Evaluation Of Individual Features Research Articles

Initialization of Feature Selection Search for Classification

Selecting the best features in a dataset improves accuracy and efficiency of classifiers in a learning process. Datasets generally have more features than necessary, some of them being irrelevant or redundant to others. For this reason, numerous feature selection methods have been developed, in which different evaluation functions and measures are applied. This paper proposes the systematic application of individual feature evaluation methods to initialize search-based feature subset selection methods. An exhaustive review of the starting methods used by genetic algorithms from 2014 to 2020 has been carried out. Subsequently, an in-depth empirical study has been carried out evaluating the proposal for different search-based feature selection methods (Sequential forward and backward selection, Las Vegas filter and wrapper, Simulated Annealing and Genetic Algorithms). Since the computation time is reduced and the classification accuracy with the selected features is improved, the initialization of feature selection proposed in this work is proved to be worth considering while designing any feature selection algorithms.

A Mixed Integer Linear Programming Support Vector Machine for Cost-Effective Group Feature Selection: Branch-Cut-and-Price Approach

Recently, cost-based feature selection has received significant attention due to its great ability to achieve promising prediction accuracy at a minimum feature acquisition cost. To further improve its predictive and economic performances, this research proposes a cost-effective 1-norm support vector machine with group feature selection as GFS-CESVM1. Its robust counterpart model, GFS-RCESVM1, is also introduced to address the cost uncertainty of features and feature groups because cost variation commonly exists in real-world problems. The proposed models are formulated as Mixed Integer Linear Programming (MILP). To efficiently solve the proposed SVM MILP models, we develop a Branch-Cut-and-Price (BCP) algorithm that considers only a limited number of variables and/or constraints, which thereby leads to rapid convergence to an optimal solution. Various experimental results on benchmark and synthetic datasets demonstrate that GFS-CESVM1 can achieve competitive outcomes by considering not only individual feature evaluation but also group structural information among features. The GFS-RCESVM1 can identify the subset of features that is immune to cost uncertainty and therefore provide feasible and optimal solutions. Furthermore, our BCP algorithm can dominantly outperform the ordinary BB algorithm for finding better objective value and integrality gap within a short period of time.

An “all-in-one” wearable application for assisting children with Autism Spectrum Disorder

BACKGROUND: Assistive technologies are playing a vital role in children with Autism Spectrum Disorder. OBJECTIVE: The current paper presents the design and assessment of an “all-in-one” wearable solution incorporating location monitoring, scheduling, communication, and physiological parameters monitoring functionalities for children with autism and their families/carers. METHODS: Twenty high-functioning children and twenty-five parents participated in the research. The final application was assessed over a period of two months. A highly patient-centred design was followed by a multidisciplinary team of experts (experts in the field of special education, speech therapy, psychology, and informatics) with significant experience in children with autism. The evaluation took place in a real and open environment. All participants were administered the PYTHEIA questionnaire in order to assess the various characteristics of the platform. Pearson’s Chi-squared test was used to examine the independence of the variables, with a statistical significance of 95%. RESULTS: The results support that the implemented features can effectively assist everyday life activities and facilitate the independent living of children with autism. The subjective evaluation of the solution revealed important considerations for further developments. Both groups of users (parents and children) were very satisfied with the implemented functionalities. The flexibility given by the PYTHEIA instrument was very important for the subjective assessment of the different functionalities of the solution. All examined axes were found excellent in terms of their adaptability in daily life activities. CONCLUSIONS: Real-world applications, able to run in an open and uncontrolled environment are critical for this population. PYTHEIA is a valuable instrument that supports the assessment of the device and application as a whole, as well as the evaluation of individual features and functions.

Probability Interpretation of Forensic Evidence

A short historical survey was carried out to compare different statistical approaches used in the field of forensics for the evaluation of individual features or sets of features, as well as the results of comparative examinations. Some advantages and disadvantages of each approach are considered, and the most appropriate trend is suggested for the improvement of statistical and mathematical tools of evidence interpretation in forensic practice

Assessing Gait Patterns of Healthy Adults Climbing Stairs Employing Machine Learning Techniques

