NEURAL NETWORK-ENHANCED STEREOSCOPIC VISUALIZATION

点击翻译按钮获取中文摘要

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Large and high-dimensional data sets mapped to low-dimensional visualizations often result in perceptual ambiguities. One such ambiguity is overlap or occlusion that occurs when the number of records exceeds the number of unique locations in visualization or when there exist two or more records that map to the same location. To lessen the affect of occlusion, nonstandard visual attributes (i.e. shading and/or transparency) are applied, or such records may be remapped to a corresponding randomly-generated jittered location. While the resulting mapping efficiently portrays the density of records it also fails to provide the insight into the relationship between the neighboring records. Our algorithms address such issues through the integration of neural-network and visualization techniques, displacing records based on their dimensional values, instead of the more common random displacement of records. Unfortunately, large data sets and detailed visualizations make these algorithms computationally intense. To address this, we extend the algorithms to provide support for computational environments consisting of a large number of clustered computers. We measure the effectiveness of new visualizations using empirical evaluation targeted at task analysis, usability evaluation and usage analysis while the effectiveness of algorithms is measured in time required to process the data.