Multi-label decision procedures are the target of the supervised learning algorithm we propose in this paper. Multi-label decision procedures map examples to a finite set of labels. Our learning algorithm extends Scha...
详细信息
ISBN:
(纸本)3540405046
Multi-label decision procedures are the target of the supervised learning algorithm we propose in this paper. Multi-label decision procedures map examples to a finite set of labels. Our learning algorithm extends Schapire and Singer's *** and produces sets of rules that can be viewed as trees like Alternating Decision Trees (invented by Freund and Mason). Experiments show that we take advantage of both performance and readability using boosting techniques as well as tree representations of large set of rules. Moreover, a key feature of our algorithm is the ability to handle heterogenous input data: discrete and continuous values and text data.
this paper presents a Neural Network approach for extracting classification rules from symbolic type of data. A scheme for generating data for the Neural Network training is proposed. the Neural Network employed is a ...
详细信息
ISBN:
(纸本)0780381858
this paper presents a Neural Network approach for extracting classification rules from symbolic type of data. A scheme for generating data for the Neural Network training is proposed. the Neural Network employed is a constructive learning Neural Network known as Resource Allocation Network (RAN). this Network learns very fast and exhibits good response to incoming test pattern making it suitable for rule extraction applications in datamining.
We present a model-based method for accurate extraction of pedestrian silhouettes from video sequences. Our approach is based on two assumptions, 1) there is a common appearance to all pedestrians, and 2) each individ...
详细信息
ISBN:
(纸本)0769519504
We present a model-based method for accurate extraction of pedestrian silhouettes from video sequences. Our approach is based on two assumptions, 1) there is a common appearance to all pedestrians, and 2) each individual looks like him/herself over a short amount of time. these assumptions allow us to learn pedestrian models that encompass both a pedestrian population appearance and the individual appearance variations. Using our models, we are able to produce pedestrian silhouettes that have fewer noise pixels and missing parts. We apply our silhouette extraction approach to the NIST gait data set and show that under the gait recognition task, our model-based sulhouettes result in much higher recognition rates than silhouettes directly extracted from background subtraction, or any non-model-based smoothing schemes.
the Branch & Bound (B&B) algorithm is a globally optimal feature selection method. the high computational complexity of this algorithm is a well-known problem. the B&B algorithm constructs a search tree, a...
详细信息
ISBN:
(纸本)3540140409
the Branch & Bound (B&B) algorithm is a globally optimal feature selection method. the high computational complexity of this algorithm is a well-known problem. the B&B algorithm constructs a search tree, and then searches for the optimal feature subset in the tree. Previous work on the B&B algorithm was focused on how to simplify the search tree in order to reduce the search complexity. Several improvements have already existed. A detailed analysis of basic B&B algorithm and existing improvements is given under a common framework in which all the algorithms are compared. Based on this analysis, an improved B&B algorithm, BBPP+, is proposed. Experimental comparison shows that BBPP+ performs best.
the RoboCup competition has brought back to attention the classification of objects in a controlled illumination environment. We present a very fast classifier to achieve image segmentation. Our methods are based on t...
详细信息
ISBN:
(纸本)354040550X
the RoboCup competition has brought back to attention the classification of objects in a controlled illumination environment. We present a very fast classifier to achieve image segmentation. Our methods are based on the machine literature, but adapted to robots equipped with low cost image-capture equipment. We then present new fast methods for object recognition, based on also rapid methods for blob formation. We describe how to extract the skeleton of a polygon and we use this for object recognition.
Support vector machines (SVMs) have been promising methods for classification and regression analysis because of their solid mathematical foundations which convery several salient properties that other methods hardly ...
详细信息
ISBN:
(纸本)9781581137378
Support vector machines (SVMs) have been promising methods for classification and regression analysis because of their solid mathematical foundations which convery several salient properties that other methods hardly provide. However, despite the prominent properties of SVMs, they are not as favored for large-scale datamining as for patternrecognition or machinelearning because the training complexity of SVMs is highly dependent on the size of a data set. Many real-world datamining applications involve millions or billions of data records where even multiple scans of the entire data are too expensive to perform. this paper presents a new method, Clustering-Based SVM (CB-SVM), which is specifically designed for handling very large data sets. CB-SVM applies a hierarchical micro-clustering algorithm that scans the entire data set only once to provide an SVM with high quality samples that carry the statistical summaries of the data such that the summaries maximize the benefit of learningthe SVM. CB-SVM tries to generate the best SVM boundary for very large data sets given limited amount of resources. Our experiments on synthetic and real data sets show that CB-SVM is highly scalable for very large data sets while also generating high classification accuracy. Copyright 2003 ACM.
the proceedings contain 112 papers. the special focus in this conference is on Audio and video-based biometric person authentication. the topics include: Robust face recognition in the presence of clutter;an image pre...
