Multi-Valued Classification of Text Data Based on an ECOC Approach Using a Ternary Orthogonal Table

Leona Suzuki,Kenta Mikawa,Masayuki Goto 대한산업공학회 2017 Industrial Engineeering & Management Systems Vol.16 No.2

Because of the advancements in information technology, a large number of document data has been accumulated on various databases and automatic multi-valued classification becomes highly relevant. This paper focuses on a multivalued classification technique that is based on Error-Correcting Output Codes (ECOC) and which combines several binary classifiers. When predicting the category of a new document data, the outputs of the binary classifiers are combined to produce a predicted value. It is a known problem that if two category sets have an imbalanced amount of training data, the prediction accuracy of a binary classifier is low. To solve this problem, a previous study proposed to employ the Reed-Muller (RM) codes in the context an ECOC approach for resolving the imbalance in the cardinality of the training data sets. However, RM codes can equalize the amount of between training data of two category sets only for a specific number of categories. We want to provide a method that can be employed for a multi-valued classification with an arbitrary number of categories. In this paper, we propose a new configuration method combining binary classifiers with categories, which are not used for classification. This method allows us to reduce the amount of training data for each binary classifier while improving the balance of the training data between two category sets for each binary classifier. As a result, the computational complexity can be decreased. We verify the effectiveness of our proposed method by conducting a document classification experiment.

Multi-Valued Classification of Text Data Based on an ECOC Approach Using a Ternary Orthogonal Table

Suzuki, Leona,Mikawa, Kenta,Goto, Masayuki Korean Institute of Industrial Engineers 2017 Industrial Engineeering & Management Systems Vol.16 No.2

Because of the advancements in information technology, a large number of document data has been accumulated on various databases and automatic multi-valued classification becomes highly relevant. This paper focuses on a multi-valued classification technique that is based on Error-Correcting Output Codes (ECOC) and which combines several binary classifiers. When predicting the category of a new document data, the outputs of the binary classifiers are combined to produce a predicted value. It is a known problem that if two category sets have an imbalanced amount of training data, the prediction accuracy of a binary classifier is low. To solve this problem, a previous study proposed to employ the Reed-Muller (RM) codes in the context an ECOC approach for resolving the imbalance in the cardinality of the training data sets. However, RM codes can equalize the amount of between training data of two category sets only for a specific number of categories. We want to provide a method that can be employed for a multi-valued classification with an arbitrary number of categories. In this paper, we propose a new configuration method combining binary classifiers with categories, which are not used for classification. This method allows us to reduce the amount of training data for each binary classifier while improving the balance of the training data between two category sets for each binary classifier. As a result, the computational complexity can be decreased. We verify the effectiveness of our proposed method by conducting a document classification experiment.

Hierarchical classification method based on selective learning of slacked hierarchy for activity recognition systems

Khowaja, S.A.,Yahya, B.N.,Lee, S.L. Pergamon ; Elsevier Science Ltd 2017 expert systems with applications Vol.88 No.-

Physical activity recognition using wearable sensors has gained significant interest from researchers working in the field of ambient intelligence and human behavior analysis. The problem of multi-class classification is an important issue in the applications which naturally has more than two classes. A well-known strategy to convert a multi-class classification problem into binary sub-problems is the error-correcting output coding (ECOC) method. Since existing methods use a single classifier with ECOC without considering the dependency among multiple classifiers, it often fails to generalize the performance and parameters in a real-life application, where different numbers of devices, sensors and sampling rates are used. To address this problem, we propose a unique hierarchical classification model based on the combination of two base binary classifiers using selective learning of slacked hierarchy and integrating the training of binary classifiers into a unified objective function. Our method maps the multi-class classification problem to multi-level classification. A multi-tier voting scheme has been introduced to provide a final classification label at each level of the solicited model. The proposed method is evaluated on two publicly available datasets and compared with independent base classifiers. Furthermore, it has also been tested on real-life sensor readings for 3 different subjects to recognize four activities i.e. Walking, Standing, Jogging and Sitting. The presented method uses same hierarchical levels and parameters to achieve better performance on all three datasets having different number of devices, sensors and sampling rates. The average accuracies on publicly available dataset and real-life sensor readings were recorded to be 95% and 85%, respectively. The experimental results validate the effectiveness and generality of the proposed method in terms of performance and parameters.

