Abstract:In various classification areas, the curse of dimensionality becomes a major challenge among the researchers. Thus, feature selection plays an important role in overcoming dimensionality problem. Relief-f is one of the filter methods to rank the most significant features based on their relevance. Although relief-f proved to be a powerful technique in filter strategy, but this method only rank the features based on their significant level. Hence, feature selection is embedded to select the most meaningful features based on their rank. Differential evolution (DE) is one of the evolutionary algorithms that are widely used in various classification domains. Simple and powerful in implementation, we combined relief-f with DE in our proposed feature selection method to solving the optimization problem. In this work, population size and generation size were adaptively determined from the number of features from relief-f. The performance of proposed method is compared with several feature selection techniques in order to prove their superiority using ten datasets obtained from UCI machine learning repository.

Monitoring and measuring the shoreline of coastal zones helps establish the boundary of a country. Such an activity entails ground survey, topographic survey, aerial photo, or remote sensing techniques to extract the shoreline. For example, the remote sensing technique to determine shorelines involves the extraction of relevant data from satellite images. Specifically, the satellite image classification enables shorelines to be extracted from land and water classes with a high degree of precision. However, extracting information from satellite images is challenging as it relies on a strong understanding of image processing, machine learning, and data mining techniques. Thus, the researchers discuss the study of the pixel-based classification of machine learning techniques to classify land and water classes in terms of accuracy, training time, and testing time. The research findings showed that the Multilayer Perceptron Artificial Neural Network (MLP ANN) was the most effective technique, compared with other techniques, hence reinforcing its importance in classifying land and water classes.

Observation satellites orbiting the Earth provide important surveillance information that helps in identifying various types of land cover. As such, the resolution of satellite images is critical to ensure high accuracy in classifying land cover types. Clearly, high-resolution images are desirable, but such images are prohibitively expensive. Hence, the use of medium-resolution satellite images seems more economical and practical. Several techniques have been developed to monitor the conditions of land covers across the world, such as aerial photography, ground survey, and remote sensing. Among the three techniques, remote sensing is the best, given its ability to monitor vast geographical areas more accurately and much faster compared to those of the other two techniques. In recent decades, many countries have been relying on remote sensing to monitor the conditions of coastal areas by extracting shorelines from satellite images. To date, several pixel-based methods have been proposed for the extraction of shorelines, but most of these methods are fraught with problems. Thus, the object-based approach is proposed using a combination of segmentation algorithms, namely Felzenswalb, Quickshift, and SLIC, together with 15 machine learning classifiers, to classify segmented images of Langkawi Island. The performance of the segmentation algorithms and machine learning classifiers were assessed in terms of segmentation time and overall accuracy in four experimental settings comprising of three different parameters. The research findings showed that the proposed hybridization of SLIC segmentation and Extra Tree classifier was the most efficient and accurate technique compared to other combinations of techniques in extracting the shoreline of the study area. Specifically, SLIC was faster than Felzenswalb and Quickshift by as much as 37 times and 500 times, respectively. Together with Extra Tree classifier, SLIC managed to achieve 100% overall accuracy in the object-based classification in three out of four settings compared to the rest of the techniques tested in the study.

Shoreline is a very important element to identify exact boundary at the coastal areas of a country. However, in order to identify land-water boundary for a large region using traditional ground survey technique is very time consuming. Alternatively, shoreline can be extracted by using satellite images that minimizes the mapping errors. The trend of extracting shoreline has been shifted from image processing to machine learning and data mining techniques. By using machine learning technique, the satellite images could be classified into land and water classes in order to extract shoreline. However, the result is meaningless if it has cloud and shadow on the water-land boundary. In this study, we compare the accuracy and Kappa Coefficient of six machine learning techniques namely Maximum Likelihood, Minimum Distance, Mahalanobis Distance, Parallelepiped, Neural Network and Support Vector Machines on three type of images; single optical multispectral, single SAR and fused image. A case study for this research is done alongside Tumpat beach, located at the Northeast Coast of Peninsular Malaysia. All the machine learning techniques have been tested on the three types of images. The experimental results show that classification using SVM on single multispectral image has the highest accuracy among all. However, the classified of fused image using SVM is considered much more accurate because it can cater the cloud and shadow problem. Additionally, the classification on 5 and 10 m fused images also tested and the result shows that with the increase of spatial resolution of fused image, the classification accuracy also increases.

No abstract