检索结果-内蒙古大学图书馆

25th IEEE International Conference on Tools with Artificial Intelligence (ICTAI)

作者： Wald, Randall Khoshgoftaar, Taghi Napolitano, Amri Florida Atlantic Univ Boca Raton FL 33431 USA

ISBN: (纸本)9781479929719

Due to the problem of high-dimensionality (datasets which contain many independent attributes or features), feature selection has become an important part of data mining research. One popular form of feature selection, wrapper selection, chooses the best features by directly addressing the question of which features build the best models. Various feature subsets are used to build classification models, and the performance of these models is the score of each feature subset. The feature subset with the best score is then used to build the final classification model. As wrappers use a classification algorithm (learner) both to select the features and to build a predictive model, it has been traditional to use the same learner for both, such that the features chosen will be those which optimize that model's performance. However, no research has considered whether having different learners operate inside and outside the wrapper (that is, for selecting the features and for building the final model) might actually result in improved classification performance. In this work, we consider five learners both inside the wrapper and for building the classification model, along with two datasets drawn from the domain of Twitter profile mining. By considering both the raw performance values and a statistical analysis, we find that contrary to intuition, usually the best performance for a given choice of external learner is not found by using the same learner within the wrapper. Instead, the Naive Bayes learner is usually the best choice for selecting features, regardless of which learner is used for the external model. We also find that Multi-Layer Perceptron is able to build consistent classification models for many different choices of internal learner. Finally, the 5-Nearest Neighbor learner gave poor results both inside and outside the wrapper.

关键词： Wrapper feature selection classification algorithm learner Twitter social bots

来源：评论

学校读者我要写书评

暂无评论

METHODS FOR classification OF NOCTURNAL MIGRATORY BIRD VOCALIZATIONS USING PSEUDO WIGNER-VILLE TRANSFORM

METHODS FOR CLASSIFICATION OF NOCTURNAL MIGRATORY BIRD VOCAL...

引用

IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Patti, Anand Williamson, Geoffrey A. IIT Chicago IL 60616 USA

ISBN: (纸本)9781479903566

Many species of birds in Americas vocalize during nocturnal migration flights. Acoustic detection and classification of the calls show potential for study of the natural history of these migrant birds. In particular, information about the species' composition and number of birds involved in migration movements may be obtainable through acoustic techniques. Other methods such as radar monitoring may have capability only to assess the number, but not the composition. Mel Frequency Cepstral Coefficients-Gaussian Mixture Model-based methods (MFCC-GMM), Mel Frequency Cepstral Coefficients-Hidden Markov Model-based methods (MFCC-HMM) and spectrogram correlation-based methods have been proposed to automate the recognition/classification of the nocturnal flight calls. Here we investigate the choice of Pseudo Wigner-Ville Transform (PWVT) on MFCC-HMM-based classifier and correlation-based classifier performance. We use a collection of recordings of nocturnal flight calls of several species of thrushes and other bird species with similar calls to evaluate and compare classifiers.

关键词： Acoustic signal detection spectrogram classification algorithm

来源：评论

学校读者我要写书评

暂无评论

FAST AND IMPROVED HAND classification USING DIMENSIONALITY REDUCTION AND TEST SET REDUCTION

FAST AND IMPROVED HAND CLASSIFICATION USING DIMENSIONALITY R...

引用

IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Boughnim, Nabil Marot, Julien Fossati, Caroline Bourennane, Salah Guerault, Frederic Ecole Cent Marseille Inst Fresnel Grp GSM Domaine Univ St Jerome Av Escadrille Normandie F-13397 Marseille France Intui Sense Technol F-13400 Aubagne France

ISBN: (纸本)9781479903566

In this paper, we consider an issue of hand posture classification. We improve a recently proposed signature, a matrix containing the distance of all contour pixels to an arbitrary reference point. Adequate pre-processings ensure the invariance properties of the signature. Candidate postures are pre-selected with a surface criterion, and Principal Component Analysis (PCA) reduces the dimensionality of the data, which improves the classification process.

