Talayeh Razzaghi

A weighted support vector machine method for control chart pattern recognition

Manual inspection and evaluation of quality control data is a tedious task that requires the undistracted attention of specialized personnel. On the other hand, automated monitoring of a production process is necessary, not only for real time product quality assessment, but also for potential machinery malfunction diagnosis. For this reason, control chart pattern recognition (CCPR) methods have received a lot of attention over the last two decades. Current state-of-the-art control monitoring methodology includes K charts which are based on support vector machines (SVM). Although K charts have some profound benefits, their performance deteriorate when the learning examples for the normal class greatly outnumbers the ones for the abnormal class. Such problems are termed imbalanced and represent the vast majority of the real life control pattern classification problems. Original SVM demonstrate poor performance when applied directly to these problems. In this paper, we propose the use of weighted support vector machines (WSVM) for automated process monitoring and early fault diagnosis. We show the benefits of WSVM over traditional SVM, compare them under various fault scenarios. We evaluate the proposed algorithm in binary and multi-class environments for the most popular abnormal quality control patterns as well as a real application from wafer manufacturing industry.

Multilevel Weighted Support Vector Machine for Classification on Healthcare Data with Missing Values

This work is motivated by the needs of predictive analytics on healthcare data as represented by Electronic Medical Records. Such data is invariably problematic: noisy, with missing entries, with imbalance in classes of interests, leading to serious bias in predictive modeling. Since standard data mining methods often produce poor performance measures, we argue for development of specialized techniques of data-preprocessing and classification. In this paper, we propose a new method to simultaneously classify large datasets and reduce the effects of missing values. It is based on a multilevel framework of the cost-sensitive SVM and the expected maximization imputation method for missing values, which relies on iterated regression analyses. We compare classification results of multilevel SVM-based algorithms on public benchmark datasets with imbalanced classes and missing values as well as real data in health applications, and show that our multilevel SVM-based method produces fast, and more accurate and robust classification results.

Weighted relaxed support vector machines

Classification of imbalanced data is challenging when outliers exist. In this paper, we propose a supervised learning method to simultaneously classify imbalanced data and reduce the influence of outliers. The proposed method is a cost-sensitive extension of the relaxed support vector machines (RSVM), where the restricted penalty free-slack is split independently between the two classes in proportion to the number samples in each class with different weights, hence given the name weighted relaxed support vector machines (WRSVM). We compare classification results of WRSVM with SVM, WSVM and RSVM on public benchmark datasets with imbalanced classes and outlier noise, and show that WRSVM produces more accurate and robust classification results.

Scalable Multilevel Support Vector Machines

Abstract Solving optimization models (including parameters fitting) for support vector machines on large-scale training data is often an expensive computational task. This paper proposes a multilevel algorithmic framework that scales efficiently to very large data sets. Instead of solving the whole training set in one optimization process, the support vectors are obtained and gradually refined at multiple levels of coarseness of the data. Our multilevel framework substantially improves the computational time without loosing the quality of classifiers. The algorithms are demonstrated for both regular and weighted support vector machines for balanced and imbalanced classification problems. Quality improvement on several imbalanced data sets has been observed.

Constraint relaxation, cost-sensitive learning and bagging for imbalanced classification problems with outliers

Supervised learning consists in developing models able to distinguish data that belong to different categories (classes). When data are available in different proportions the problem becomes imbalanced and the performance of standard classification methods deteriorates significantly. Imbalanced classification becomes even more challenging in the presence of outliers. In this paper, we study several algorithmic modifications of support vector machines classifier for tackling imbalanced problems with outliers. We provide computational evidence that the combined use of cost sensitive learning with constraint relaxation performs better, on average, compared to algorithmic tweaks that involve bagging, a popular approach for dealing with imbalanced problems or outliers separately. The proposed technique is embedded and requires the solution of a single convex optimization problem with no outlier detection preprocessing.