Special issue on "Learning in data science: theory, methods and applications"-preface by the guest editors
Special issue on "Learning in data science: theory, methods and applications"-preface by the guest editors
复制标题
《数据科学学习：理论、方法与应用》特刊——客座编辑序言
DOI：
10.1007/s11634-020-00431-6
复制
发表时间：
2020
影响因子：
1.6
通讯作者：
Baier D
中科院分区：
文献类型：
--
作者：
Baier D
最近，数据科学所涉及的学科之间的相互作用，最明显的是统计学和计算机科学。通过开发和应用越来越复杂的数据、数据流、文本或图像处理方法，统计、深度和机器学习（监督和无监督）取得了令人印象深刻的进步。它们现在被进一步开发并用于许多应用领域，例如工程，金融，基因组学，工业自动化，工业4.0，营销，个性化医疗或医疗保健，系统生物学。这些学习方法中的许多即将进入现实世界的使用，在许多情况下，有很好的理由，严格的法律的要求。这重新要求方法不仅要准确，而且要让从业者获得关于学习过程和结果的重要见解。因此，在应用领域中，确保可解释性至关重要，在这些应用领域中，对系统的信任对于系统的接受至关重要，并且故障可能导致法律的责任。因此，在这期特刊中，我们征集描述和应用深度，机器和统计学习领域新发展的稿件，讨论和评估各种学习方法的可解释性和/或其对不确定性的评估，使用可理解的降维和表示
Recently, the interplay of disciplines involved in data science, most notably statistics and computer science has intensified. Impressive advances in statistical, deep, and machine learning (both supervised and unsupervised) have been achieved by developing and applying more and more complex methods for data, data stream, text, or image processing. They are now further developed and used in many fields of applications like, eg, engineering, finance, genomics, industrial automation, industry 4.0, marketing, personalised medicine or health care, systems biology. Many of these learning methods are about to make their way into real-world usage now with—in many cases and for good reasons—strict legal requirements. This has renewed the demand that methods should not only be accurate but allow the practitioner to obtain important insights about the learning process and results at hand. So, ensuring interpretability is of central importance in application domains where trust into the system is essential for its acceptance and where malfunctioning may result in legal liability.In this special issue, we therefore solicit contributions that describe and apply new developments in the field of deep, machine, and statistical learning, discuss and evaluate the interpretability of various types of learning methods and/or their assessment of uncertainty, use dimension reduction and representations which are understandable