关键词： Hand posture hand recognition classification algorithm principal component analysis biometrics

来源：评论

学校读者我要写书评

暂无评论

Using Pattern Position Distribution for Software Failure Detection

引用

INTERNATIONAL JOURNAL OF COMPUTATIONAL INTELLIGENCE SYSTEMS 2013年第2期6卷 234-243页

作者： Li, Chunping Chen, Ziniu Du, Hao Wang, Hui Wilkie, George Augusto, Juan C. Liu, Jun Tsinghua Univ Sch Software Tsinghua Natl Lab Informat Sci & Technol Beijing 100084 Peoples R China Univ Ulster Fac Comp & Engn Jordanstown BT37 0QB North Ireland

We present a novel approach for using the pattern position distribution as features to detect software failure. In this approach, we divide an execution sequence into several sections and compute the pattern distribution in each section. The distribution of all patterns is then used as features to train a classifier. This approach outperforms conventional frequency based methods by more effectively identifying software failures occurring through misused software patterns. Comparative experiments show the effectiveness of our approach.

关键词： Sequential Patterns classification algorithm Software Failure Anomaly Detection

来源：评论

学校读者我要写书评

暂无评论

Design of a testing strategy using non-animal based test methods: Lessons learnt from the ACuteTox project

引用

TOXICOLOGY IN VITRO 2013年第4期27卷 1395-1401页

作者： Kopp-Schneider, Annette Prieto, Pilar Kinsner-Ovaskainen, Agnieszka Stanzel, Sven German Canc Res Ctr Dept Biostat D-69009 Heidelberg Germany European Commiss Joint Res Ctr Inst Hlth & Consumer Protect I-21027 Ispra VA Italy

In the framework of toxicology, a testing strategy can be viewed as a series of steps which are taken to come to a final prediction about a characteristic of a compound under study. The testing strategy is performed as a single-step procedure, usually called a test battery, using simultaneously all information collected on different endpoints, or as tiered approach in which a decision tree is followed. Design of a testing strategy involves statistical considerations, such as the development of a statistical prediction model. During the EU FP6 ACuteTox project, several prediction models were proposed on the basis of statistical classification algorithms which we illustrate here. The final choice of testing strategies was not based on statistical considerations alone. However, without thorough statistical evaluations a testing strategy cannot be identified. We present here a number of observations made from the statistical viewpoint which relate to the development of testing strategies. The points we make were derived from problems we had to deal with during the evaluation of this large research project. A central issue during the development of a prediction model is the danger of overfitting. Procedures are presented to deal with this challenge. (C) 2012 Elsevier Ltd. All rights reserved.

关键词： classification algorithm Overfitting Prediction model Prediction error Testing strategy

来源：评论

学校读者我要写书评

暂无评论

Hand posture recognition using jointly optical flow and dimensionality reduction

引用

EURASIP JOURNAL ON ADVANCES IN SIGNAL PROCESSING 2013年第1期2013卷 1-22页

作者： Boughnim, Nabil Marot, Julien Fossati, Caroline Bourennane, Salah Aix Marseille Univ CNRS UMR 7249 Fresnel InstEcole Cent MarseilleDU St Jerome F-13397 Marseille 20 France

Hand posture recognition is generally addressed by using either YCbCr (luminance and chrominance components) or HSV (hue, saturation, value) mappings which assume that a hand can be distinguished from the background from some colorfulness and luminance properties. This can hardly be used when a dark hand, or a hand of any color, is under study. In addition, existing recognition processes rely on descriptors or geometric shapes which can be reliable;this comes at the expense of an increased computational complexity. To cope with these drawbacks, this paper proposes a four-step method recognition technique consisting of (i) a pyramidal optical flow for the detection of large movements and hence determine the region of interest containing the expected hand, (ii) a preprocessing step to compute the hand contour while ensuring geometric and illumination invariance, (iii) an image scanning method providing a signature which characterizes non-star-shaped contours with a one-pixel precision, and (iv) a posture classification method where a sphericity criterion preselects a set of candidate postures, principal component analysis reduces the dimensionality of the data, and Mahalanobis distance is used as a criterion to identify the hand posture in any test image. The proposed technique has been assessed in terms of its performances including the computational complexity using both visual and statistical results.

关键词： Hand posture Contour signature classification algorithm Optical flow Principal component analysis

来源：评论

学校读者我要写书评

暂无评论

Interest Mining algorithm based on Blog Information

Interest Mining Algorithm based on Blog Information

引用

3rd International Conference on Applied Mechanics, Materials and Manufacturing (ICAMMM 2013)

作者： Ou, Guohua Xu, Changjian Zhan, Haoxun Qin, Yong Huang, Han South China Univ Technol Sch Software Engn Guang Zhou Guang Dong Peoples R China Dongguan Univ Technol Sch Comp Dong Guan Guangdong Peoples R China

ISBN: (纸本)9783037858882

Recently, the amount of blogs on the Internet rises sharply. Hence, mining valuable information in blogs possesses realistic significance for improving user experience, network services, etc. This paper proposes a mining algorithm with blog authors' interests based on classification techniques, which introduces an evaluation standard of non-empty intersection. This algorithm can also improve the hit ratio of recommendation service based on blog authors' interests by means of the interest collection from expanding prediction;therefore, it can reach a higher degree of satisfaction. In addition, this paper performs experiments with the data set from Sina Blog and NetEase Blog, whose result illustrates the higher accuracy of our algorithm.

