on which the probability distribution of the CLs can be obtained easily. Furthermore, the validity and universality of our statistical analysis method have been confirmed in numerical experiments. Our contributions may shed light on a new perspective in the interdisciplinary field of complexity science and biological science and would be of great and general interest to network physiology.

RS Transformer: A Two-Stage Region Proposal Using Swin Transformer for Few-Shot Pest Detection in Automated Agricultural Monitoring Systems

Agriculture is pivotal in national economies, with pest classification significantly influencing food quality and quantity. In recent years, pest classification methods based on deep learning have made progress. However, there are two problems with these methods. One is that there are few multi-scale pest detection algorithms, and they often lack effective global information integration and discriminative feature representation. The other is the lack of high-quality agricultural pest datasets, leading to insufficient training samples. To overcome these two limitations, we propose two methods called RS Transformer (a two-stage region proposal using Swin Transformer) and the Randomly Generated Stable Diffusion Dataset (RGSDD). Firstly, we found that the diffusion model can generate high-resolution images, so we developed a training strategy called the RGSDD, which was used to generate agricultural pest images and was mixed with real datasets for training. Secondly, RS Transformer uses Swin Transformer as the backbone to enhance the ability to extract global features, while reducing the computational burden of the previous Transformer. Finally, we added a region proposal network and ROI Align to form a two-stage training mode. The experimental results on the datasets show that RS Transformer has a better performance than the other models do. The RGSDD helps to improve the training accuracy of the model. Compared with methods of the same type, RS Transformer achieves up to 4.62% of improvement.

Global Filter: Augmenting Images to Support Seeing the “Big Picture” for People with Local Interference

Some neurodivergent people prioritize visual details over the “big picture”. While excellent attention to detail has many advantages, some contexts require the rapid integration of global and local information. A local processing style can be so strong that local details interfere with the fluid integration of global information required for processing of information rapidly displayed on user interfaces. This disconnect between context of an interaction and processing style can be termed local interference. Personalization of visual stimuli can promote a more accessible computing experience. We describe how technological interventions can support shifting of visual attention from local to global features to make them more accessible. We present two empirical studies. One study with one autistic adult revealed a significant shift in eye gaze fixation, and the other study with 20 autistic children revealed filters that visually emphasize primary aspects encouraged more global comments about the image content.

Ensemble of Learner for Network Intrusion Detection System

The uses of the internet have improved drastically for online communication and working from home. Data sharing and integration of global information bring network security risks. To protect private data and information, network security is becoming a very important research topic. An intrusion detection system is generally used as safe operational tool. It excellently detects and prevents intruders in a network by issuing a warning before the attack is launched in a network. An ensemble technique is extensively used to employ intrusion detection systems. In this paper, the stacking method of the ensemble has been proposed for the intrusion detection system. Three base classifiers have been stacked using the Meta classifier. Multilayer Perceptron, Ripple Down Rule learner, and RepTree Decision Tree have been used as Base Classifiers. These Base Classifiers are stacked using Logistic Regression with ridge estimator Meta Classifier. These learners have been trained and tested using the NSL dataset. A genetic algorithm has been applied for choosing relevant and most corrected features which helped in reducing the dimension of the dataset. Experimental results demonstrate that the proposed stacked classifier gives accuracies of 79.36%, 99.72%, and 99.64% on a test, train dataset, and cross-validation respectively. It is observed that the proposed stacked classifiers do better than existing hybrid intrusion detection systems.

A Pose-Aware Global Representation Network for Human Parsing

Many recognition tasks including image/video classification, segmentation and object detection can be improved by the integration of global information. Although global information may be better represented in some recognition tasks than the others, it is worth exploring how global information from related tasks can be effectively used to improve the performance of a target task. The task of pose estimation predicts the locations of human joints, thus providing global information about the human body. In this paper, we propose a pose-aware global representation network model (PAGRnet) that exploits global information from pose estimation to enhance feature learning in human parsing. In our PAGRnet model, a novel learning module with three integrated parts is used to learn global information. The first part generates a global joint representation, while the second part learns the relationship between the pixels and joints. By integrating the global joint representation with the pixel-joint relationship, the resulting pose-aware global representation is augmented for the parsing task. Our experimental results show competitive performance of our method on the LIP, the Pascal-person-part and the ATR datasets, with reduced computation costs in comparison to other proposals of global information fusion. We also demonstrate the advantages of our feature fusion model over concatenation, pixel-wise and channel-wise relation models.

