Retno Kusumaningrum

Integrated visual vocabulary in latent Dirichlet allocation–based scene classification for IKONOS image

Abstract Scene classification based on latent Dirichlet allocation (LDA) is a more general modeling method known as a bag of visual words, in which the construction of a visual vocabulary is a crucial quantization process to ensure success of the classification. A framework is developed using the following new aspects: Gaussian mixture clustering for the quantization process, the use of an integrated visual vocabulary (IVV), which is built as the union of all centroids obtained from the separate quantization process of each class, and the usage of some features, including edge orientation histogram, CIELab color moments, and gray-level co-occurrence matrix (GLCM). The experiments are conducted on IKONOS images with six semantic classes (tree, grassland, residential, commercial/industrial, road, and water). The results show that the use of an IVV increases the overall accuracy (OA) by 11 to 12% and 6% when it is implemented on the selected and all features, respectively. The selected features of CIELab color moments and GLCM provide a better OA than the implementation over CIELab color moment or GLCM as individuals. The latter increases the OA by only ∼ 2 to 3%. Moreover, the results show that the OA of LDA outperforms the OA of C4.5 and naive Bayes tree by ∼ 20 % .

Classification of Indonesian news articles based on Latent Dirichlet Allocation

A massive number of news articles leads to the potential problem in automatic classification task. The discussions on classification of English news articles have been widely studied. However, it is in contrast to automatic classification of Indonesian news articles. The classification method that has been implemented is limited to conventional methods, such as Naïve Bayes and Support Vector Machine. Both methods is rigid in classify a document into one topic. Therefore, we implement one of Topic Modeling methods which represent a document as a distribution of topics and a topic is represented by a set of words. The method is Latent Dirichlet Allocation. The experimental study based on 10-fold cross validation strategy is conducted by employing several parameter includes number of topics (5, 10, and 15) and both LDAs hyperparameters (0.001, 0.01, and 0.1). The result shows that the best overall accuracy is about 70% for classifying documents of Indonesian news articles into 5 classes, i.e. economic, tourism, criminal, sport, and politics.

Parallel rules based classifier using DNA strand displacement for multiple molecular markers detection

The detection of molecular markers such as micro ribonucleic acid (miRNA) expression levels in the cells are essential for diagnosis of a disease, especially cancer. Recently, a huge amount of molecular markers on cell is being extracted by single molecular detection method, as results, excessively detection process and time is required. Meanwhile DNA computing has capabilities to interact with biological nucleic acids and massively parallel processing, we propose parallel rule-based classifier using DNA strand displacement reaction as method of detecting multiple molecular markers. Two DNA reaction of parallel sensing and encoding system of the IF-THEN rules have been developed in our proposed model. In this paper, we compared between the proposed method and the AND gate method for the classification result. Moreover, based on the simulation results show that our method is possible as an alternative solution for the programmable fast big molecular markers detection and diagnosis.