PERFORMANCE OF MEE OVER TDNN IN A TIME SERIES PREDICTION
... For this the time-delay neural network with back propagation learning is used to predict the time series. The primary objective is to implement the MEE cost function over TDNN and verify the improvement this would provide over the traditional MSE cost function. The idea behind using the MEE is to up ...
... For this the time-delay neural network with back propagation learning is used to predict the time series. The primary objective is to implement the MEE cost function over TDNN and verify the improvement this would provide over the traditional MSE cost function. The idea behind using the MEE is to up ...
Prediction of maximum surface settlement caused by earth pressure
... Underground tunneling for the development of underground railway lines as a rapid, clean, and efficient way to transport passengers in megacities has received a great deal of attention. Since such tunnels are generally excavated beneath important structures in urban zones, estimating the surface set ...
... Underground tunneling for the development of underground railway lines as a rapid, clean, and efficient way to transport passengers in megacities has received a great deal of attention. Since such tunnels are generally excavated beneath important structures in urban zones, estimating the surface set ...
77
... empirical risk minimization, whilst SVMs use structural risk minimization. The reason that SVMs often outperform Bayesian net models in practice is that they deal with the biggest problem with Bayesian net models, SVMs are less prone to overfitting". In contrast to Bayesian Net modelling, SVMs autom ...
... empirical risk minimization, whilst SVMs use structural risk minimization. The reason that SVMs often outperform Bayesian net models in practice is that they deal with the biggest problem with Bayesian net models, SVMs are less prone to overfitting". In contrast to Bayesian Net modelling, SVMs autom ...
Book Recommending Using Text Categorization
... spelling errors, and there is a wide amount of variance in the length and quality of book descriptions. Performance Measures To evaluate performance, we ran 10-fold cross-validation and examined two performance measures, binary classification accuracy and Spearman’s rank correlation coefficient (r,) ...
... spelling errors, and there is a wide amount of variance in the length and quality of book descriptions. Performance Measures To evaluate performance, we ran 10-fold cross-validation and examined two performance measures, binary classification accuracy and Spearman’s rank correlation coefficient (r,) ...
Analysis and Improvement of Multiple Optimal Learning Factors for
... The K-fold validation procedure is used to calculate the average training and validation errors. ...
... The K-fold validation procedure is used to calculate the average training and validation errors. ...
comparison of purity and entropy of k-means
... The K-means algorithm is implemented and the respective clusters are obtained. These clusters are compared with the true label data set and the values of Purity and Entropy calculated clusters generated by K-means clusters and Fuzzy C means. The implementation is done in PERL. The Fuzzy C means is i ...
... The K-means algorithm is implemented and the respective clusters are obtained. These clusters are compared with the true label data set and the values of Purity and Entropy calculated clusters generated by K-means clusters and Fuzzy C means. The implementation is done in PERL. The Fuzzy C means is i ...
- White Rose Research Online
... the bagging prediction models on imbalanced data-sets. Most research on existing bagging-based sampling schemes for imbalanced data, e.g. (Li 2007; Hido, Kashima, and Takahashi 2009), focused on using sampling methods to provide a set of equally balanced or average-balanced training sub-sets for tra ...
... the bagging prediction models on imbalanced data-sets. Most research on existing bagging-based sampling schemes for imbalanced data, e.g. (Li 2007; Hido, Kashima, and Takahashi 2009), focused on using sampling methods to provide a set of equally balanced or average-balanced training sub-sets for tra ...
Improved Gaussian Mixture Density Estimates Using Bayesian
... In this section we discuss the averaging of several Gaussian mixtures to yield improved probability density estimation. The averaging over neural network ensembles has been applied previously to regression and classification tasks ([PC93]) . There are several different variants on the simple averagi ...
... In this section we discuss the averaging of several Gaussian mixtures to yield improved probability density estimation. The averaging over neural network ensembles has been applied previously to regression and classification tasks ([PC93]) . There are several different variants on the simple averagi ...
Learning with Perceptrons and Neural Networks
... – Which weights have greatest effect on error? – Effectively, partial derivatives of error wrt weights • In turn, depend on other weights => chain rule ...
