CAREER: A Foundation for Applied Machine Learning

本项目旨在为机器学习在实践中的应用奠定基础。研究部分有两个目标。首先，开发数据收集和精化方法，使机器学习算法能够从数据中提取结构。其次，对机器学习算法进行经验和理论分析，以加深对现有方法和新方法的理解，并揭示使其适用于特定学习算法的数据集的特征。该项目的教育部分旨在通过更深入地了解各种学习算法可以检测到的结构类型，创建收集数据的工具，让学生亲身体验在实践中应用机器学习，以及通过跨学科研究项目，为从业者提供在实践中成功应用机器学习所需的基础。这些研究和教育活动是在几个应用领域开展的，包括根据遥感数据绘制全球土地覆盖图，对肺部进行高分辨率CAT扫描的基于内容的图像检索，学习检测计算机安全方面的异常情况，以及对行星数据进行船上处理。除了对应用机器学习的贡献外，这个项目还有一个好处，那就是机器学习在拟议的应用领域的成功应用将导致对医学、计算机系统开发和地理的巨大贡献。Http://mow.ecn.purdue.edu/Brodley/

This project aims to provide a foundation for applying machine learning in practice. The research component has two goals. Firstly, to develop data collection and refinement methods that allow machine learning algorithm to extract structure from data. Secondly, to perform empirical and theoretical analyses of machine learning algorithms to develop a deeper understanding of current and new methods, and to uncover the characteristics of datasets that make them amenable to particular learning algorithms. The education component of this project aims to provide practitioners with the foundation required to apply machine learning successfully in practice by providing a deeper understanding of the type of structure that various learning algorithms can detect, by creating tools for collecting data, by giving students hands on experience applying machine learning in practice, and through inter-disciplinary research projects. These research and education activities are being performed in the context of several application domains, including mapping global land cover of the Earth from remotely sensed data, content-based image retrieval of high resolution CAT scans of the lung, learning to detect anomalies for computer security, and on-board processing of planetary data. In addition to its contribution to applied machine learning, this project has the benefit that the successful application of machine learning to proposed application domains will result in non-trivial contributions to medicine, computer systems development, and geography. http://mow.ecn.purdue.edu/ brodley/