Results 11  20
of
59
The nearest subclass classifier: A compromise between the nearest mean and nearest neighbor classifier
 IEEE Trans on Pattern Analysis and Machine Intelligence
, 2005
"... Abstract We present the Nearest Subclass Classifier (NSC), which is a classification algorithm that unifies the flexibility of the nearest neighbor classifier with the robustness of the nearest mean classifier. The algorithm is based on the Maximum Variance Cluster algorithm and as such it belongs ..."
Abstract

Cited by 10 (0 self)
 Add to MetaCart
Abstract We present the Nearest Subclass Classifier (NSC), which is a classification algorithm that unifies the flexibility of the nearest neighbor classifier with the robustness of the nearest mean classifier. The algorithm is based on the Maximum Variance Cluster algorithm and as such it belongs to the class of prototypebased classifiers. The variance constraint parameter of the cluster algorithm serves to regularise the classifier, that is, to prevent overfitting. With a low variance constraint value the classifier turns into the nearest neighbor classifier and with a high variance parameter it becomes the nearest mean classifier with the respective properties. In other words, the number of prototypes ranges from the whole training set to only one per class. In the experiments, we compared the NSC with regard to its performance and data set compression ratio to several other prototypebased methods. On several data sets the NSC performed similarly to the knearest neighbor classifier, which is a wellestablished classifier in many domains. Also concerning storage requirements and classification speed, the NSC has favorable properties, so it gives a good compromise between classification performance and efficiency.
Boosting interval based literals
, 2001
"... A supervised classification method for time series, even multivariate, is presented. It is based on boosting very simple classifiers: clauses with one literal in the body. The background predicates are based on temporal intervals. Two types of predicates are used: i) relative predicates, such as “ ..."
Abstract

Cited by 6 (4 self)
 Add to MetaCart
A supervised classification method for time series, even multivariate, is presented. It is based on boosting very simple classifiers: clauses with one literal in the body. The background predicates are based on temporal intervals. Two types of predicates are used: i) relative predicates, such as “increases” and “stays”, and ii) region predicates, such as “always” and “sometime”, which operate over regions in the domain of the variable. Experiments on different data sets, several of them obtained from the UCI ML and KDD repositories, show that the proposed method is highly competitive with previous approaches.
Learning Fuzzy Classifiers with Evolutionary Algorithms
, 2001
"... This paper illustrated an evolutionary algorithm which learns classifiers, represented as sets of fuzzy rules, from a data set containing past experimental observations of a phenomenon. The approach is applied to a benchmark dataset made available by the machine learning community. ..."
Abstract

Cited by 5 (3 self)
 Add to MetaCart
This paper illustrated an evolutionary algorithm which learns classifiers, represented as sets of fuzzy rules, from a data set containing past experimental observations of a phenomenon. The approach is applied to a benchmark dataset made available by the machine learning community.
Simple Learning Algorithms for Training Support Vector Machines
, 1998
"... Support Vector Machines (SVMs) have proven to be highly effective for learning many real world datasets but have failed to establish themselves as common machine learning tools. This is partly due to the fact that they are not easy to implement, and their standard implementation requires the use of ..."
Abstract

Cited by 4 (0 self)
 Add to MetaCart
Support Vector Machines (SVMs) have proven to be highly effective for learning many real world datasets but have failed to establish themselves as common machine learning tools. This is partly due to the fact that they are not easy to implement, and their standard implementation requires the use of optimization packages. In this paper we present simple iterative algorithms for training support vector machines which are easy to implement and guaranteed to converge to the optimal solution. Furthermore we provide a technique for automatically finding the kernel parameter and best learning rate. Extensive experiments with real datasets are provided showing that these algorithms compare well with standard implementations of SVMs in terms of generalisation accuracy and computational cost, while being significantly simpler to implement. 1 Introduction Since their introduction by Vapnik and coworkers [38, 7], Support Vector Machines (SVMs) have been successfully applied to a number of real ...
Cooperative Evolution of a Neural Classifier and Feature Subset
 Second AsiaPacific Conference on Simulated Evolution and Learning (SEAL’98
, 1999
"... . This paper describes a novel feature selection algorithm which utilizes a genetic algorithm to select a feature subset in conjunction with the weights for a threelayer feedforward network classifier. On the "ionosphere " data set from UC Irvine, this approach produces results comparable ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
. This paper describes a novel feature selection algorithm which utilizes a genetic algorithm to select a feature subset in conjunction with the weights for a threelayer feedforward network classifier. On the "ionosphere " data set from UC Irvine, this approach produces results comparable to those reported for other algorithms on the same data, but using fewer input features and a simpler neural network architecture. These results indicate that tailoring a neural network classifier to a specific subset of features has the potential to produce a classifier with low classification error, good generalizability, and relatively low computational overhead. Keywords Genetic algorithm; neural network; classification; ionosphere 1 Introduction Feature selection is the process of selecting an optimum subset of features from the enormous set of potentially useful features available in a given problem domain [2]. The "optimum subset of features" which is the aim of the feature extraction algori...
Adaptive Kernels for Support Vector Classification
, 2002
"... Support vector machines (SVM) are a recent addition to the set of machine learning techniques available to the data miner for classification. The SVM is based on the theory of structural risk minimization [1] and as such has good generalization properties that have been demonstrated both theoretical ..."
Abstract