... – Which weights have greatest effect on error? – Effectively, partial derivatives of error wrt weights • In turn, depend on other weights => chain rule ...
A New Ensemble Model based Support Vector Machine for
... features. However, boosting generates diversity by combing a sequence of classifiers. Marina compare three combining techniques, i.e., bagging, random subspace and boosting, with linear discriminant analysis carried out for several data sets [23]. Though ensemble methods often perform better than a ...
... features. However, boosting generates diversity by combing a sequence of classifiers. Marina compare three combining techniques, i.e., bagging, random subspace and boosting, with linear discriminant analysis carried out for several data sets [23]. Though ensemble methods often perform better than a ...
NeuralNets
... Perceptron as Hill Climbing • The hypothesis space being search is a set of weights and a threshold. • Objective is to minimize classification error on the training set. • Perceptron effectively does hill-climbing (gradient descent) in this space, changing the weights a small amount at each point t ...
... Perceptron as Hill Climbing • The hypothesis space being search is a set of weights and a threshold. • Objective is to minimize classification error on the training set. • Perceptron effectively does hill-climbing (gradient descent) in this space, changing the weights a small amount at each point t ...
Large-scale attribute selection using wrappers
... forward selections, one for each of the training sets in the cross-validation. The training data is used to decide which attribute is added in each iteration of forward selection, and the test data is only used to evaluate the “best” m best subsets of a particular size. To determine the “optimal” su ...
... forward selections, one for each of the training sets in the cross-validation. The training data is used to decide which attribute is added in each iteration of forward selection, and the test data is only used to evaluate the “best” m best subsets of a particular size. To determine the “optimal” su ...
Large-scale attribute selection using wrappers
... forward selections, one for each of the training sets in the cross-validation. The training data is used to decide which attribute is added in each iteration of forward selection, and the test data is only used to evaluate the “best” m best subsets of a particular size. To determine the “optimal” su ...
... forward selections, one for each of the training sets in the cross-validation. The training data is used to decide which attribute is added in each iteration of forward selection, and the test data is only used to evaluate the “best” m best subsets of a particular size. To determine the “optimal” su ...
Modeling the probability of a binary outcome
... because maximum likelihood coefficients are large sample estimates. A minimum of 50 cases per predictor is recommended. ...
... because maximum likelihood coefficients are large sample estimates. A minimum of 50 cases per predictor is recommended. ...
Prediction of pedal cyclists and pedestrian fatalities from
... Artificial Neural Network (ANN) is used for modeling the statistical data. This method is a reliable way of representing the non-linear relations between the inputs and outputs of a system. ANN can consider the complex relations inside the data, and it tries to generalize. A typical ANN model can be ...
... Artificial Neural Network (ANN) is used for modeling the statistical data. This method is a reliable way of representing the non-linear relations between the inputs and outputs of a system. ANN can consider the complex relations inside the data, and it tries to generalize. A typical ANN model can be ...
Initialization of Big Data Clustering
... Fig. 3: Initialization and search phases wall times for parallellized Algorithm 1. gorithm 1 occasionally gives smaller errors than the repeated, full K-means++, especially for the smaller values of k. A strong variation of the SSE difference for the dataset S1 is most likely a consequence of higher ...
... Fig. 3: Initialization and search phases wall times for parallellized Algorithm 1. gorithm 1 occasionally gives smaller errors than the repeated, full K-means++, especially for the smaller values of k. A strong variation of the SSE difference for the dataset S1 is most likely a consequence of higher ...
PDF
... where the +1 element denotes the position of the output within the time series, whereas the negative values in the mask denote the relative positions of the three inputs. The FIR modeling engine searches through all possible masks up to a given mask depth, creating for each mask an input/output tabl ...
... where the +1 element denotes the position of the output within the time series, whereas the negative values in the mask denote the relative positions of the three inputs. The FIR modeling engine searches through all possible masks up to a given mask depth, creating for each mask an input/output tabl ...
Tutorial on Pattern Classification in Cell Recording
... identities (see, for example, chapters 7 and 10), different object categories, different object positions or viewpoints, the same objects under different experimental manipulations (e.g., attention/no attention) (see, for example, chapter 3), and so on. In order for population decoding methods to wo ...
