IEEE Transactions on Computational Social Systems
IEEE Transactions on Computational Social Systems
复制标题
IEEE 计算社会系统汇刊
DOI：
--
复制
发表时间：
2023
影响因子：
5
通讯作者：
V. Sánchez
中科院分区：
文献类型：
--
作者：
S. Najib;J. Rodríguez;M. Ríos;M. Muniaín;R. Goberna;V. Sánchez
近年来，出现了许多先进的机器学习技术，如深度学习、迁移学习等。深度学习方法在图像和视频分析、自然语言处理、语音识别等方面取得了巨大的成功，近年来开始在人类行为分析的认知计算中得到应用。迁移学习利用在解决一个问题中获得的数据或知识来帮助解决另一个不同但相关的问题。迁移学习在认知计算中可以特别有用，以应对不同个体或任务之间的差异，加速学习和提高绩效。深度学习和迁移学习也可以结合起来，以利用这两个领域的优势。虽然利用先进的机器学习方法进行人类行为分析的研究越来越多，但到目前为止，仍有许多基本问题没有得到解决。例如，深度学习如何从多个形态代表人类的外表和行为？我们如何将数据从一种通道映射到另一种通道，以实现跨通道人类行为分析？我们如何识别和利用来自两个或多个不同模式的元素之间的关系，以进行全面的行为分析？我们如何融合来自两个或更多个医疗模式的信息，以执行更准确的预测？我们如何在模态和它们的表示之间传递知识？在已观察到的情况下，我们如何恢复丢失的通道数据？如何延长人类行为分析设备和网络的使用寿命并增强其可用性？在过去的十年里，几种机器学习模型已经被开发出来，并在一些现实世界的例子中显示出良好的结果，例如多媒体描述和检索，这有助于我们开发和开发基于认知计算的高级机器学习算法，以解决人类行为分析的基本问题。本期特刊旨在从认知计算的角度为研究人员提供一个论坛，介绍人类行为分析中最先进的方法和应用的最新进展。可能的主题包括但不限于：·用于人类行为分析的卷积/递归神经网络·用于人类行为分析的深度前馈/信念/残差网络·用于人类行为分析的极端学习机·用于人类行为分析的生成性对抗网络·用于人类行为分析的长期短期记忆·用于人类行为分析的迁移学习·用于人类行为分析的领域适应·用于人类行为分析的协变量转移·用于人类行为分析的深度转移学习·用于人类行为分析的认知无线充电器网络
Recently, many advanced machine learning technologies have appeared, such as deep learning, transfer learning, and so on. Deep learning method has achieved great success in image and video analysis, natural language processing, speech recognition, etc., and recently has been starting to find applications in cognitive computing for human behavior analysis. Transfer learning makes use of data or knowledge gained in solving one problem to help solve a different, albeit related problem. Transfer learning can be particularly useful in cognitive computing to cope with the variability across individuals or tasks, accelerating learning and improving performance. Deep learning and transfer learning can also be integrated to take advantage of both domains. Although the studies of human behavior analysis using the advanced machine learning methods become more and more popular, there are many fundamental problems unsolved so far. For example, how does deep learning represent human appearance and behaviors from multiple modalities? How do we map data from one modality to another to achieve crossmodality human behavior analysis? How do we identify and utilize relations between elements from two or more different modalities for comprehensive behavior analysis? How do we fuse information from two or more modalities to perform a more accurate prediction? How do we transfer knowledge between modalities and their representations? And how do we recover missing modality data given the observed ones? How to extend the lifetime of and enhance the usability of human behavior analysis devices and networks? In the past decade, several machine learning models have been developed and shown promising results in some realworld examples, such as multimedia description and retrieval, which facilitates us to exploit and develop advanced machine learning algorithms on cognitive computing for addressing fundamental issues of human behavior analysis. This special issue aims to provide a forum for researchers from the perspective of cognitive computing to present recent progress on state-of-the-art methods and applications to human behavior analysis. The list of possible topics includes, but not limited to: • Convolutional/recurrent neural networks for human behavior analysis • Deep feedforward/belief/residual networks for human behavior analysis • Extreme learning machines for human behavior analysis • Generative adversarial networks for human behavior analysis • Long short-term memory for human behavior analysis • Transfer learning for human behavior analysis • Domain adaptation for human behavior analysis • Covariate shift for human behavior analysis • Deep-transfer learning for human behavior analysis • Cognitive wireless charger networks for human behavior analysis