Federated learning (FL) is an appealing model training technique that utilizes heterogeneous datasets and user devices, ensuring user data privacy. Existing FL research proposed device selection schemes to balance the...
详细信息
ISBN:
(纸本)9798350368543;9798350368536
Federated learning (FL) is an appealing model training technique that utilizes heterogeneous datasets and user devices, ensuring user data privacy. Existing FL research proposed device selection schemes to balance the computing speeds of devices. However, we observe that these schemes compromise prediction accuracy by similar to 57.7%. To solve this problem, we present Harmonia that enhances prediction accuracy, while also balancing the diverse computing speeds of devices. Our evaluation shows that Harmonia improves prediction accuracy by similar to 1.7x over existing schemes.
Undergraduate engineering programs are typically considered some of the most challenging as their curricula require students to have an aptitude for math, science, and engineering. The resources (time, effort, funds) ...
详细信息
ISBN:
(纸本)9798350372977;9798350372984
Undergraduate engineering programs are typically considered some of the most challenging as their curricula require students to have an aptitude for math, science, and engineering. The resources (time, effort, funds) required to finish an engineering degree is substantial. Therefore, it is imperative that the engineering students are supported with well-informed academic guidance as early in their education as possible so that these resources can be used most effectively. Analytical and data-driven methods such as machinelearning techniques can be used to inform this guidance process by predicting student success based on features such as individual traits and academic performance. In that direction, we investigated the effectiveness of using machinelearning in predicting engineering student success based on academic performance in core math, physics, and engineering courses in three undergraduate engineering programs. The data categories selected for training and testing of the machinelearning models in this study are common to most engineering programs nationwide and can be customized in a straightforward manner for other engineering disciplines. The methodology and results outlined in this preliminary study shows promise for predicting degree and cumulative GPA in our three engineering programs.
Missing values are a significant problem in data analysis and machinelearning applications. This study looks at the efficacy of machinelearning (ML) - based imputation strategies for dealing with missing data. K-nea...
详细信息
Distribution grids are evolving due to rising electricity demand and renewable energy integration, requiring efficient operation and effective planning. To achieve this, one essential step is translating the available...
详细信息
ISBN:
(纸本)9798350318562;9798350318555
Distribution grids are evolving due to rising electricity demand and renewable energy integration, requiring efficient operation and effective planning. To achieve this, one essential step is translating the available load data into actionable insights. machinelearning (ML) approaches have emerged as promising solutions, leveraging increasing availability of data and computational capabilities. While research papers exist on applications of ML in power grids, a review in low-voltage substation-level is missing, an aspect that will be explored in this paper. The significance of emphasis at this level is twofold: ensuring privacy protection while gaining insights into consumption behavior, and eliminating the need for installing new meters or adjusting communication infrastructure. The paper covers three main ML algorithms, supervised, unsupervised, and reinforcement learning, their applications, while providing a critical discussion of their strengths and limitations. Furthermore, the paper provides recommendations for future studies.
This paper introduces a 3D model-based approach to calculate the coverage of target data, which aims to provide an intuitive and efficient way to analyze and evaluate the performance of remote sensing data acquisition...
详细信息
ISBN:
(纸本)9798400709234
This paper introduces a 3D model-based approach to calculate the coverage of target data, which aims to provide an intuitive and efficient way to analyze and evaluate the performance of remote sensing data acquisition of targets. We classify existing data by load and target, build a 3D model of the target, and project the data onto a hemisphere to visually show the coverage of the data. This method combines datascience, GIS technology and the application of the *** plug-in to provide a powerful tool for research and decision-making in the field of remote sensing.
The research aims to optimize agriculture using datascience techniques. Agriculture is a critical sector for sustaining life on earth, and optimizing it can enhance food security and increase the profitability of far...
详细信息
The advancements in computational techniques regarding parallel computing and machinelearning are revolutionizing stock market prediction, This study explores the effectiveness of parallel computing architectures in ...
详细信息
ISBN:
(纸本)9798350395839;9798350395846
The advancements in computational techniques regarding parallel computing and machinelearning are revolutionizing stock market prediction, This study explores the effectiveness of parallel computing architectures in predicting stock market movements. Existing literature reveals a huge shift towards employing machinelearning models, especially in handling larger financial datasets, yet there remains a gap in understanding the full potential of parallel computing in this domain. Our research aims to bridge this gap by developing a comparative analysis between two Random Forest models: one utilizing parallel processing and the other based on sequential computation. Employing a comprehensive dataset that includes financial data from 2018 with 225 indicators of the US stock market, the data has been pre-processed to ensure its suitability for analysis. The methodology involves constructing and training both models, with the parallel model utilizing the multi-core capability of an Apple M1 chip and evaluating them based on accuracy and training time. The findings reveal that while both models achieve an impressive 100% accuracy, the parallel processing model significantly reduces training time, demonstrating the efficiency of parallel computing in rapid data processing. This research not only highlights the potential of parallel computing in enhancing the speed and accuracy of financial market predictions but also contributes to the broader field of financial analytics by suggesting new avenues for future research, including the application of deep learning models and the integration of a wider range of financial indicators.
Effective machinelearning algorithms require meticulous data preparation. In contemporary microprocessors, ensuring predictable and efficient memory access is crucial to mitigate cache misses. Therefore, compact data...
详细信息
ISBN:
(纸本)9783031606373;9783031606380
Effective machinelearning algorithms require meticulous data preparation. In contemporary microprocessors, ensuring predictable and efficient memory access is crucial to mitigate cache misses. Therefore, compact data structures with high information density are crucial for optimal computation. This paper discusses the interplay of musical meter and microtiming, as long-term expectation of pattern repetition and short-term deviations. A review of musicological, cognitive and computational models is presented. The literature regards meter perception of rhythmic patterns as a manifestation of human attentive behavior, which is also culturally dependent. Additionally, the paper discusses dissimilarity measures of rhythmic patters. The author introduces two algorithms featuring normalized binary representations of rhythmic patterns of inter-onset intervals, activation functions for computing linear, and logarithmic time differences. Bitwise operations are implemented for the bit-sequences of rhythmic patterns. The author offers an implementation of mutual information calculation as bitwise operations.
Neuromorphic computing is a new paradigm that emerges from the structure and function of the human brain and aims to revolutionize computing. The technology is designed to simulate the high speed, low power consumptio...
详细信息
暂无评论