A high-dimensional feature clustering approach to support knowledge-assisted visualization

Abstract

a b s t r a c tThe ever-growing arsenal of methods and parameters available for data visualization can be daunting to the casual user and even to domain experts. Furthermore, comprehensive expertise is often not available in a centralized venue, but distributed over sub-communities. As a means to overcome this inherent problem, efforts have begun to store visualization expertise directly with the visualization method and possibly the dataset, to then be utilized for user guidance in the data visualization, suggesting to the user both the visualization method and its best parameters for the data and task at hand. While this is certainly an immensely useful and promising development, one requirement remains -the matching of a newly acquired dataset with the appropriate segment of the library storing the expert knowledge. This requires one to detect and recognize the dataset's category at some level of granularity and then use this information as a library index. We describe a possible framework for accomplishing the first stage of this process, namely the data categorization, using data classification via a rich set of feature vectors sufficiently sensitive to detect critical variations. We demonstrate the utility of our framework by ways of a set of medical and computational datasets and visualize the resulting categorization as a layout in 2D.

Interactive visual exploration of overlapping similar structures for three-dimensional microscope images

“…In Figure 7(c), because similar structures in the data are highlighted, it is difficult to present the endpoints of one linear structure in a visually distinguishable form. Applying machine learning or clustering approaches [28] are interesting methods to automate TF configuration, but they require prior knowledge to learn the target structures.…”

“…Because the high degree of freedom in multidimensional TFs makes it difficult for users to obtain visualization results through manual parameter settings, a variety of user interfaces and automatic TF generation methods have been investigated [25-27]. In the case of visual exploration, there are many situations for which feature descriptors have not been formulated [28]. Some researchers have focused on this issue and have investigated methods for exploring high-dimensional feature space.…”

Computer Aided Classification of Neuroblastoma Histological Images Using Scale Invariant Feature Transform with Feature Encoding

“…In this stage, a SIFT descriptor is calculated for each keypoint as a histogram of orientation in eight directions (as shown in Figure 8 ). According to [ 15 , 18 , 19 ], it is common that a 16 × 16 neighborhood region around each keypoint is divided into 4 × 4 sub-regions. The gradient vectors are aggregated into 8-bin histogram over the matrix of subregions.…”

Automatic fuzzy comprehensive approach toward the determination of processing methods for the components based on manufacturing features

“…In general, typical parts are divided into shaft parts, plate parts, case parts, abnormity parts, and so on. Every typical component has its own manufacturing features, such as hole feature, tank feature, and flat feature [10] . To some extent, the geometric characteristics of manufacturing features describe a certain engineering significance, which is the main feature of product information model and can be used as the carrier of non-geometric features information for manufacturing features, such as precision characteristic, material characteristic, assembly characteristic, management characteristic.…”