Statistical Discriminability Estimation for Pattern Classification Based on Neural Incremental Attribute Learning

2014 ◽  
Vol 5 (2) ◽  
pp. 37-57
Author(s):  
Ting Wang ◽  
Sheng-Uei Guan ◽  
Sadasivan Puthusserypady ◽  
Prudence W. H. Wong
Keyword(s):  
Machine Learning ◽  
Genetic Algorithm ◽  
Learning Approach ◽  
Classification Problems ◽  
Dynamical Feature ◽  
Discrimination Ability ◽  
Machine Learning Approach ◽  
Attribute Learning ◽  
Predictive Metric ◽  
Significant Data

Feature ordering is a significant data preprocessing method in Incremental Attribute Learning (IAL), a novel machine learning approach which gradually trains features according to a given order. Previous research has shown that, similar to feature selection, feature ordering is also important based on each feature's discrimination ability, and should be sorted in a descending order of their discrimination ability. However, such an ordering is crucial for the performance of IAL. As the number of feature dimensions in IAL is increasing, feature discrimination ability also should be calculated in the corresponding incremental way. Based on Single Discriminability (SD), where only the feature discrimination ability is computed, a new filter statistical feature discrimination ability predictive metric, called the Accumulative Discriminability (AD), is designed for the dynamical feature discrimination ability estimation. Moreover, a criterion that summarizes all the produced values of AD is employed with a GA (Genetic Algorithm)-based approach to obtain the optimum feature ordering for classification problems based on neural networks by means of IAL. Compared with the feature ordering obtained by other approaches, the method proposed in this paper exhibits better performance in the final classification results. Such a phenomenon indicates that, (i) the feature discrimination ability should be incrementally estimated in IAL, and (ii) the feature ordering derived by AD and its corresponding approaches are applicable with IAL.

scholarly journals Korpusbasierte Exploration der Variation der nominalen Genitivmarkierung

2014 ◽  
Vol 42 (3) ◽  
Author(s):  
Noah Bubenhofer ◽  
Sandra Hansen-Morath ◽  
Marek Konopka
Keyword(s):  
Machine Learning ◽  
Decision Trees ◽  
Explanatory Power ◽  
Learning Approach ◽  
Machine Learning Approach ◽  
Singular Noun ◽  
Corpus Data ◽  
Significant Data

AbstractThe variation of the strong genitive marker of the singular noun has been treated by diverse accounts. Still there is a consensus that it is to a large extent systematic but can be approached appropriately only if many heterogeneous factors are taken into account. Over thirty variables influencing this variation have been proposed. However, it is actually unclear how effective they can be, and above all, how they interact. In this paper, the potential influencing variables are evaluated statistically in a machine learning approach and modelled in decision trees in order to predict the genitive marking variants. Working with decision trees based exclusively on statistically significant data enables us to determine what combination of factors is decisive in the choice of a marking variant of a given noun. Consequently the variation factors can be assessed with respect to their explanatory power for corpus data and put in a hierarchized order.

OPTIMIZED NEURAL INCREMENTAL ATTRIBUTE LEARNING FOR CLASSIFICATION BASED ON STATISTICAL DISCRIMINABILITY

2014 ◽  
Vol 13 (04) ◽  
pp. 1450019 ◽  
Author(s):  
TING WANG ◽  
SHENG-UEI GUAN ◽  
KA LOK MAN ◽  
T. O. TING ◽  
ALEXEI LISITSA
Keyword(s):  
Neural Networks ◽  
Data Preprocessing ◽  
Ability Estimation ◽  
Classification Problems ◽  
Discrimination Ability ◽  
Time Consumption ◽  
Preprocessing Method ◽  
Attribute Learning ◽  
Lower Dimensional ◽  
Significant Data

Feature ordering is a significant data preprocessing method in incremental attribute learning (IAL), where features are gradually trained according to a given order. Previous research showed feature ordering is crucial to the IAL performance. It is relevant to each feature's discrimination ability, which can be calculated by single discriminability (SD). However, when feature dimensions increase, feature discrimination ability should also be calculated incrementally, because discrimination ability in lower dimensional spaces is different from that in higher spaces. Thus based on SD, accumulative discriminability (AD), a new statistical metric for incremental feature discrimination ability estimation, is designed. Moreover, a criterion that summarizes all the produced values of AD is employed to obtain the optimum feature ordering for classification problems based on neural networks by means of IAL. In addition, in order to reduce the time consumption, an effective feature ordering approach is developed. Compared with the feature ordering obtained by other approaches, the method outlined in this paper obtained good final classification results, which indicates that, firstly, feature discrimination ability should be incrementally estimated in IAL; and secondly, feature ordering derived by AD and its corresponding approaches are applicable with IAL.

A Genetic Algorithm-Based, Hybrid Machine Learning Approach to Model Selection

2006 ◽  
Vol 33 (2) ◽  
pp. 195-221 ◽  
Author(s):  
Robert R. Bies ◽  
Matthew F. Muldoon ◽  
Bruce G. Pollock ◽  
Steven Manuck ◽  
Gwenn Smith ◽  
...  
Keyword(s):  
Machine Learning ◽  
Genetic Algorithm ◽  
Model Selection ◽  
Learning Approach ◽  
Machine Learning Approach ◽  
Hybrid Machine

scholarly journals Development and Initial Validation of Two Brief Measures of Left-wing Authoritarianism: A Machine Learning Approach

2021 ◽  
Author(s):  
Thomas H Costello ◽  
Christopher Patrick
Keyword(s):  
Machine Learning ◽  
Genetic Algorithm ◽  
Political Violence ◽  
Community Sample ◽  
Learning Approach ◽  
Left Wing ◽  
Initial Validation ◽  
Machine Learning Approach ◽  
Automated Machine Learning ◽  
Political Spectrum

Although authoritarianism has predominantly been studied among conservatives, newer work on left-wing authoritarianism (LWA) has suggested that authoritarian individuals exist on both poles of the political spectrum. A 39-item multidimensional measure, the Left-wing Authoritarianism Index, was recently developed to measure LWA. The present study used a fully automated machine learning approach (i.e., a genetic algorithm) in a large, demographically representative American sample (N = 834) to generate two abbreviated versions of the LWA Index. A second community sample (N = 477) was used to conduct extensive validational tests of the abbreviated measures, which comprise 25- and 13-items. The abbreviated forms demonstrated remarkable convergence with the full LWA Index in terms of their psychometric (e.g., internal consistency) and distributional (e.g., mean, standard deviation, skew, kurtosis) properties. Further, this convergence extended to virtually identical cross-measure patterns of correlations with 14 external criteria, including need for chaos, political violence, anomia, and low institutional trust. In light of these results, the LWA-25 and LWA-13 scales appeared to function effectively as measures of LWA. We conclude by examining the items retained (vs. excluded) by the genetic algorithm to clarify the central vs. peripheral conceptual elements of LWA.

Sign in / Sign up

Export Citation Format

Share Document