Data-driven Multi-class Classifier and its Applications in Biomedicine
数据驱动的多类分类器设计及其在生物医学中的应用
报告人：Dr. Jie Zhou(Associate Professor,Northern Illinois University)
报告摘要：
Multi-class classifiers have wide usages in pattern recognition problems
including those in biology and medicine. We present a method of designing a
multi-class classifier which we called Data-driven Error Correcting Output
Coding (DECOC). By exploring the distribution of data classes and optimizing
both the composition and the number of base learners, DECOC is capable of
decomposing a multi-class problem into multiple binary problems via an
effective and compact code matrix. We apply DECOC to the classification of
cancer tissue types based on microarray gene expressions. Results show that
the proposed DECOC is able to deliver competitive accuracy compared with other
multi-classifiers based on ECOC and pairwise coupling. In addition, I will
also briefly discuss two projects of biological image annotation and
classification of brain computer interface (BCI) signals.
报告人简介：
Jie Zhou is a tenured Associate Professor at Department of Computer Science of
Northern Illinois University, USA. She obtained her B.S. degree and M. S.
degree in biomedical engineering from Southeast University, China in 1993 and
1996, respectively. She obtained her Ph.D. degree in computer science from
Concordia University, Canada in 2000. She is an IEEE senior member and an
associate editor of Pattern Recognition journal. She is also a reviewer of
many journals and international conferences in the field of pattern
recognition and its applications in biomedicine. She has given invited talks
in many research institutes and conferences including Lawrence Berkeley
National Lab and IEEE international conference of medicine and biology.
时间：2010年1月12日（周二）下午2：30
地点：逸夫科技馆生物电子学国家重点实验室3楼会议室