Blended Glial Cell’s Spiking Neural Network

Spiking Neural Networks (SNNs), the third generation of artificial neural networks, have been widely employed. However, the realization of advanced artificial intelligence is challenging due to the dearth of efficient spatiotemporal information integration models. Inspired by brain neuroscientists, this paper proposes a novel spiking neural network - Blended Glial Cell’s Spiking Neural Network (BGSNN). BGSNN introduces glial cells as spatiotemporal information processing units based on neurons and synapses, and also provides four new network dynamics connection models which extend the information processing dimension, enhance the network global information integration in the spatiotemporal domain, as well as the plasticity of neurons and synapses. In this paper, a BGSNN application - Sudoku solver is designed and implemented on the "WenTian" neuromorphic prototype. On the Easybrain dataset, the BGSNN solver achieves 100% accuracy, outperforming the same structure SNN solver by 97% at the Evil difficulty level, and has faster converges speed compared with the SOTA Sudoku solver LSGA. On the kaggle dataset, the BGSNN solver achieves over 99.99% accuracy, outperforming the publicly available optimal DNN solver under this dataset by 3.82%. In addition, BGSNN exhibits good parallelism and sparsity, decreasing computation by at least 92.9% compared to serial solvers and reducing sparsity by 88% compared to the equal fully dense DNN. BGSNN improves the expression, feedback, and regulation capabilities of neural networks while maintaining the advantages of SNN parallel sparsity, making it simpler to implement advanced artificial intelligence.

Normalization effect of levodopa on hierarchical brain function in Parkinson's disease.

Hierarchical brain organization, in which the rich club and diverse club situate in core position, is critical for global information integration in the human brain network. Parkinson’s disease (PD), a common movement disorder, has been conceptualized as a network disorder. Levodopa is an effective treatment for PD. Whether there is a functional divergence in the hierarchical brain system under PD pathology, and how this divergence is regulated by immediate levodopa therapy, remains unknown. We constructed a functional network in 61 PD patients and 89 normal controls and applied graph theoretical analyses to examine the neural mechanism of levodopa short response from the perspective of brain hierarchical configuration. The results revealed the following: (a) PD patients exhibited disrupted function within rich-club organization, while the diverse club preserved function, indicating a differentiated brain topological organization in PD. (b) Along the rich-club derivate hierarchical system, PD patients showed impaired network properties within rich-club and feeder subnetworks, and decreased nodal degree centrality in rich-club and feeder nodes, along with increased nodal degree in peripheral nodes, suggesting distinct functional patterns in different types of nodes. And (c) levodopa could normalize the abnormal network architecture of the rich-club system. This study provides evidence for levodopa effects on the hierarchical brain system with divergent functions.

Frequency-Resolved Connectome Hubs and Their Test-Retest Reliability in the Resting Human Brain.

Functional hubs with disproportionately extensive connectivities play a crucial role in global information integration in human brain networks. However, most resting-state functional magnetic resonance imaging (R-fMRI) studies have identified functional hubs by examining spontaneous fluctuations of the blood oxygen level-dependent signal within a typical low-frequency band (e.g., 0.01-0.08 Hz or 0.01-0.1 Hz). Little is known about how the spatial distributions of functional hubs depend on frequency bands of interest. Here, we used repeatedly measured R-fMRI data from 53 healthy young adults and a degree centrality analysis to identify voxelwise frequency-resolved functional hubs and further examined their test-retest reliability across two sessions. We showed that a wide-range frequency band (0.01-0.24 Hz) accessible with a typical sampling rate (fsample = 0.5 Hz) could be classified into three frequency bands with distinct patterns, namely, low-frequency (LF, 0.01-0.06 Hz), middle-frequency (MF, 0.06-0.16 Hz), and high-frequency (HF, 0.16-0.24 Hz) bands. The functional hubs were mainly located in the medial and lateral frontal and parietal cortices in the LF band, and in the medial prefrontal cortex, superior temporal gyrus, parahippocampal gyrus, amygdala, and several cerebellar regions in the MF and HF bands. These hub regions exhibited fair to good test-retest reliability, regardless of the frequency band. The presence of the three frequency bands was well replicated using an independent R-fMRI dataset from 45 healthy young adults. Our findings demonstrate reliable frequency-resolved functional connectivity hubs in three categories, thus providing insights into the frequency-specific connectome organization in healthy and disordered brains.

