Abstract Semi-Supervised Learning of Gaussian Classifiers
In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled... more In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled data. Training with unlabeled data introduces efficiency in terms of time and energy spent for labeling the data. We present experiments on different data sets to illustrate the effect of unlabeled data on the performance of the classifiers. We will try to show that under specific conditions unlabeled data contains valuable information for the target function. The algorithm we utilize, first trains the classifier with limited number of unlabeled data and then with the obtained classifier labels unlabeled data and re-trains the classifier with newly labeled data and proceeds the iteration. 1.
In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled... more In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled data. Training with unlabeled data introduces efficiency in terms of time and energy spent for labeling the data. We present experiments on different data sets to illustrate the effect of unlabeled data on the performance of the classifiers. We will try to show that under specific conditions unlabeled data contains valuable information for the target function. The algorithm we utilize, first trains the classifier with limited number of unlabeled data and then with the obtained classifier labels unlabeled data and re-trains the classifier with newly labeled data and proceeds the iteration.
Abstract Semi-Supervised Learning of Gaussian Classifiers
In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled... more In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled data. Training with unlabeled data introduces efficiency in terms of time and energy spent for labeling the data. We present experiments on different data sets to illustrate the effect of unlabeled data on the performance of the classifiers. We will try to show that under specific conditions unlabeled data contains valuable information for the target function. The algorithm we utilize, first trains the classifier with limited number of unlabeled data and then with the obtained classifier labels unlabeled data and re-trains the classifier with newly labeled data and proceeds the iteration. 1.
In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled... more In this paper we present an approach that trains Gaussian classifiers using labeled and unlabeled data. Training with unlabeled data introduces efficiency in terms of time and energy spent for labeling the data. We present experiments on different data sets to illustrate the effect of unlabeled data on the performance of the classifiers. We will try to show that under specific conditions unlabeled data contains valuable information for the target function. The algorithm we utilize, first trains the classifier with limited number of unlabeled data and then with the obtained classifier labels unlabeled data and re-trains the classifier with newly labeled data and proceeds the iteration.
Uploads
Papers by Faysal Başçi