ISBN:
(纸本)9783540403029
the proceedings contain 112 papers. the special focus in this conference is on Audio and video-based biometric person authentication. the topics include: Robust face recognition in the presence of clutter;an image preprocessing algorithm for illumination invariant face recognition;quad phase minimum average correlation energy filters for reduced memory illumination tolerant face authentication;component-based face recognition with 3D morphable models;a comparative study of automatic face verification algorithms on the BANCA database;assessment of time dependency in face recognition;constraint shape model using edge constraint and gabor wavelet based search;expression-invariant 3D face recognition;automatic estimation of a priori speaker dependent thresholds in speaker verification;a bayesian network approach for combining pitch and reliable spectral envelope features for robust speaker verification;cluster-dependent feature transformation for telephone-based speaker verification;LUT-Based adaboost for gender classification;independent component analysis and support vector machine for face feature extraction;real-time emotion recognition using biologically inspired models;a dual-factor authentication system featuring speaker verification and token technology;wavelet-based 2-parameter regularized discriminant analysis for face recognition;face tracking and recognition from stereo sequence;face recognition system using accurate and rapid estimation of facial position and scale;fingerprint enhancement using oriented diffusion filter;fusion of statistical and structural fingerprint classifiers;learning features for fingerprint classification;fingerprint matching with registration pattern inspection and biometric template selection.
the development of an on line computer based classification system for the real time classification of different composites is addressed in this study. Different parameters were collected simultaneously when embedded ...
详细信息
ISBN:
(纸本)354040550X
the development of an on line computer based classification system for the real time classification of different composites is addressed in this study. Different parameters were collected simultaneously when embedded sensors (dielectric, optical fibre, and piezoelectric sensors) were used within two different composite matrices during the curing process. the measurements were used by an algorithm-based software as a logged data file, resulting in to induction of a decision tree. Later, a systematic software is designed based on the rules derived from this decision tree, to recognise the type of composites used in the experiment together withrecognition of their physical and mechanical characteristics. this is a new approach to data acquisition in intelligent materials produced by smart manufacturing system.
mining frequent patterns has been a topic of active research because it is computationally the most expensive step in association rule discovery. In this paper, we discuss the use of compact data structure design for ...
详细信息
the proceedings contain 39 papers. the special focus in this conference is on Unsupervised learning, Matching, Probabilistic Modelling, Segmentation, Grouping, Shape Modelling and Reconstruction. the topics include: S...
ISBN:
(纸本)3540404988
the proceedings contain 39 papers. the special focus in this conference is on Unsupervised learning, Matching, Probabilistic Modelling, Segmentation, Grouping, Shape Modelling and Reconstruction. the topics include: Stochastic search for optimal linear representations of images on spaces with orthogonality constraints;curve matching using the fast marching method;EM algorithm for clustering an ensemble of graphs with comb matching;information force clustering using directed trees;active sampling strategies for multihypothesis testing;likelihood based hierarchical clustering and network topology identification;learning mixtures of tree-unions by minimizing description length;hierarchical annealing for random image synthesis;semi-supervised image segmentation by parametric distributional clustering;path variation and image segmentation;a fast snake segmentation method applied to histopathological sections;a compositionality architecture for perceptual feature grouping;using prior shape and points in medical image segmentation;separating a texture from an arbitrary background using pair wise grey level cooccurrences;surface recovery from 3D point data using a combined parametric and geometric flow approach;curvature vector flow to assure convergent deformable models for shape modelling;a MAP estimation algorithm using IIR recursive filters;estimation of rank deficient matrices from partial observations;contextual and non-combinatorial approach to feature extraction;generalized multi-camera scene reconstruction using graph cuts and graph matching using spectral seriation.
暂无评论