Cited by 3 (0 self)
 Add to MetaCart
Support vector machines (SVM) are a recent addition to the set of machine learning techniques available to the data miner for classification. The SVM is based on the theory of structural risk minimization [1] and as such has good generalization properties that have been demonstrated both theoretically and empirically. The flexibility of the SVM is provided by the use of kernel functions that implicitly map the data to a higher, possibly infinite, dimensional space. A solution linear in the features in the higher dimensional space corresponds to a solution nonlinear in the original features.
Dimensionality Reduction Using a Novel Neural Network Based Feature Extraction Method
 IJCNN
, 1999
"... A novel neural network based method for feature extraction is proposed. The method achieves dimensionality reduction of input vectors used for supervised learning problems. Combinations of the original features are formed that maximize the sensitivity of the network's outputs with respect to va ..."
Abstract

Cited by 3 (0 self)
 Add to MetaCart
A novel neural network based method for feature extraction is proposed. The method achieves dimensionality reduction of input vectors used for supervised learning problems. Combinations of the original features are formed that maximize the sensitivity of the network's outputs with respect to variations of its inputs. The method exhibits some similarity to Principal Component Analysis, but also takes into account supervised character of the learning task. It is applied to classification problems leading to efficient dimensionality reduction and increased generalization ability. 2. Introduction Methods for dimensionality reduction concentrate either on selecting from the original set of features a smaller subset of salient features, or on combining the original features in such a way as to extract a small number of salient features. Application of such methods to data analysis or pattern recognition problems has distinct advantages in terms of generalization properties and processing s...
CLIP4: Hybrid inductive machine learning algorithm that generates inequality rules
, 2004
"... The paper describes a hybrid inductive machine learning algorithm called CLIP4. The algorithm first partitions data into subsets using a tree structure and then generates production rules only from subsets stored at the leaf nodes. The unique feature of the algorithm is generation of rules that invo ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
The paper describes a hybrid inductive machine learning algorithm called CLIP4. The algorithm first partitions data into subsets using a tree structure and then generates production rules only from subsets stored at the leaf nodes. The unique feature of the algorithm is generation of rules that involve inequalities. The algorithm works with the data that have large number of examples and attributes, can cope with noisy data, and can use numerical, nominal, continuous, and missingvalue attributes. The algorithm's flexibility and e#ciency are shown on several wellknown benchmarking data sets, and the results are compared with other machine learning algorithms. The benchmarking results in each instance show the CLIP4's accuracy, CPU time, and rule complexity. CLIP4 has builtin features like tree pruning, methods for partitioning the data (for data with large number of examples and attributes, and for data containing noise), dataindependent mechanism for dealing with missing values, genetic operators to improve accuracy on small data, and the discretization schemes. CLIP4 generates model of data that consists of wellgeneralized rules, and ranks attributes and selectors that can be used for feature selection.
LESS: a ModelBased Classifier for Sparse Subspaces
"... Abstract In this paper we specifically focus on high dimensional data sets for which the number of dimensions is an order of magnitude higher than the number of objects. From a classifier design standpoint, such small sample size problems have some interesting challenges. The first challenge is to f ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
Abstract In this paper we specifically focus on high dimensional data sets for which the number of dimensions is an order of magnitude higher than the number of objects. From a classifier design standpoint, such small sample size problems have some interesting challenges. The first challenge is to find, from all hyperplanes that separate the classes, a separating hyperplane which generalizes well for future data. A second important task is to determine which features are required to distinguish the classes. To attack these problems, we propose the LESS (Lowest Error in a Sparse Subspace) classifier that efficiently finds linear discriminants in a sparse subspace. In contrast with most classifiers for high dimensional data sets, the LESS classifier incorporates a (simple) data model. Further, by means of a regularization parameter the classifier establishes a suitable tradeoff between subspace sparseness and classification accuracy. In the experiments we show how LESS performs on several high dimensional data sets and compare its performance to related stateoftheart classifiers like among others linear ridge regression with the LASSO and the Support Vector Machine. It turns out that LESS performs competitively while using fewer dimensions.
Genetic Algorithm Optimized Feature Transformation –A Comparison with Different Classifiers
 in Proc. of Genetic and Evolutionary Computation
"... Abstract. When using a Genetic Algorithm (GA) to optimize the feature space of pattern classification problems, the performance improvement is not only determined by the data set used, but also depends on the classifier. This work compares the improvements achieved by GAoptimized feature transforma ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
Abstract. When using a Genetic Algorithm (GA) to optimize the feature space of pattern classification problems, the performance improvement is not only determined by the data set used, but also depends on the classifier. This work compares the improvements achieved by GAoptimized feature transformations on several simple classifiers. Some traditional feature transformation techniques, such as Principal Components Analysis (PCA) and Linear Discriminant Analysis (LDA) are also tested to see their effects on the GA optimization. The results based on some realworld data and five benchmark data sets from the UCI repository show that the improvements after GAoptimized feature transformation are in reverse ratio with the original classification rate if the classifier is used alone. It is also shown that performing the PCA and LDA transformations on the feature space prior to the GA optimization improved the final result. 1