Open-World Classification or Machine Learning

(Open Classification or Open Learning)
A form of Lifelong Learning


New Book: "Lifelong Machine Learning." by Z. Chen and B. Liu, Morgan & Claypool Publishers, November 2016.

Open world machine learning (a.k.a. open world classification or open classification) is getting very important as the learning agent is increasingly working in the real-life open and dynamic environment, e.g., chatbots and self-driving cars. We can no longer make the closed world assumption (Fei and Liu 2016), which is made by the classic machine learning paradigm.

Most existing research on supervised learning or classification makes the closed world assumption, which focuses on designing accurate classifiers under the assumption that all test classes are known at training time. A more realistic scenario is to expect unseen classes during testing (open world). In this case, the goal is to design a learning system that classifies data of the known classes into their respective classes and also to reject/detect instances from unknown classes. This problem is called open-world classification. Apart from detecting the unseen classes, we also want to incrementally learn the new classes. In fact, an open world learner should be able to do the following:

In the process, the system accumulates more and more knowledge (Fei et al., 2016). The learner is self-motivated and it knows what it does and does not know.

Publications

         TextBook: Zhiyuan Chen and Bing Liu. Lifelong Machine Learning. Morgan & Claypool, 2016.

Created on Jan 24, 2018 by Bing Liu.