张敏灵
博士 教授 博士生导师
东南大学 计算机科学与工程学院
从事机器学习与数据挖掘领域的研究工作
个性化签名
 姓名：张敏灵
 目前身份：在职研究人员
 担任导师情况：博士生导师
 学位：博士

学术头衔：
博士生导师
 职称：高级教授

学科领域：
人工智能
 研究兴趣：从事机器学习与数据挖掘领域的研究工作
张敏灵分别于2001年6月、2004年6月和2007年9月于南京大学计算机科学与技术系获学士、硕士和博士学位。自2001年起长期从事机器学习与数据挖掘领域的研究工作，现为东南大学计算机科学与工程学院教授。先后承担本科生专业主干课程《软件工程》、 “东南大学—Monash大学联合研究生院”全英文课程《Pattern Recognition》的教学工作。已发表学术论文20余篇，部分论文发表在领域内重要国际期刊，如《Data Mining and Knowledge Discovery》、《IEEE Trans. Pattern Analysis and Machine Intelligence》、《IEEE Trans. Knowledge and Data Engineering》、《IEEE Trans. Systems, Man, and Cybernetics – Part B: Cybernetics》等，以及领域内重要国际会议，如IJCAI、KDD、AAAI、ICDM等。担任重要国际期刊《Machine Learning》客座编辑，应邀担任ACML’14 Workshop CoChair、KDD’12 Student Volunteer共同主席，IJCAI’13、SDM’13等高级程序委员，以及ICML’14、AAAI’13/’12、KDD’11/’10等重要国际会议程序委员。现（曾）任中国人工智能学会机器学习专委会常务委员、中国计算机学会人工智能与模式识别专委会委员、江苏省计算机学会人工智能专委会秘书长等。先后主持国家自然科学基金、教育部博士点基金等的研究工作。

主页访问
23

关注数
1

成果阅读
121

成果数
26
【期刊论文】Improve MultiInstance Neural Networks through Feature Selection
Neural Processing Letters，2004，19（）：1–10
2004年02月01日
Multiinstance learning is regarded as a new learning framework where the training examples are bags composed of instances without labels, and the task is to predict the labels of unseen bags through analyzing the training bags with known labels. Recently, a multiinstance neural network BPMIP was proposed. In this paper, BPMIP is improved through adopting two different feature selection techniques, i.e. feature scaling with Diverse Density and feature reduction with principal component analysis. In detail, before feature vectors are fed to a BPMIP neural network, they are scaled by the feature weights found by running Diverse Density on the training data, or projected by a linear transformation matrix formed by principal component analysis. Experiments show that these feature selection mechanisms can significantly improve the performance of BPMIP.
无
0

9浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】Adapting RBF neural networks to multiinstance learning
Neural Processing Letters，2006，23（）：1–26
2006年02月01日
In multiinstance learning, the training examples are bags composed of instances without labels, and the task is to predict the labels of unseen bags through analyzing the training bags with known labels. A bag is positive if it contains at least one positive instance, while it is negative if it contains no positive instance. In this paper, a neural network based multiinstance learning algorithm named RBFMIP is presented, which is derived from the popular radial basis function (RBF) methods. Briefly, the first layer of an RBFMIP neural network is composed of clusters of bags formed by merging training bags agglomeratively, where Hausdorff metric is utilized to measure distances between bags and between clusters. Weights of second layer of the RBFMIP neural network are optimized by minimizing a sumofsquares error function and worked out through singular value decomposition (SVD). Experiments on realworld multiinstance benchmark data, artificial multiinstance benchmark data and natural scene image database retrieval are carried out. The experimental results show that RBFMIP is among the several best learning algorithms on multiinstance problems.
无
0

9浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】Multilabel Neural Networks with Applications to Functional Genomics and Text Categorization
IEEE Transactions on Knowledge and Data Engineering，2006，18（10）：1338  135
2006年08月28日
In multilabel learning, each instance in the training set is associated with a set of labels and the task is to output a label set whose size is unknown a priori for each unseen instance. In this paper, this problem is addressed in the way that a neural network algorithm named BPMLL, i.e., backpropagation for multilabel learning, is proposed. It is derived from the popular backpropagation algorithm through employing a novel error function capturing the characteristics of multilabel learning, i.e., the labels belonging to an instance should be ranked higher than those not belonging to that instance. Applications to two realworld multilabel learning problems, i.e., functional genomics and text categorization, show that the performance of BPMLL is superior to that of some wellestablished multilabel learning algorithms
无
0

6浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】Solving multiinstance problems with classifier ensemble based on constructive clustering
Knowledge and Information Systems volume，2006，11（）：155–170
2006年08月10日
In multiinstance learning, the training set is composed of labeled bags each consists of many unlabeled instances, that is, an object is represented by a set of feature vectors instead of only one feature vector. Most current multiinstance learning algorithms work through adapting singleinstance learning algorithms to the multiinstance representation, while this paper proposes a new solution which goes at an opposite way, that is, adapting the multiinstance representation to singleinstance learning algorithms. In detail, the instances of all the bags are collected together and clustered into d groups first. Each bag is then rerepresented by d binary features, where the value of the ith feature is set to one if the concerned bag has instances falling into the ith group and zero otherwise. Thus, each bag is represented by one feature vector so that singleinstance classifiers can be used to distinguish different classes of bags. Through repeating the above process with different values of d, many classifiers can be generated and then they can be combined into an ensemble for prediction. Experiments show that the proposed method works well on standard as well as generalized multiinstance problems.
无
0

5浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】MLKNN: A lazy learning approach to multilabel learning
Pattern Recognition，2007，40（7）：20382048
2007年07月01日
Multilabel learning originated from the investigation of text categorization problem, where each document may belong to several predefined topics simultaneously. In multilabel learning, the training set is composed of instances each associated with a set of labels, and the task is to predict the label sets of unseen instances through analyzing training instances with known label sets. In this paper, a multilabel lazy learning approach named MLKNN is presented, which is derived from the traditional Knearest neighbor (KNN) algorithm. In detail, for each unseen instance, its K nearest neighbors in the training set are firstly identified. After that, based on statistical information gained from the label sets of these neighboring instances, i.e. the number of neighboring instances belonging to each possible class, maximum a posteriori (MAP) principle is utilized to determine the label set for the unseen instance. Experiments on three different realworld multilabel learning problems, i.e. Yeast gene functional analysis, natural scene classification and automatic web page categorization, show that MLKNN achieves superior performance to some wellestablished multilabel learning algorithms.
Machine learning， Multilabel learning， Lazy learning， Knearest neighbor， Functional genomics， Natural scene classification， Text categorization
0

5浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】Mlrbf: RBF Neural Networks for MultiLabel Learning
Neural Processing Letters，2009，29（）：61–74
2009年02月10日
Multilabel learning deals with the problem where each instance is associated with multiple labels simultaneously. The task of this learning paradigm is to predict the label set for each unseen instance, through analyzing training instances with known label sets. In this paper, a neural network based multilabel learning algorithm named Mlrbf is proposed, which is derived from the traditional radial basis function (RBF) methods. Briefly, the first layer of an Mlrbf neural network is formed by conducting clustering analysis on instances of each possible class, where the centroid of each clustered groups is regarded as the prototype vector of a basis function. After that, second layer weights of the Mlrbf neural network are learned by minimizing a sumofsquares error function. Specifically, information encoded in the prototype vectors corresponding to all classes are fully exploited to optimize the weights corresponding to each specific class. Experiments on three realworld multilabel data sets show that Mlrbf achieves highly competitive performance to other wellestablished multilabel learning algorithms.
无
0

3浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】MIMLRBF: RBF neural networks for multiinstance multilabel learning
Neurocomputing，2009，72（1618）：39513956
2009年10月01日
In multiinstance multilabel learning (MIML), each example is not only represented by multiple instances but also associated with multiple class labels. Several learning frameworks, such as the traditional supervised learning, can be regarded as degenerated versions of MIML. Therefore, an intuitive way to solve MIML problem is to identify its equivalence in its degenerated versions. However, this identification process would make useful information encoded in training examples get lost and thus impair the learning algorithm's performance. In this paper, RBF neural networks are adapted to learn from MIML examples. Connections between instances and labels are directly exploited in the process of first layer clustering and second layer optimization. The proposed method demonstrates superior performance on two realworld MIML tasks.
Machine learning， Multiinstance multilabel learning， Radial basis function， Scene classification， Text categorization
0

5浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】Multiinstance clustering with applications to multiinstance prediction
Applied Intelligence ，2008，31（）：47–68
2008年01月23日
In the setting of multiinstance learning, each object is represented by a bag composed of multiple instances instead of by a single instance in a traditional learning setting. Previous works in this area only concern multiinstance prediction problems where each bag is associated with a binary (classification) or realvalued (regression) label. However, unsupervised multiinstance learning where bags are without labels has not been studied. In this paper, the problem of unsupervised multiinstance learning is addressed where a multiinstance clustering algorithm named Bamic is proposed. Briefly, by regarding bags as atomic data items and using some form of distance metric to measure distances between bags, Bamic adapts the popular k Medoids algorithm to partition the unlabeled training bags into k disjoint groups of bags. Furthermore, based on the clustering results, a novel multiinstance prediction algorithm named Bartmip is developed. Firstly, each bag is rerepresented by a kdimensional feature vector, where the value of the ith feature is set to be the distance between the bag and the medoid of the ith group. After that, bags are transformed into feature vectors so that common supervised learners are used to learn from the transformed feature vectors each associated with the original bag’s label. Extensive experiments show that Bamic could effectively discover the underlying structure of the data set and Bartmip works quite well on various kinds of multiinstance prediction problems.
无
0

6浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】Feature selection for multilabel naive bayes classification
Information Sciences，2009，179（19）：32183229
2009年09月09日
In multilabel learning, the training set is made up of instances each associated with a set of labels, and the task is to predict the label sets of unseen instances. In this paper, this learning problem is addressed by using a method called Mlnb which adapts the traditional naive Bayes classifiers to deal with multilabel instances. Feature selection mechanisms are incorporated into Mlnb to improve its performance. Firstly, feature extraction techniques based on principal component analysis are applied to remove irrelevant and redundant features. After that, feature subset selection techniques based on genetic algorithms are used to choose the most appropriate subset of features for prediction. Experiments on synthetic and realworld data show that Mlnb achieves comparable performance to other wellestablished multilabel learning algorithms.
无
0

2浏览

0点赞

0收藏

0分享

0下载

0评论

引用
【期刊论文】CoTrade: Confident CoTraining With Data Editing
IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics)，2011，41（6）：1612  162
2011年06月23日
Cotraining is one of the major semisupervised learning paradigms that iteratively trains two classifiers on two different views, and uses the predictions of either classifier on the unlabeled examples to augment the training set of the other. During the cotraining process, especially in initial rounds when the classifiers have only mediocre accuracy, it is quite possible that one classifier will receive labels on unlabeled examples erroneously predicted by the other classifier. Therefore, the performance of cotraining style algorithms is usually unstable. In this paper, the problem of how to reliably communicate labeling information between different views is addressed by a novel cotraining algorithm named COTRADE. In each labeling round, COTRADE carries out the label communication process in two steps. First, confidence of either classifier's predictions on unlabeled examples is explicitly estimated based on specific data editing techniques. Secondly, a number of predicted labels with higher confidence of either classifier are passed to the other one, where certain constraints are imposed to avoid introducing undesirable classification noise. Experiments on several realworld datasets across three domains show that COTRADE can effectively exploit unlabeled data to achieve better generalization performance.
无
0

4浏览

0点赞

0收藏

0分享

0下载

0评论

引用