关键词： Blog mining Interest mining classification algorithm Non-empty intersection

来源：评论

学校读者我要写书评

暂无评论

The Coal Production Anomaly Detection Based on Data Mining

The Coal Production Anomaly Detection Based on Data Mining

引用

International Conference on Measurement, Instrumentation and Automation (ICMIA 2012)

作者： Wang Guanghui Kuang Yali Wang Zhangguo China Univ Min & Technol Sch Chem Engn & Technol Xuzhou 221008 Peoples R China

ISBN: (纸本)9783037855454

Choose data Mining to study the anomaly detection in coal preparation, using ash of raw coal, rapid ash and yields of raw coal which density below 1.45, and ash and actual yields of fine coal in the database as sample attribute of coal production anomaly detection model, based on Box-plot analysis, the evaluating values range of five attribute above are determined. On this condition, by using SVM and KNN, the identification model of anomaly detection in coal preparation is established. The Receiver Operating Characteristic curves analysis result shows judging production target Abnormal Conditions using SVM will be more accurate in coal preparation.

关键词： Anomaly Detection classification algorithm Coal Preparation KNN SVM

来源：评论

学校读者我要写书评

暂无评论

An Evolutionary Data-Conscious Artificial Immune Recognition System 13

An Evolutionary Data-Conscious Artificial Immune Recognition...

引用

15th Genetic and Evolutionary Computation Conference (GECCO)

作者： Tay, Darwin Poh, Chueh Loo Kitney, Richard I. Nanyang Technol Univ Div Bioengn Singapore 639798 Singapore Imperial Coll London Dept Bioengn London SW7 2AZ England

ISBN: (纸本)9781450319638

Artificial Immune Recognition System (AIRS) algorithm offers a promising methodology for data classification. It is an immune-inspired supervised learning algorithm that works efficiently and has shown comparable performance with respect to other classifier algorithms. For this reason, it has received escalating interests in recent years. However, the full potential of the algorithm was yet unleashed. We proposed a novel algorithm called the evolutionary data-conscious AIRS (EDC-AIRS) algorithm that accentuates and capitalizes on 3 additional immune mechanisms observed from the natural immune system. These mechanisms are associated to the phenomena exhibited by the antibodies in response to the concentration, location and type of foreign antigens. Bio-mimicking these observations empower EDC-AIRS algorithm with the ability to robustly adapt to the different density, distribution and characteristics exhibited by each data class. This provides competitive advantages for the algorithm to better characterize and learn the underlying pattern of the data. Experiments on four widely used benchmarking datasets demonstrated promising results - outperforming several state-of-the-art classification algorithms evaluated. This signifies the importance of integrating these immune mechanisms as part of the learning process.

关键词： Artificial immune recognition system classification algorithm Evolutionary computation

来源：评论

学校读者我要写书评

暂无评论

Predicting Preferred Topics of Authors based on Co-Authorship Network

Predicting Preferred Topics of Authors based on Co-Authorshi...

引用

IEEE RIVF International Conference on Computing and Communication Technologies - Research, Innovation, and Vision for the Future (RIVF)

作者： Nguyen Le Hoang Pham Vu Dang Khoa Do Phuc Univ Informat Technol VNU HCMC Ho Chi Minh City Vietnam

ISBN: (纸本)9781479913497;9781479913503

This paper focuses a common question in Social Network Analysis - evaluating how much a person prefers or non-prefers a specific issue. To realize this problem, we use the ILPnet2 database and model it as a co-authorship network in which the graph's nodes represent the authors and the links between two nodes means the two corresponding authors have some common papers. And what we have to do is predicting the preferred topics of authors in this network. Based on the original algorithm in [8], we propose a general algorithm with some basic assumptions and definitions and apply it to solve our problem. Finally, we use the ROC Analysis and Regression Estimation model to evaluate the Degree of Accuracy of the algorithm

关键词： social network ILPnet2 co-authorship classification algorithm relaxation labelling predicting preferred topic ROC analysis

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：