Inheritance Attention Matrix-Based Universal Adversarial Perturbations on Vision Transformers

Universal Adversarial Perturbations (UAPs), which is type of image-agnostic adversarial attack, has been deeply investigated for Convolutional Neural Networks due to its high efficiency. On the other hand, as an architecture based on self-attention mechanism, Vision Transformers (ViTs) have boomed and been widely applied to solve various computer vision problems since most recent years. In this letter, we delve into the robustness of ViTs against universal adversarial attack and propose an inheritance attention matrix-based UAPs (IAM-UAP). Specifically, we introduce the inheritance attention weight matrix (IAM), which represents the integration of global information of the input patch sequence. Further, we propose a perturbation optimization objective based on IAM to confuse the global information integration of ViTs. The empirical results confirm the attacking capability of IAM-UAP on ViTs with a moderate attacking rate. In addition, we also disclose that the patch size of ViTs is a latent factor influencing the robustness against the universal attack.

Long-range connections are more severely damaged and relevant for cognition in multiple sclerosis.

An efficient network such as the human brain features a combination of global integration of information, driven by long-range connections, and local processing involving short-range connections. Whether these connections are equally damaged in multiple sclerosis is unknown, as is their relevance for cognitive impairment and brain function. Therefore, we cross-sectionally investigated the association between damage to short- and long-range connections with structural network efficiency, the functional connectome and cognition. From the Amsterdam multiple sclerosis cohort, 133 patients (age = 54.2 ± 9.6) with long-standing multiple sclerosis and 48 healthy controls (age = 50.8 ± 7.0) with neuropsychological testing and MRI were included. Structural connectivity was estimated from diffusion tensor images using probabilistic tractography (MRtrix 3.0) between pairs of brain regions. Structural connections were divided into short- (length < quartile 1) and long-range (length > quartile 3) connections, based on the mean distribution of tract lengths in healthy controls. To determine the severity of damage within these connections, (i) fractional anisotropy as a measure for integrity; (ii) total number of fibres; and (iii) percentage of tract affected by lesions were computed for each connecting tract and averaged for short- and long-range connections separately. To investigate the impact of damage in these connections for structural network efficiency, global efficiency was computed. Additionally, resting-state functional connectivity was computed between each pair of brain regions, after artefact removal with FMRIB's ICA-based X-noiseifier. The functional connectivity similarity index was computed by correlating individual functional connectivity matrices with an average healthy control connectivity matrix. Our results showed that the structural network had a reduced efficiency and integrity in multiple sclerosis relative to healthy controls (both P < 0.05). The long-range connections showed the largest reduction in fractional anisotropy (z = -1.03, P < 0.001) and total number of fibres (z = -0.44, P < 0.01), whereas in the short-range connections only fractional anisotropy was affected (z = -0.34, P = 0.03). Long-range connections also demonstrated a higher percentage of tract affected by lesions than short-range connections, independent of tract length (P < 0.001). Damage to long-range connections was more strongly related to structural network efficiency and cognition (fractional anisotropy: r = 0.329 and r = 0.447. number of fibres r = 0.321 and r = 0.278. and percentage of lesions: r = -0.219; r = -0.426, respectively) than damage to short-range connections. Only damage to long-distance connections correlated with a more abnormal functional network (fractional anisotropy: r = 0.226). Our findings indicate that long-range connections are more severely affected by multiple sclerosis-specific damage than short-range connections. Moreover compared to short-range connections, damage to long-range connections better explains network efficiency and cognition.

Using Partial Directed Coherence to Study Alpha-Band Effective Brain Networks during a Visuospatial Attention Task.

