Optimal kernel choice for domain adaption learning

Abstract

In this paper, a kernel choice method is proposed for domain adaption, referred to as Optimal Kernel Choice Domain Adaption (OKCDA). It learns a robust classier and parameters associate with Multiple Kernel Learning side by side. Domain adaption kernel-based learning strategy has shown outstanding performance. It embeds two domains of different distributions, namely, the auxiliary and the target domains, into Hilbert Space, and exploits the labeled data from the source domain to train a robust kernel-based SVM classier for the target domain. We reduce the distributions mismatch by setting up a test statistic between the two domains based on the Maximum Mean Discrepancy (MMD) algorithm and minimize the Type II error, given an upper bound on error I. Simultaneously, we minimize the structural risk functional. In order to highlight the advantages of the proposed method, we tackle a text classication problem on 20 Newsgroups dataset and Email Spam dataset. The results demonstrate that our method exhibits outstanding performance.

Twitter mining for ontology-based domain discovery incorporating machine learning

“…For demonstration purposes, this overview will consider the domain-based classification at the user level. LR (Al-Tahrawi, 2015;Yen et al, 2011), decision tree (Sharef et al, 2015) and SVM (Altınel et al, 2015;Dong et al, 2016) in particular have been used for text categorisations. Also these approaches are more narrow and computationally simpler than recently developed machine learning approaches, such as the deep learning or deep networks approaches.…”

“…Support vector machine is commonly used for conducting binary classification tasks (Boser et al, 1992) particularly involving with the confusion matrix analysis (true-positive [TP] and false-negative [FN]). SVM is relatively new and was designed for applications involving text categorization and recognition (see for example Altınel et al, 2015;Dong et al, 2016).…”