4 September 2017 in IEEE Transactions on Industrial Informatics
Multiple clustering analysis has the clear advantages to discover latent data pattern in big data from different views, so it has tremendous practical values in automation industries. However, most of current algorithms are difficult to group heterogeneous data to multiple clusterings according to the requirements of different applications. This paper presents a flexibly multiple clustering analytic and service framework, and a novel tensor-based multiple clusterings (TMC) approach. Heterogeneous data objects in Cyber-Physical-Social Systems (CPSS) are first represented as low-order tensors and a weight tensor construction approach is proposed to measure the importance of attributes combinations in heterogeneous feature spaces. Then, a selective weighted tensor distance (SWTD) is explored to cluster tensorized data objects for different applications. This paper, through real world applications such as a gas processing system and a smart bike maintenance system, illustrates TMC and evaluates its clustering performance. Experimental results reveal TMC can obtain higher-quality clustering results but with lower redundancies to meet different equirements of applications in automation systems.
20 April 2021 in IEEE Transactions on Industrial Informatics
Data in modern industrial applications and data science present multidimensional progressively, the dimension and the structural complexity of these data are becoming extremely high, which renders existing data analysis methods and machine learning algorithms inadequate to the extent. In addition, high-dimensional data in actual scenarios often share some common latent components and patterns, it is necessary and significant to analyze such data in an associative manner, rather than treating them independently. Considering the problem of data islands and data privacy that is prevalent in the industry. In this article, we propose the first joint high-order orthogonal iterative (J-HOOI) algorithm for simultaneous tensor decomposition and federated tensor decomposition (FTD) model for feature extraction and dimension reduction of high-dimensional industrial data under the federated learning framework. Moreover, we also develop a secure federated computation process based on the J-HOOI method. Using this method, multiple participants iteratively calculate the local factor matrices and transfer the local information to the parameter server, which aggregates the local information to generate the globally updated factor matrices. Finally, each client generates globally compressed features by projecting local data onto these common potential spaces. We have demonstrated with real-world industrial datasets that our approach is similar to a centralized training model in decomposition accuracy and classification accuracy while respecting privacy.
15 March 2018 in IEEE Transactions on Industrial Informatics
To uncover latent data structure in big data, a high-order clustering algorithm by fast search and find of density peaks has emerged recently, and will bring great application values in Industrial Internet-of-Things data management and analysis. With the popularity of cloud computing, it provides users with the convenience of outsourcing calculations while bringing the risk of privacy disclosure. Aiming at the problem above, from the characteristics of the secure cloud service system, this paper proposes a secure high-order clustering algorithm by fast search and find of density peaks on hybrid cloud. In the proposed scheme, the client first builds the encrypted object tensors with user data using homomorphic encryption, then uploads them to the cloud to completely implement the high-order clustering computing by fast search and find of density peaks over the encrypted object tensors. In the end, clustering results perturbed with random number are returned to client for remove perturbation. The performance of the proposed method is evaluated on a smart grid data set in terms of clustering accuracy, efficiency and speedup ratio. Experimental results reveal that the proposed approach can accurately and effectively cluster big data without disclosing user privacy while ensuring that the client is very lightweight. Therefore, the proposed scheme with high security and scalability is suitable for clustering Industrial Internet-of-Things big data.
19 September 2018 in IEEE Transactions on Industrial Informatics
Aiming at discovering hidden different data structures in big data from different perspectives, a tensorbased multiple clustering method has been developed recently, which can be widely used in Industrial IoT to improve production and service quality. However, due to the high computational cost and huge volume of data, outsourcing computing to relatively inexpensive cloud servers can greatly save local costs, but there is a high risk of revealing user privacy. To address the problem above, a privacy preserving tensor-based multiple clustering method on the secure hybrid cloud is proposed. The proposed scheme utilizes homomorphic cryptosystem to encrypt object tensors, then employs cloud servers to completely implement multiple clustering calculation over encrypted object tensors. Furthermore, a series of related security sub-protocols are proposed to support privacy preserving tensor-based multiple clusterings. In the proposed scheme, only encryption and removing perturbation are performed on the client, which is very lightweight for users. Experimental results show the proposed scheme is accurate and efficient when clustering objects to different groups, while no private or additional information is leaked. Moreover, when employing more cloud nodes, the scheme has high scalability, thus it is very suitable for clustering Industrial IoT big data.
15 July 2022 in IEEE Transactions on Industrial Informatics
With the deepening of industry 4.0 paradigm in logistics systems, artificial intelligent has been widely used to improve the quality of logistics services. Considering that data collected in comprehensive logistics service system usually integrate multistage complex information such as traffic flow records and spatiotemporal trajectory, it is inevitable that the data are incomplete and partially missing due to equipment failures, communication interruptions, etc. As an effective spatiotemporal completion tool in logistics systems, low-rank tensor completion has aroused extensive research interest thanks to its excellent performance on data recovery. Although existing tensor completion methods effectively capture the complex associations/dependencies of multidimensional inputs, they fail to exploit the potential characteristics of spatiotemporal data, such as the periodicity. In this article, we propose a jointly low-rank tensor completion method for logistics data completion, which constructs multiple periodic subtensors by setting an appropriate time window, then performs jointly low-rank completion and imputation. In addition, we also provide an optimization algorithm based on Alternating Direction Method of Multiplier framework for the proposed problem. Experimental results on four logistics-related datasets have further demonstrated the promising performance of the proposed method compared with other state-of-the-art competitors. We believe that the proposed approach not only effectively maintains the advantages of classical completion methods, but also fully excavates the multidimensional correlation and hidden patterns behind records, and further provides a novel and effective strategy for data completion and imputation in logistics systems.
3 February 2023 in IEEE Transactions on Computational Social Systems
As an important part of cyber–physical–social intelligence, artificial intelligence (AI)-driven smart healthcare is committed to promoting the application of human–machine hybrid augmented intelligence in the medical field, including AI-assisted medical image analysis and lesion recognition. Among them, deep learning models represented by fully convolutional networks (FCNs) have achieved excellent performance in medical image segmentation. However, limited by the complex structure of segmentation networks and the inherently redundant characteristics of convolutional operation, the scale of these models is extremely large. To further promote the application of machine intelligence in the field of medical image analysis, we propose an attention U-Net based on Bi-ConvLSTM (AUBC-Net) for accurate segmentation of medical images in this article. Different from classical U-Net, the proposed model deals with the potential association between decoding features and encoding features by bidirectional convolution LSTM. Furthermore, for the inherent redundancy characteristics of FCNs, we propose a lightweight feature generation strategy and optimize the calculation process of Bi-ConvLSTM based on tensor multilinear algebra, which can greatly reduce the number of network parameters. In addition, we have conducted the image segmentation experiments on two benchmark medical datasets, and the experimental results demonstrate that the proposed model can not only achieve better performance than existing methods, but also effectively compress network parameters while ensuring performance, which greatly facilitates AI-driven smart medical applications.