Abstract. Information fusion is currently a very active research topic aimed at improving the performance of biometric systems. This paper proposes a novel method for optimizing the parameters of a score fusion model based on maximizing an index related to the Area Under the ROC Curve. This approach has the convenience that the fusion parameters are learned without having to specify the client and impostor priors or the costs for the different errors. Empirical results on several datasets show the effectiveness of the proposed approach. 1

Dimensionality reduction, image segmentation, metric learning, pairwise constraints, semi-supervised learning, visual object classification. This paper describes a semi-supervised distance metric learning algorithm which uses pairwise equivalence (similarity and dissimilarity) constraints to discover the desired groups within high-dimensional data. As opposed to the traditional full rank distance metric learning algorithms, the proposed method can learn nonsquare projection matrices that yield low rank distance metrics. This brings additional benefits such as visualization of data samples and reducing the storage cost, and it is more robust to overfitting since the number of estimated parameters is greatly reduced. Our method works in both the input and kernel induced-feature space, and the distance metric is found by a gradient descent procedure that involves an eigen-decomposition in each step. Experimental results on high-dimensional visual object classification problems show that the computed distance metric improves the performance of the subsequent clustering algorithm. 1

I would like to thank my thesis advisor, Sridhar Mahadevan. Sridhar has been such a wonderful advisor, and every aspect of this thesis has benefitted from his guidance and support throughout my graduate studies. I also like to thank Sridhar for giving me the flexibility to explore many different ideas and research topics. I am appreciative of the support offered by my other thesis committee members, Andrew McCallum, Erik Learned-Miller, and Weibo Gong. Andrew helped me on CRF, MALLET and topic modeling. Erik helped me on computer vision. Weibo has many brilliant ideas on how brains work. I got a lot of inspirations from him. I am grateful for many other professors and staff members, who helped me along. Andy Barto offered me many insightful comments and advice on my research. David Kulp and Oliver Brock helped me on bioinformatics. Stephen Scott brought me to this country, taught me machine learning/bioinformatics and offers me constant support. Vadim Gladyshev helped me on biochemistry. Mauro Maggioni helped me on diffusion wavelets. I also thank Gwyn Mitchell and Leanne Leclerc for their help with my questions over the years. I am deeply thankful to my Master thesis advisor, Zhuzhi Yuan and other teachers in Nankai University for guiding my development as

Abstract. This paper describes a novel framework for learning discriminative features, where both labeled and unlabeled data are used to map the data instances to a lower dimensional space, preserving both class separability and data manifold topology. In contrast to linear discriminant analysis (LDA) and its variants (like semi-supervised discriminant analysis), which can only return c−1featuresforaproblemwithc classes, the proposed approach can generate d features, where d is bounded only by the dimensionality of the original problem. The proposed framework can be used with both two class and multiple class problems. It can also be adapted to problems where class labels are continuous. We describe and evaluate the new approach both theoretically and experimentally, and compare its performance with other state of the art methods.

Abstract. Face recognition is currently a very active research topic due to the great variety of applications it can offer. Moreover, nowadays itis very common for people to have mobile devices such as smart phones or laptops which have an integrated digital camera. This gives the opportunity to develop face recognition applications for this type of devices. This paper treats the problem of face verification in mobile devices. The problem is discussed and analysed, observing which are the difficulties that are encountered. Some algorithms for face recognition are analysed and optimised so that they are better suited for the resource constrains of mobile devices. 1

A protection scheme for enhancing biometric template security and discriminability 9

Abstract. Prototype classifiers have been studied for many years. But most methods adopt single vectors as prototypes to represent the original data. It is difficult to learn the local information [1] with such prototypes. In this paper, we propose an incremental classifier named Subspace Based Prototype Classifier (SBPC) to handle the issue. SBPC designs an augmented strategy to enhance traditional prototype classifiers. Instead of using single vector, each prototype of SBPC represents a subset of input data using a subspace. And we employ an incremental subspace representation method to learn a subspace for each prototype. By designing a self-adaptive threshold policy, SBPC automatically learns the number and value of prototypes without any prior knowledge. Through adopting both condensing scheme and editing scheme [3], the prototypes are incremental learned, automatically adjusted (condensing scheme) and removed (editing scheme). Results of experiments described herein show that the proposed SBPC accommodates the non-stationary data environment and provides good recognition performance and storage efficiency.

These are the proceedings of the second workshop “From Local Patterns to Global Models”. The key idea of the so-called LeGo-model of data mining (Knobbe, Crémilleux, Fürnkranz & Scholz 2008) is the observation that in many different domains and application scenarios, the use of local patterns as features may considerably improve the performance of a global model. Many papers study the phases of the LeGo-Process (local pattern discovery, pattern selection, and global modeling) individually, and applications that follow this model typically assemble separate, independent components. However, optimal choices for individual phases may not necessarily lead to an optimal choice in the entire process. The key idea of the LeGo workshops is to investigate the interactions and dependencies of these phases in a data mining process. These proceedings contain 10 papers that span the entire process in various guises, from frequent pattern set mining to clustering, from subgroup discovery to classification, from rule learning to probability estimation. Applications range from spam mail filtering to game playing.

Abstract—Person recognition using facial features, e.g., mug-shot images, has long been used in identity documents. However, due to the widespread use of web-cams and mobile devices embedded with a camera, it is now possible to realize facial video recognition, rather than resorting to just still images. In fact, facial video recognition offers many advantages over still image recognition; these include the potential of boosting the system accuracy and deterring spoof attacks. This paper presents an evaluation of person identity verification using facial video data, organized in conjunction with the International Conference on Biometrics (ICB 2009). It involves 18 systems submitted by seven academic institutes. These systems provide for a diverse set of assumptions, including feature representation and preprocessing variations, allowing us to assess the effect of adverse conditions, usage of quality information, query selection, and template construction for video-to-video face authentication. Index Terms—Biometric authentication, face video recognition. I.

Most text mining tasks, including clustering and topic detection, are based on statistical methods that treat text as bags of words. Semantics in the text is largely ignored in the mining process, and mining results often have low interpretability. One particular challenge faced by such approaches lies in short text understanding, as short texts lack enough content from which statistical conclusions can be drawn easily. In this paper, we improve text understanding by using a probabilistic knowledgebase that is as rich as our mental world in terms of the concepts (of worldly facts) it contains. We then develop a Bayesian inference mechanism to conceptualize words and short text. We conducted comprehensive experiments on conceptualizing textual terms, and clustering short pieces of text such as Twitter messages. Compared to purely statistical methods such as latent semantic topic modeling or methods that use existing knowledgebases (e.g., WordNet, Freebase and Wikipedia), our approach brings significant improvements in short text understanding as reflected by the clustering accuracy. 1