Results 1 - 10
of
16
Logistic Regression by Means of Evolutionary Radial Basis Function Neural Networks
"... Abstract — This paper proposes a hybrid multilogistic method-ology, named logistic regression using initial and radial basis function (RBF) covariates. The process for obtaining the co-efficients is carried out in three steps. First, an evolutionary programming (EP) algorithm is applied, in order to ..."
Abstract
-
Cited by 2 (0 self)
- Add to MetaCart
(Show Context)
Abstract — This paper proposes a hybrid multilogistic method-ology, named logistic regression using initial and radial basis function (RBF) covariates. The process for obtaining the co-efficients is carried out in three steps. First, an evolutionary programming (EP) algorithm is applied, in order to produce an RBF neural network (RBFNN) with a reduced number of RBF transformations and the simplest structure possible. Then, the initial attribute space (or, as commonly known as in logistic regression literature, the covariate space) is transformed by adding the nonlinear transformations of the input variables given by the RBFs of the best individual in the final generation. Finally, a maximum likelihood optimization method determines the coefficients associated with a multilogistic regression model built in this augmented covariate space. In this final step, two different multilogistic regression algorithms are applied: one considers all initial and RBF covariates (multilogistic initial-RBF regression) and the other one incrementally constructs the model and applies cross validation, resulting in an automatic covariate selection [simplelogistic initial-RBF regression (SLIRBF)]. Both methods include a regularization parameter, which has been also optimized. The methodology proposed is tested using 18 bench-mark classification problems from well-known machine learning problems and two real agronomical problems. The results are compared with the corresponding multilogistic regression meth-ods applied to the initial covariate space, to the RBFNNs obtained by the EP algorithm, and to other probabilistic classifiers, including different RBFNN design methods [e.g., relaxed variable kernel density estimation, support vector machines, a sparse clas-sifier (sparse multinomial logistic regression)] and a procedure similar to SLIRBF but using product unit basis functions. The SLIRBF models are found to be competitive when compared with the corresponding multilogistic regression methods and the RBFEP method. A measure of statistical significance is used, which indicates that SLIRBF reaches the state of the art.
Study on Efficient Search in Evolutionary Computation
"... Enhancing the search capability of evolutionary computation (EC) and increas-ing its optimization performance are important but have not completed yet. EC is applicable to high dimensional, non-linear, non-differentiable, and/or other hard problems. However, obtaining an optimal performance is still ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
Enhancing the search capability of evolutionary computation (EC) and increas-ing its optimization performance are important but have not completed yet. EC is applicable to high dimensional, non-linear, non-differentiable, and/or other hard problems. However, obtaining an optimal performance is still hard for practical EC applications. For example, user fatigue is a serious issue of applying interactive EC, and reducing fatigue is a practical requirement for its applications. As implementing an efficient search method in EC algorithm is one of the methods for reducing user fatigue, it is valuable to study on the efficient search methods for EC. In this dissertation, we propose six novel approaches on this subject and dis-cuss them within three research directions. They are: (1) approximating fitness landscape in lower dimensional search space and elite local search, (2) Fourier anal-ysis on fitness landscape and its enhancement methods, (3) Fourier niche method for multi-modal optimization, (4) triple and quadruple comparison-based interac-tive differential evolution (IDE) and differential evolution (DE), (5) EC acceleration
PROPOSAL FOR A STRATEGIC PLANNING FOR THE REPLACEMENT OF PRODUCTS IN STORES BASED ON SALES FORECAST
, 2010
"... ABSTRACT. This paper presents a proposal for strategic planning for the replacement of products in stores of a supermarket network. A quantitative method for forecasting time series is used for this, the Artificial Radial Basis Neural Networks (RBFs), and also a qualitative method to interpret the f ..."
Abstract
- Add to MetaCart
ABSTRACT. This paper presents a proposal for strategic planning for the replacement of products in stores of a supermarket network. A quantitative method for forecasting time series is used for this, the Artificial Radial Basis Neural Networks (RBFs), and also a qualitative method to interpret the forecasting results and establish limits for each product stock for each store in the network. The purpose with this strategic planning is to reduce the levels of out-of-stock products (lack of products on the shelves), as well as not to produce overstocking, in addition to increase the level of logistics service to customers. The results were highly satisfactory reducing the Distribution Center (DC) to shop out-of-stock levels, in average, from 12% to about 0.7 % in hypermarkets and from 15 % to about 1.7 % in supermarkets, thereby generating numerous competitive advantages for the company. The use of RBFs for forecasting proved to be efficient when used in conjunction with the replacement strategy proposed in this work, making effective the operational processes.
A Hybrid Framework using RBF and SVM for Direct Marketing
"... Abstract—one of the major developments in machine learning in the past decade is the ensemble method, which finds highly accurate classifier by combining many moderately accurate component classifiers. This paper addresses using an ensemble of classification methods for direct marketing. Direct mark ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract—one of the major developments in machine learning in the past decade is the ensemble method, which finds highly accurate classifier by combining many moderately accurate component classifiers. This paper addresses using an ensemble of classification methods for direct marketing. Direct marketing has become an important application field for data mining. In direct marketing, companies or organizations try to establish and maintain a direct relationship with their customers in order to target them individually for specific product offers or for fund raising. A variety of techniques have been employed for analysis ranging from traditional statistical methods to data mining approaches. In this research work, new hybrid classification method is proposed by combining classifiers in a heterogeneous environment using arcing classifier and their performances are analyzed in terms of accuracy. A Classifier ensemble is designed using Radial Basis Function (RBF) and Support Vector Machine (SVM) as base classifiers. Here, modified training sets are formed by resampling from original training set; classifiers constructed using these training sets and then combined by voting. Empirical results illustrate that the proposed hybrid systems provide more accurate direct marketing system.
