No abstract

Data-driven machine learning models, compared to numerical models, demonstrated promising improvements in detecting damage in structural health monitoring (SHM) applications. In such approaches, sensors’ data are used to train a model either in a centralized model (server) or locally inside each sensor unit node (client). The centralized learning model often leads to computing and privacy issues such as wireless transmission costs and data-sensitive vulnerability, especially in real-time settings. The decentralized model also poses different challenges such as feature correlations and relationships loss in decentralized learning settings. To handle the shortcomings of both models, we propose a new Personalized federated learning (FL) model augmented with tensor data fusion to learn and detect damage in SHM. Our approach employs FL which enables the central machine learning model to gain experience from diverse datasets located at different sensor locations. Furthermore, our proposed model addresses the problems associated non-i.i.d. data by employing the Moreau envelopes as a regularized loss function in the learning process of client’s models. Our methods help in decoupling the client models from the central one which improves personalized in FL. Our experimental evaluation on real structural datasets demonstrates promising damage detection accuracy without the need to transmit the actual data to the centralized learning model. The results also show that the data correlations and relationships from all participating sensors are preserved.

No abstract

Single-cell data analysis can transform the practice of personalised medicine by facilitating characterisation of disease-associated molecular changes across every single cell. Advanced single-cell multimodal assays can now simultaneously measure different types of molecules (e.g., DNA, RNA, Protein) across hundreds of thousands of individual cells to provide a complete molecular readout of the state of the cell. A key analytical challenge is to integrate single-cell measurements across different modalities. In recent years, different methods have been developed to integrate different molecular modalities of single cell measurements, yet there is no systematic evaluation on the performance of different techniques with respect to different pre-processig strategies. Here, we consider a general pipeline of single-cell data analyses comprising normalisation and data integration followed by the most popular dimensionality reduction methods where several algorithms can be adopted to each module in the pipeline. The performance of different algorithm combinations often depends on the sizes and characteristics of datasets. This paper explores the optimal framework of different algorithms using six datasets across diverse modalities, tissues and organisms. We use three evaluation metrics (Silhouette Coefficient Score, Adjusted Rand Index and Calinski-Harabasz Index) to explore the model's overall performance from the two aspects, including clustering performance and time efficiency. We conduct experiments based on combinations of seven normalization methods, four dimensional reduction methods, and five integration methods. Our results demonstrate that for the data integration module, the clustering performance of Seurat and Harmony are more prominent, but the time efficiency of Harmony was better. At the same time, the performance of Seurat for small data sets is superior. For the dimensionality reduction module, the UMAP method shows promising results in compatibility with the integration methods. The data normalization method could be differentiated according to different integration methods.

No abstract