Comparison of Various Criteria for Designing ECOC

석경하,이승철,전갑동 한국데이터정보과학회 2006 한국데이터정보과학회지 Vol.17 No.2

Error Correcting Output Coding(ECOC) is used to solve multi-class problem. It is known that it improves the classification accuracy. In this paper, we compared various criteria to design code matrix while encoding. In addition. we prorpose an ensemble which uses the ability of each classifier while decoding. We investigate the justification of the proposed method through real data and synthetic data.

Data-Adaptive ECOC for Multicategory Classification

Seok, Kyung-Ha 한국데이터정보과학회 2008 한국데이터정보과학회지 Vol.19 No.1

Error Correcting Output Codes (ECOC) can improve generalization performance when applied to multicategory classification problem. In this study we propose a new criterion to select hyperparameters included in ECOC scheme. Instead of margins of a data we propose to use the probability of misclassification error since it makes the criterion simple. Using this we obtain an upper bound of leave-one-out error of OVA(one vs all) method. Our experiments from real and synthetic data indicate that the bound leads to good estimates of parameters.

Data-Adaptive ECOC for Multicategory Classification

석경하 한국데이터정보과학회 2008 한국데이터정보과학회지 Vol.19 No.1

Error Correcting Output Codes(ECOC) can improve generalization performance when applied to multicategory classification problem. In this study we propose a new criterion to select hyperparameters included in ECOC scheme. Instead of margins of a data we propose to use the probability of misclassification error since it makes the criterion simple. Using this we obtain an upper bound of leave-one-out error of OVA(one vs all) method. Our experiments from real and synthetic data indicate that the bound leads to good estimates of parameters.

Hyperparameter Selection for APC-ECOC

Seok, Kyung-Ha 한국데이터정보과학회 2008 한국데이터정보과학회지 Vol.19 No.4

The main object of this paper is to develop a leave-one-out(LOO) bound of all pairwise comparison error correcting output codes (APC-ECOC). To avoid using classifiers whose corresponding target values are 0 in APC-ECOC and requiring pilot estimates we developed a bound based on mean misclassification probability(MMP). It can be used to tune kernel hyperparameters. Our empirical experiment using kernel mean squared estimate(KMSE) as the binary classifier indicates that the bound leads to good estimates of kernel hyperparameters.

Hyperparameter Selection for APC-ECOC

석경하 한국데이터정보과학회 2008 한국데이터정보과학회지 Vol.19 No.4

The main object of this paper is to develop a leave-one-out(LOO) bound of all pairwise comparison error correcting output codes (APC-ECOC). To avoid using classifiers whose corresponding target values are 0 in APC-ECOC and requiring pilot estimates we developed a bound based on mean misclassification probability(MMP). It can be used to tune kernel hyperparameters. Our empirical experiment using kernel mean squared estimate(KMSE) as the binary classifier indicates that the bound leads to good estimates of kernel hyperparameters.

Comparison of Various Criteria for Designing ECOC

Seok, Kyeong-Ha,Lee, Seung-Chul,Jeon, Gab-Dong 한국데이터정보과학회 2006 한국데이터정보과학회지 Vol.17 No.2

Error Correcting Output Coding(ECOC) is used to solve multi-class problem. It is known that it improves the classification accuracy. In this paper, we compared various criteria to design code matrix while encoding. In addition. we prorpose an ensemble which uses the ability of each classifier while decoding. We investigate the justification of the proposed method through real data and synthetic data.

Data-Adaptive ECOC for Multicategory Classification

Kyung Ha Seok 한국데이터정보과학회 2008 한국데이터정보과학회지 Vol.19 No.1