With increasing data volumes, the bottleneck in obtaining data for training a given learning task is the cost of manually labeling instances within the data. To alleviate this issue, various reduced label settings hav...
详细信息
With increasing data volumes, the bottleneck in obtaining data for training a given learning task is the cost of manually labeling instances within the data. To alleviate this issue, various reduced label settings have been considered including semi-supervised learning, partial- or incomplete-labellearning, multiple-instancelearning, and active learning. Here, we focus on multiple-instance multiple-label learning with missing bag labels. Little research has been done for this challenging yet potentially powerful variant of incomplete supervision learning. We introduce a novel discriminative probabilistic model for missing labels in multiple-instance multiple-label learning. To address inference challenges, we introduce an efficient implementation of the EM algorithm for the model. Additionally, we consider an alternative inference approach that relies on maximizing the label-wise marginal likelihood of the proposed model instead of the joint likelihood. Numerical experiments on benchmark datasets illustrate the robustness of the proposed approach. In particular, comparison to state-of-the-art methods shows that our approach introduces a significantly smaller decrease in performance when the proportion of missing labels is increased.
Frog call classification has received increasing attention due to its importance for ecosystem. Traditionally, the classification of frog calls is solved by means of the single-instance single-label classification cla...
详细信息
ISBN:
(纸本)9783319336183;9783319336176
Frog call classification has received increasing attention due to its importance for ecosystem. Traditionally, the classification of frog calls is solved by means of the single-instance single-label classification classifier. However, since different frog species tend to call simultaneously, classifying frog calls becomes a multiple-instance multiple-label learning problem. In this paper, we propose a novel method for the classification of frog species using multiple-instancemultiple-label (MIML) classifiers. To be specific, continuous recordings are first segmented into audio clips (10 s). For each audio clip, acoustic event detection is used to segment frog syllables. Then, three feature sets are extracted from each syllable: mask descriptor, profile statistics, and the combination of mask descriptor and profile statistics. Next, a bag generator is applied to those extracted features. Finally, three MIML classifiers, MIML-SVM, MIML-RBF, and MIML-kNN, are employed for tagging each audio clip with different frog species. Experimental results show that our proposed method can achieve high accuracy (81.8% true positive/negatives) for frog call classification.
Automated annotation of skin biopsy histopathological images provides valuable information and supports for diagnosis, especially for the discrimination between malignant and benign lesions. Currently, computer-aid an...
详细信息
Automated annotation of skin biopsy histopathological images provides valuable information and supports for diagnosis, especially for the discrimination between malignant and benign lesions. Currently, computer-aid analysis of skin biopsy images mostly relied on some human-designed features, which requires expensive human efforts and experiences in problem domains. In this study, we propose an annotation framework for automated skin biopsy image analysis which makes use of a deep model for image feature representation. A convolutional neural network (CNN) is designed for local regions of skin biopsy images which learns potential high-level features automatically from input raw pixels. The annotation model is constructed in the multiple-instancemultiple-label (MIML) learning framework with the features learned through the network. We achieve significant improvement of the model performance on a real world clinical skin biopsy image dataset and a benchmark dataset. Moreover, our study indicates that deep learning based model could achieve better performance than human designed features.
暂无评论