Random Forests for Multiclass Classification: Random
"... Abstract—Several supervised learning algorithms are suited to classify instances into a multiclass value space. MultiNomial Logit (MNL) is recognized as a robust classifier and is commonly applied within the CRM (Customer Relationship Management) domain. Unfortunately, to date, it is unable to handl ..."
Abstract
- Add to MetaCart
Abstract—Several supervised learning algorithms are suited to classify instances into a multiclass value space. MultiNomial Logit (MNL) is recognized as a robust classifier and is commonly applied within the CRM (Customer Relationship Management) domain. Unfortunately, to date, it is unable to handle huge feature spaces typical of CRM applications. Hence, the analyst is forced to immerse himself into feature selection. Surprisingly, in sharp contrast with binary logit, current software packages lack any feature selection algorithm for MultiNomial Logit. Conversely, Random Forests, another algorithm learning multi class problems, is ju st like MNL robust but unlike MNL it easily handles high-dimensional feature spaces. This paper investigates the potential of applying the Random Forests principles to the MNL framework. We propose the Random MultiNomial Logit (RMNL), i.e. a random forest of MNLs, and compare its predictive performance to that of a) MNL with expert feature selection, b) Random Forests of classification trees. We illustrate the Random MultiNomial Logit on a cross-sell CRM problem within the home-appliances industry. The results indicate a substantial increase in model accuracy of the RMNL model to that of the MNL model with expert feature selection. Keywords—multiclass classifier design and evaluation, feature evaluation and selection, data mining methods and algorithms, customer relationship management (CRM) 2 M
Evolving an Automatic Defect Classification Tool
"... Abstract. Automatic Defect Classification (ADC) is a well-developed technology for inspection and measurement of defects on patterned wafers in the semiconductors industry. The poor training data and its high dimensionality in the feature space render the defect-classification task hard to solve. In ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract. Automatic Defect Classification (ADC) is a well-developed technology for inspection and measurement of defects on patterned wafers in the semiconductors industry. The poor training data and its high dimensionality in the feature space render the defect-classification task hard to solve. In addition, the continuously changing environment—comprising both new and obsolescent defect types encountered during an imaging machine’s lifetime—require constant human intervention, limiting the technology’s effectiveness. In this paper we design an evolutionary classification tool, based on genetic algorithms (GAs), to replace the manual bottleneck and the limited human optimization capabilities. We show that our GA-based models attain significantly better classification performance, coupled with lower complexity, with respect to the human-based model and a heavy random search model. 1
Model Selection for Support Vector Machines
, 2005
"... The accuracy, efficiency, and robustness of machine learning systems depends mainly on prior knowledge that is being utilized. In Support Vector Machines, prior knowledge is typically incorporated by means of choosing a regularization parameter C and a kernel function, which defines a similarity mea ..."
Abstract
- Add to MetaCart
The accuracy, efficiency, and robustness of machine learning systems depends mainly on prior knowledge that is being utilized. In Support Vector Machines, prior knowledge is typically incorporated by means of choosing a regularization parameter C and a kernel function, which defines a similarity measure between all data points. Since in general it is not always possible to convert prior knowledge into such a form, one uses model selection. In model selection, a model consisting of C and an appropriate kernel function in a space of kernel functions is being automatically selected according to prior assumptions and a performance measure. In this work, we discuss spaces of basic kernel functions, a number of performance measures which estimate the generalization ability of a learning system with a given model, and search strategies which optimize a model with respect to a performance measure. Additionally, we present several heuristics, that can be used to make the search more accurate, more efficient, or more robust. In our experiments we compare all methods that we presented with regard to three benchmark datasets. Based on these insights we develop a model selection system for Support Vector Machines, which we finally compare to a state-of-the-art model selection system for Radial Basis Function networks.
Developing System to Filter Unwanted Texts and Images from Social Network User Wall
"... Abstract — In these days, internet is spreading worldwide. One of the chief purpose of use of internet is online social networks (OSNs). OSNs are used on large basis to market products, promote brands, connect people with their family, friends, & society through online communication. But, due t ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract — In these days, internet is spreading worldwide. One of the chief purpose of use of internet is online social networks (OSNs). OSNs are used on large basis to market products, promote brands, connect people with their family, friends, & society through online communication. But, due to the deficiency of categorization & filtering tools, user acquires all messages which are published on his private space. so, user do not have whole control on the contents that are being posted. So, proposed work explains rule based system that gives permission to users to channelize the filtering criterion applied to their private space, and machine learning classifier able to title the messages automatically based on their content. Unwanted images are possible to filter by extracting and classifying their features like edge, color and shape.
Customizable Content Based Filtering Method to Filter Texts and Images from Social Network User Wall
"... Abstract — Today online social media are most valuable and essential member of human life. People can online communicate with their family, society and friends to exchange several type of information including text, images, audio & video data. Therefore, there is need of control of user over the ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract — Today online social media are most valuable and essential member of human life. People can online communicate with their family, society and friends to exchange several type of information including text, images, audio & video data. Therefore, there is need of control of user over the contents which are published on his wall. Hence, user is able to avoid undesirable content that is displayed on his wall. But, currently social networks provide this service in very small extent. To provide this service, we plan a user defined filtering rules and machine learning categorization technique in this paper. Filtering rules grant users to personalize the filtering norms that are employed to the contents which are published on their walls. Automatic categorization based on content of messages into proposed categories is possible through machine learning technique. Also, proposed system can restrict the undesired images that are published on online social network (OSN) user private space by using content wise image filtering along with K-Nearest Neighbour (KNN) classifier.
