报告时间：

2013年9月8号星期天上午10:00

报告地点：

智能感知与图像理解教育部重点实验室报告厅（主楼二区227）

报告人简介：

陈欢欢，男，2004年获中国科技大学学士学位，2008年获英国伯明翰大学博士学位。海外高水平学者，现为中国科学技术大学计算机学院教授。研究方向为机器学习、数据挖掘、进化计算及其应用等。主持自然基金项目1项，基金委与英国皇家学会交流项目1项，英国The Wellcome Trust项目1项。获2011年IEEE计算智能协会优秀博士论文奖、全英杰出博士论文奖、《IEEE Transactions on Neural Networks》2009年度优秀论文奖（2012年颁奖）。发表期刊论文包括《IEEE Transactions on Neural Networks》、《IEEE Transactions on Knowledge and Data Engineering》和《IEEE Transactions on Evolutionary Computation》等；会议论文包括KDD，IJCAI，ECAI等。

报告题目：

模型空间中的学习及其在认知故障诊断与时间序列分析中的应用(Learning in the Model Space for Cognitive Fault Diagnosis and Time Series Analysis)

报告摘要：

Traditional machine learning relies on the metric space of data, i.e.

Euclidean space. However, the dimensionality of data space is often high

and the noise level is often significant. To address these problems, we

propose a learning framework "learning in the model space" by generating

a series of models to fit the nearby data. The learning algorithms can

be constructed in the generated model space by defining the distance

between these models using function analysis. Following this framework,

we illustrate how to use the learning in the model space for cognitive

fault diagnosis and for effective time series analysis. The theoretical

and empirical results confirm the effectiveness of the proposed framework.