Search

We aimed to implement four data partitioning strategies evaluated with four federated learning (FL) algorithms and investigate the impact of data distribution on FL model performance in detecting steatosis using B-mode US images. A private dataset (153 patients; 1530 images) and a public dataset (55 patient; 550 images) were included in this retrospective study. The datasets contained patients with metabolic dysfunction-associated fatty liver disease (MAFLD) with biopsy-proven steatosis grades and control individuals without steatosis. We employed four data partitioning strategies to simulate FL scenarios and we assessed four FL algorithms. We investigated the impact of class imbalance and the mismatch between the global and local data distributions on the learning outcome. Classification performance was assessed with area under the receiver operating characteristic curve (AUC) on a separate test set. AUCs were 0.93 (95% CI 0.92, 0.94) for source-based partitioning scenario with FedAvg, 0.90 (95% CI 0.89, 0.91) for a centralized model, and 0.83 (95% CI 0.81, 0.85) for a model trained in a single-center scenario. When data was perfectly balanced on the global level and each site had an identical data distribution, the model yielded an AUC of 0.90 (95% CI 0.88, 0.92). When each site contained data exclusively from one single class, irrespective of the global data distribution, the AUC fell in the range of 0.34-0.70. FL applied to B-mode US images provide performance comparable to a centralized model and higher than single-center scenario. Global data imbalance and local data heterogeneity influenced the learning outcome., (© 2024. The Author(s).)

We present a new parallel density-based spatial clustering of applications with noise (DBSCAN) algorithm for spark on the google cloud platform (GCP). Statistical analysis is applied to determine DBSCAN's optimal parameters to enhance clustering performance. for scalability cost-based, R-tree partitioning is selected based on the distribution of the dataset into balanced workloads. Parallel DBSCAN consists of three parts: local DBSCAN, partitioning, and merging. Optimizing the partitioning of parallel DBSCAN is important to save time and space compared to serial DBSCAN. This approach can improve the performance and time cost of large datasets. the modified statistical cost-based (SCbs-DBSCAN) is applied to the UCI (university of california irvine) standard datasets, basic benchmark clustering and large different scales of data. For clustering performance and time cost, the experimental results show that the proposed algorithm achieve 10~15% more efficiently, and can run about 1.5x~3x faster than alternative Parallel DBSCAN method on Spark without sacrificing clustering quality. [ABSTRACT FROM AUTHOR]

This paper presents a parallel varied density-based clustering algorithm with optimized data partition (PVDB). First, we improve the partition with reduced boundary points algorithm using shared nearest neighbour (SNN) methods and propose the reachable partition with reduced boundary points algorithm. Second, we introduce a layered grouping grid structure and propose an efficient k nearest neighbour (kNN) search method. This method enhances the efficiency of kNN searches and determines whether kNNs are in their own partitions. Third, we propose a new merging strategy for connecting clusters in different partitions, based on the reachable point concept. Meanwhile, the strategy avoids connecting clusters with varying densities by SNN as occurs with SNN-based clustering methods. Our algorithm is implemented and compared with DBSCANMR and GriDBSCAN using the MapReduce paradigm and shows better varied density clustering capability and scalability. In addition, varied applications show our algorithm's capability of discerning spatial patterns and extending to many fields. [ABSTRACT FROM AUTHOR]

The pairwise alignment (PA) algorithm is widely used in bioinformatics to analyze biological sequence. With the advance of sequencer technology, a massive amount of DNA fragments are sequenced much quicker and cheaper. Thus, the alignment algorithm needs to be parallelized to be able to align them in a shorter time. Many previous researches have parallelized PA algorithm using various data partitioning schema, but it is unknown which one is the best. The data partitioning schema is important for parallel PA performance, because this algorithm uses dynamic programming technique that needs intense inter-thread communication. In this paper, we compared four partitioning schemas to find the best performing one on shared memory system. Those schemas are: blocked columnwise, rowwise, antidiagonal, and blocked columnwise with manual scheduling and loop unrolling. The testing is done on quad-core processor using DNA sequence of 1000 to 16000 bp as the input. The blocked columnwise with manual scheduling and loop unrolling schema gave the best performance of 89% efficiency. The synchronization time is minimized to get the best performance possible. This result provided high performance parallel PA with fine-grain parallelism that can be used further to develop parallels multiple sequence alignment (MSA). [ABSTRACT FROM AUTHOR]

Aimed at the problem that the mean shift algorithm sometimes can not compute large volumes of image data, or the data-computing speed may be very slow, by analyzing the principle of the mean-shift image segmentation, this paper presents an image-segmentation idea by processing the data in parallel computing environment, i.e. Data Partition Method with Buffer (DPMB), then we can compute different parts of data separately. By this method, we can avoid the limit of data amounts and memory errors problem of computer, which improves the efficiency of data segmentation. [ABSTRACT FROM AUTHOR]

This work presents a method for characterizing and correcting the geometric errors of the movement of the lateral stage of Imaging Confocal Microscope (CM) in extended topography measurement. For an extended topography measurement, a defined number of 2D images are taken and stitched by correlation methods. Inaccuracies due to linear displacement, vertical and horizontal straightness errors, angular errors, and squareness errors based on the assumption of the rigid body kinematics are described. A mathematical model for the scale calibration of the X- and Y- coordinates is derived according to the system kinematics, the axis chain vector of CM, and the geometric error functions and their approximations by Legendre polynomials. The correction coefficients of the kinematic modelling are determined by the measured and certified data of a dot grid target standard artefact. To process the measurement data, algorithms for data partitions, fittings of cylinder centers, and determinations of coefficients are developed and validated. During which methods such as form removal, K-means clustering, linear and non-linear Least Squares are implemented. Results of the correction coefficients are presented in Part II based on the experimental studies. The mean residual reduces 29.6% after the correction of the lateral stage errors. [ABSTRACT FROM AUTHOR]