Previous studies have shown that the neural mechanisms underlying visual spatial attention rely on top-down control information from the frontal and parietal cortexes, which ultimately amplifies sensory processing of stimulus occurred at the attended location relative to those at unattended location. However, the modulations of effective brain networks in response to stimulus at attended and unattended location are not yet clear. In present study, we collected event-related potentials (ERPs) from 15 subjects during a visual spatial attention task, and a partial directed coherence (PDC) method was used to construct alpha-band effective brain networks of two conditions (targets at attended and nontargets at unattended location). Flow gain mapping, effective connectivity pattern, and graph measures including clustering coefficient (C), characteristic path length (L), global efficiency (Eglobal), and local efficiency (Elocal) were compared between two conditions. Flow gain mapping showed that the frontal region seemed to serve as the main source of information transmission in response to targets at attended location while the parietal region served as the main source in nontarget condition. Effective connectivity pattern indicated that in response to targets, there existed obvious top-down connections from the frontal, temporal, and parietal cortexes to the visual cortex compared with in response to nontargets. Graph theory analysis was used to quantify the topographical properties of the brain networks, and results revealed that in response to targets, the brain networks were characterized by significantly smaller characteristic path length and larger global efficiency than in response to nontargets. Our findings suggested that smaller characteristic path length and larger global efficiency could facilitate global integration of information and provide a substrate for more efficient perceptual processing of targets at attended location compared with processing of nontargets at ignored location, which revealed the neural mechanisms underlying visual spatial attention from the perspective of effective brain networks and graph theory for the first time and opened new vistas to interpret a cognitive process.

PAN: Projective Adversarial Network for Medical Image Segmentation.

Adversarial learning has been proven to be effective for capturing long-range and high-level label consistencies in semantic segmentation. Unique to medical imaging, capturing 3D semantics in an effective yet computationally efficient way remains an open problem. In this study, we address this computational burden by proposing a novel projective adversarial network, called PAN, which incorporates high-level 3D information through 2D projections. Furthermore, we introduce an attention module into our framework that helps for a selective integration of global information directly from our segmentor to our adversarial network. For the clinical application we chose pancreas segmentation from CT scans. Our proposed framework achieved state-of-the-art performance without adding to the complexity of the segmentor.

Higher levels of trait emotional awareness are associated with more efficient global information integration throughout the brain: a graph-theoretic analysis of resting state functional connectivity.

Previous studies have suggested that trait differences in emotional awareness (tEA) are clinically relevant, and associated with differences in neural structure/function. While multiple leading theories suggest that conscious awareness requires widespread information integration across the brain, no study has yet tested the hypothesis that higher tEA corresponds to more efficient brain-wide information exchange. Twenty-six healthy volunteers (13 females) underwent a resting state functional magnetic resonance imaging scan, and completed the Levels of Emotional Awareness Scale (LEAS; a measure of tEA) and the Wechsler Abbreviated Scale of Intelligence (WASI-II; a measure of general intelligence quotient [IQ]). Using a whole-brain (functionally defined) region of interest (ROI) atlas, we computed several graph theory metrics to assess the efficiency of brain-wide information exchange. After statistically controlling for differences in age, gender and IQ, we first observed a significant relationship between higher LEAS scores and greater average degree (i.e. overall whole-brain network density). When controlling for average degree, we found that higher LEAS scores were also associated with shorter average path lengths across the collective network of all included ROIs. These results jointly suggest that individuals with higher tEA display more efficient global information exchange throughout the brain. This is consistent with the idea that conscious awareness requires global accessibility of represented information.

Finding influential nodes for integration in brain networks using optimal percolation theory

Global integration of information in the brain results from complex interactions of segregated brain networks. Identifying the most influential neuronal populations that efficiently bind these networks is a fundamental problem of systems neuroscience. Here, we apply optimal percolation theory and pharmacogenetic interventions in vivo to predict and subsequently target nodes that are essential for global integration of a memory network in rodents. The theory predicts that integration in the memory network is mediated by a set of low-degree nodes located in the nucleus accumbens. This result is confirmed with pharmacogenetic inactivation of the nucleus accumbens, which eliminates the formation of the memory network, while inactivations of other brain areas leave the network intact. Thus, optimal percolation theory predicts essential nodes in brain networks. This could be used to identify targets of interventions to modulate brain function.

Role of the ATLAS Grid Information System (AGIS) in Distributed Data Analysis and Simulation

In modern high-energy physics experiments, particular attention is paid to the global integration of information and computing resources into a unified system for efficient storage and processing of experimental data. Annually, the ATLAS experiment performed at the Large Hadron Collider at the European Organization for Nuclear Research (CERN) produces tens of petabytes raw data from the recording electronics and several petabytes of data from the simulation system. For processing and storage of such super-large volumes of data, the computing model of the ATLAS experiment is based on heterogeneous geographically distributed computing environment, which includes the worldwide LHC computing grid (WLCG) infrastructure and is able to meet the requirements of the experiment for processing huge data sets and provide a high degree of their accessibility (hundreds of petabytes). The paper considers the ATLAS grid information system (AGIS) used by the ATLAS collaboration to describe the topology and resources of the computing infrastructure, to configure and connect the high-level software systems of computer centers, to describe and store all possible parameters, control, configuration, and other auxiliary information required for the effective operation of the ATLAS distributed computing applications and services. The role of the AGIS system in the development of a unified description of the computing resources provided by grid sites, supercomputer centers, and cloud computing into a consistent information model for the ATLAS experiment is outlined. This approach has allowed the collaboration to extend the computing capabilities of the WLCG project and integrate the supercomputers and cloud computing platforms into the software components of the production and distributed analysis workload management system (PanDA, ATLAS).