So far, stair climbing has not been studied as extensively as gait has, although the significance of the prevention of falling on stairs has been well recognized. Based on acceleration data taken from 25 healthy subjects climbing up and down a set of 13 stairs with an accelerometer placed on the lumbo-sacral joint, this paper aims to assess gait patterns of younger and older adults climbing stairs using a machine learning approach. A total of 14 gait features were extracted and analyzed. The performance of six representative classification models: Multilayer Perceptron MLP, KStar, Support Vector Machine SVM, Naive Bayesian NB, C4.5 Decision Trees, and Random Forests were evaluated in terms of their ability to discriminate between younger and older adults climbing up- and downstairs. MLP was found to provide the highest accuracy for classification. Accuracy of 95.7% was found for classifying a subject walking either up or down the stairs and an accuracy of 80.6% for classifying whether the subject was younger or older. An evaluation of individual features showed poor performance of classification for younger and older subjects climbing up- and downstairs, and in most cases failed to distinguish between the two classes. To access which set of features derived from a triaxial accelerometer can better describe the performance differences between younger and older adults climbing up- and downstairs, two feature selection algorithms, sequential feature selection and correlation-based feature selection, were implemented. Results show that 10 features derived from correlation-based feature selection were able to produce a 96.8% accuracy for classification between subjects climbing up and down. A subset of seven features achieved a performance of 84.9% accuracy for classification between younger and older subjects.

Empirical study of feature selection methods based on individual feature evaluation for classification problems

The use of feature selection can improve accuracy, efficiency, applicability and understandability of a learning process and its resulting model. For this reason, many methods of automatic feature selection have been developed. By using a modularization of feature selection process, this paper evaluates a wide spectrum of these methods. The methods considered are created by combination of different selection criteria and individual feature evaluation modules. These methods are commonly used because of their low running time. After carrying out a thorough empirical study the most interesting methods are identified and some recommendations about which feature selection method should be used under different conditions are provided.

Osteoarthritis of the first carpometacarpal joint: a study of radiology and clinical epidemiology: Results from the Copenhagen Osteoarthritis Study

The radiological and epidemiological data from the Copenhagen Osteoarthritis Study (COS) were analysed in order to assess the prevalence of osteoarthritis (OA) of the first carpometacarpal joint (CMCJ). Another aim of the study was to analyse relationships between radiologic CMCJ OA and self-reported pain. The third aim was to analyse if additional information could be obtained applying a new method of correlating individual radiological features to self-reported pain, compared to Kellgren and Lawrence's (K-L's) radiologic OA classification. Between 1992 and 1994 standardised radiographs of both hands were recorded in 3,355 participants of the COS cohort. Subjects with known rheumatoid arthritis, other inflammatory arthritis or earlier fractures of the hand were excluded. OA of the CMCJ was assessed according to K-L's radiologic classification by two senior radiologists at our institution. The radiologists further evaluated individual radiologic features of CMCJ OA as recommended by K-L according to the text attached to each picture in their radiologic atlas of OA. To estimate inter- and intraobserver reproducibility a subset of 100 radiographs was reread. Our analyses demonstrated that the K-L method was not able to classify all X-rays. In 608 (18.1%) cases, combinations of joint space width (JSW) measurements, the graduation of osteofytes, sclerosis and cysts fell outside the classification. The radiological evaluation of individual features of OA demonstrated an acceptable reproducibility, intrapersonal (kappa=0.79) as well as interpersonal (kappa=0.65). The prevalence of each radiological feature increased after the fifth decade, progressively more so among women (P<0.001), with the highest prevalence (36.0%) of grades 3 and 4 JSW reduction among women>80 years. A significant correlation was found between signs of radiologic degeneration and self-reported pain (P<0.001); however, different combinations of OA features had different relations to symptoms. Logistic regression analyses revealed sclerosis to have an independent influence on pain in the thumb compared with the presence of osteofytes, cysts and diminished JSW. Body mass index (BMI) was positively related to radiological changes. In logistic regression analyses BMI did not demonstrate an independent positive relation to OA. Radiological degenerative changes in the CMCJ by age especially among women are quite common. However, it is demonstrated that global radiologic classifications of OA of the CMCJ have serious limitations in epidemiological studies. Not all cases fit into classification based on the K-L-atlas. Among the radiological features, subchondral sclerosis is significantly related to self-reported pain. Specific radiologic data should be incorporated in epidemiological studies on hand OA.

Structure of confusions in short-term memory between English consonants.

Data on confusions in short-term memory between English consonants [W. A. Wickelgren, J. Acoust. Soc. Am. 39, 388–398 (1966)] have been reanalyzed within the framework of binary features. A new method of analysis is described that involves a similarity metric. The method produces an evaluation of individual features and associates a confidence level with each statement made about the data. With a confidence level greater than 0.99, it is shown that long frication, continuant, sonorant, sibilant, and voiced are features present in the data of Wickelgren. The issues of feature independence, completeness of a feature system, and the derivation of optimal features are treated. The advantages of the similarity metric over rank-order statistics, multidimensional scaling, and information transfer are discussed.