... identities (see, for example, chapters 7 and 10), different object categories, different object positions or viewpoints, the same objects under different experimental manipulations (e.g., attention/no attention) (see, for example, chapter 3), and so on. In order for population decoding methods to wo ...
Neural Network Approach to Predict Quality of Data Warehouse
... proposed neural network and the criteria on which the performance of the proposed neural network is evaluated. A. Architecture/Learning algorithm The general architecture of the present NN model shown in figure 2 is described in this section. The model can be viewed as a directed graph composed of n ...
... proposed neural network and the criteria on which the performance of the proposed neural network is evaluated. A. Architecture/Learning algorithm The general architecture of the present NN model shown in figure 2 is described in this section. The model can be viewed as a directed graph composed of n ...
Improving DCNN Performance with Sparse Category
... the entire training set. For each neuron j, we first calculate the L2 norm of its response values for all the training samples belonging to category !k . Then the L2 norm values across all the categories !1 , . . . , !K of the training set are added up together to achieve the L2,1 norm computation. ...
... the entire training set. For each neuron j, we first calculate the L2 norm of its response values for all the training samples belonging to category !k . Then the L2 norm values across all the categories !1 , . . . , !K of the training set are added up together to achieve the L2,1 norm computation. ...
072-31
... to assemble more stable results through averaging. This can help offset the risk of widely-varying results that could be produced by single-model estimates. Boosting, in particular, addresses the problems that are produced in trying to work with rare instances in data. In a sense, boosting does what ...
... to assemble more stable results through averaging. This can help offset the risk of widely-varying results that could be produced by single-model estimates. Boosting, in particular, addresses the problems that are produced in trying to work with rare instances in data. In a sense, boosting does what ...
Attribute Selection in Software Engineering Datasets for Detecting
... Forward selection is much faster than backward elimination and therefore scales better to large data sets. A wide range of search strategies can be used: best–first, branch– and–bound, simulated annealing, genetic algorithms (see Kohavi and John [10] for a review). In [4], different search strategie ...
... Forward selection is much faster than backward elimination and therefore scales better to large data sets. A wide range of search strategies can be used: best–first, branch– and–bound, simulated annealing, genetic algorithms (see Kohavi and John [10] for a review). In [4], different search strategie ...
A Comparative Study of Classification Methods for Microarray Data
... whether LibSVM and another method is better. Though Table 2 give a large average accuracy difference between an ensemble method and LibSVM, we do not know wether LibSVM and an ensemble method will perform better on a data set. This is because that SVM and decision trees are two different types of cl ...
... whether LibSVM and another method is better. Though Table 2 give a large average accuracy difference between an ensemble method and LibSVM, we do not know wether LibSVM and an ensemble method will perform better on a data set. This is because that SVM and decision trees are two different types of cl ...
DATA MINING IN FINANCE AND ACCOUNTING: A - delab-auth
... ANOVA. The authors report better results for the NNs and decision trees models for both the human judgment based and the ANOVA feature selection. Finally, the authors propose a hybrid algorithm employing weighted voting of different classifiers. Marginally better performance is reported for the hyb ...
... ANOVA. The authors report better results for the NNs and decision trees models for both the human judgment based and the ANOVA feature selection. Finally, the authors propose a hybrid algorithm employing weighted voting of different classifiers. Marginally better performance is reported for the hyb ...
Analogy-based Reasoning With Memory Networks - CEUR
... function l(el , er ) = zTl M zr , where zl and zr are the concatenated word embeddings xs , xvl , xo and xs , xvr , xo , respectively, and parameter matrix M ∈ R3d×3d . We denote this model as Bai2009. We also test three neural network architecture that were proposed in different contexts. The model ...
... function l(el , er ) = zTl M zr , where zl and zr are the concatenated word embeddings xs , xvl , xo and xs , xvr , xo , respectively, and parameter matrix M ∈ R3d×3d . We denote this model as Bai2009. We also test three neural network architecture that were proposed in different contexts. The model ...