Angular default mode network connectivity across working memory load.

Initially identified during no-task, baseline conditions, it has now been suggested that the default mode network (DMN) engages during a variety of working memory paradigms through its flexible interactions with other large-scale brain networks. Nevertheless, its contribution to whole-brain connectivity dynamics across increasing working memory load has not been explicitly assessed. The aim of our study was to determine which DMN hubs relate to working memory task performance during an fMRI-based n-back paradigm with parametric increases in difficulty. Using a voxel-wise metric, termed the intrinsic connectivity contrast (ICC), we found that the bilateral angular gyri (core DMN hubs) displayed the greatest change in global connectivity across three levels of n-back task load. Subsequent seed-based functional connectivity analysis revealed that the angular DMN regions robustly interact with other large-scale brain networks, suggesting a potential involvement in the global integration of information. Further support for this hypothesis comes from the significant correlations we found between angular gyri connectivity and reaction times to correct responses. The implication from our study is that the DMN is actively involved during the n-back task and thus plays an important role related to working memory, with its core angular regions contributing to the changes in global brain connectivity in response to increasing environmental demands. Hum Brain Mapp 38:41-52, 2017. © 2016 Wiley Periodicals, Inc.

Disrupted structural and functional rich club organization of the brain connectome in patients with generalized tonic-clonic seizure.

Network studies have demonstrated that a small set of highly connected regions may play a central role in global information integration, together forming a rich club organization. Given that generalized tonic-clonic seizure (GTCS) has been conceptualized as a network disorder, we hypothesized that the rich club disturbances may be related to the network abnormalities of GTCS. Here, we used graph theoretical analysis to investigate the rich club organization of both structural and functional connectome in patients with GTCS (n = 50) and healthy controls (n = 60). We further measured the level of global efficiency and clustering in rich club and non-rich club organization. We, respectively, identified a small number of highly connected hubs as rich club organization from structural and functional networks. Patients were found to exhibit significantly reduced rich club connectivity among the central hubs. Meanwhile, both structural and functional network showed changed levels of global efficiency and clustering of rich club organization in GTCS. Furthermore, in patients, lower levels of rich club connectivity were found to be correlated with longer duration of illness and seizure frequency. Together, these findings suggest that GTCS is characterized by a selective disruption of rich club organization due to the long-term injurious effects of epileptic actions on the central hub regions, which potentially contribute to a reduced level of brain integration capacity among different functional domains and an added effect of illness on a preexisting vulnerability. Our findings emphasize a central role for abnormal rich club organization in the pathophysiological mechanism underlying GTCS. Hum Brain Mapp 37:4487-4499, 2016. © 2016 Wiley Periodicals, Inc.

DinTucker: Scaling Up Gaussian Process Models on Large Multidimensional Arrays

Tensor decomposition methods are effective tools for modelling multidimensional array data (i.e., tensors). Among them, nonparametric Bayesian models, such as Infinite Tucker Decomposition (InfTucker), are more powerful than multilinear factorization approaches, including Tucker and PARAFAC, and usually achieve better predictive performance. However, they are difficult to handle massive data due to a prohibitively high training cost. To address this limitation, we propose Distributed infinite Tucker (DinTucker), a new hierarchical Bayesian model that enables local learning of InfTucker on subarrays and global information integration from local results. We further develop a distributed stochastic gradient descent algorithm, coupled with variational inference for model estimation. In addition, the connection between DinTucker and InfTucker is revealed in terms of model evidence. Experiments demonstrate that DinTucker maintains the predictive accuracy of InfTucker and is scalable on massive data: On multidimensional arrays with billions of elements from two real-world applications, DinTucker achieves significantly higher prediction accuracy with less training time, compared with the state-of-the-art large-scale tensor decomposition method, GigaTensor.