User Tools

Site Tools


seminars:datasci:210504

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

seminars:datasci:210504 [2021/04/23 12:39] – created qyuseminars:datasci:210504 [2021/04/23 12:40] (current) qyu
Line 1: Line 1:
 +<WRAP centeralign>##Data Science Seminar##\\ Hosted by Department of Mathematical Sciences</WRAP>
 +
 +  * Date: Tuesday, May 4, 2021
 +  * Time: 12:00pm -- 1:00pm
 +  * Room: Zoom meeting
 +  * Speaker: Zhou Wang (Binghamton University)
 +  * Title: Multiclass Anomaly Detector: the CS++Support Vector Machine
 +
 +<WRAP center box 80%>
 +<WRAP centeralign>**//Abstract//**</WRAP>
 +A new support vector machine (SVM) variant, called CS++SVM, is presented
 +combining multiclass classification and anomaly detection in a single-step process to
 +create a trained machine that can simultaneously classify test data belonging to
 +classes
 +represented in the training set and label as anomalous test data belonging to
 +classes not
 +represented in the training set. A theoretical analysis of the properties of the new
 +method, showing how it combines properties inherited both from the
 +conic-segmentation SVM
 +(CS-SVM) and the 1-class SVM. Finally, experimental results are presented to
 +demonstrate
 +the effectiveness of the algorithm for both simulated and real-world data.
 +
 +</WRAP>