计算机科学
人工智能
机器学习
Python(编程语言)
特征学习
监督学习
领域(数学)
代表(政治)
外部数据表示
桥接(联网)
数据科学
人工神经网络
操作系统
政治
数学
法学
纯数学
计算机网络
政治学
出处
期刊:Northeastern University - Digital Repository Service
[Northeastern University]
日期:2023-01-01
摘要
The rapidly developing field of sensory signal processing, particularly in physiological data, faces numerous challenges, including complexity, high dimensionality, and often the lack of labeled data in clinical settings. Traditional approaches in feature extraction, limited by manual feature engineering, struggle to effectively capture the complex temporal dependencies inherent in these signals. This research asserts that representation learning, particularly through supervised and self-supervised learning techniques, can substantially improve the extraction of informative features from physiological signals, thereby enhancing the performance of machine learning models in real-world applications. The research focuses on the application of advanced representation learning methodologies for supervised learning. In parallel, it explores the domain of self-supervised learning. These methods are analyzed for their ability to address the challenges of nonlinearities, high-dimensional data, and the lack of annotated data in physiological signal processing. A significant part of the research is devoted to the practical examination of these techniques. This includes applying these methods to real-world physiological signals, such as pain databases, to evaluate their efficacy in practical scenarios. Furthermore, this research introduces pyKCN, a novel Python toolkit for keyword co-occurrence analysis.The research explores the potential of representation learning to navigate the complexities of physiological signals, intending to foster advancements in prognosis, diagnosis, and treatment methodologies. By bridging the gap between theoretical understanding and practical application, this research stands to provide insights into the efficient processing of physiological data and contribute to the broader landscape of machine learning in sensory signals. In essence, this thesis underscores the potential of representation learning in sensory signal processing proposing innovative methods and applications that could redefine the landscape of sensory data analysis.--Author's abstract
科研通智能强力驱动
Strongly Powered by AbleSci AI