SPIE '95 Conference on Visual Data Exploration and Analysis II
               San Jose Convention Center, San Jose, CA
                         February 6-10, 1995

             Robert F. Erbacher and Georges G. Grinstein
	     Institute for Visualization and Perception Research
	     University of Massachusetts Lowell
	     Lowell MA 01854

A wide range of interface, display, and computation technologies are now available for users to visualize, explore or interpret data. Commercial interactive visualization tools are available which produce 2D images and animations, 2.5 D presentations on stereoscopic displays and 3D presentations in virtual environments. Newer experimental tools and environments have been developed in research laboratories and experimentation with these is important to assist other members of the community in determining what techniques and software best satisfy a given need. At the rate at which the technology in this field is progressing it has thus become increasingly important to present and disseminate this research. This justifies having a yearly visual data analysis and exploration conference which presents results in a timely manner.

The conference this year covered topics ranging from novel interdisciplinary approaches to visualization systems to specific application solutions. The focus of the conference consisted of 7 main areas that included: Perception and Interaction, Algorithms, Field Visualization Techniques, High Performance Computing, Modeling, Image Reconstruction, and Applications.

There were several issues that were identified as being especially interesting and important to this field and presented in the papers. First noted was the differentiation between exploratory and explanatory visualization which becomes important when defining the applications for which a technique is applicable. Second, the use of supercomputers and other high performance systems in the generation of visual displays, the performance of such systems, and the visualization of the systems themselves - typically for performance analysis - is becoming increasingly important as the amount of data which needs to be visualized increases. The goals of high performance computing work in visualization include: being able to execute computationally complex applications and have them generate displays in real-time; and providing tools to aid in analyzing the performance of such applications. Modeling and algorithm papers provided key examples of the support required for visualization applications and also showed how some visualizations are migrating towards highly interactive synthetic and virtual environments for data exploration. Lastly, issues in perception and interaction were presented, in particular, the use of sonification as a medium for data analysis and the creation of models for the description of sound spaces. This goes well in hand with the development of perceptually-based presentations, a relatively new key research area in visualization.

Exploratory Visualization Vs. Explanatory Visualization

The differentiation made by David Southard, in "Vector quantization: a tool for exploration and analysis of multivariate images", between exploratory visualization and explanatory visualization is valuable for specifying tools requirements. Whereas explanatory visualization deals with presenting information directly in order to show its characteristics, exploratory visualization attempts to derive information not directly available through current techniques. Exploratory visualization aids in developing the techniques that will later be used to present displays in an explanatory visualization environment.

High Performance Computing

The papers in this session provided system and application examples. Two of the presented papers analyzed the performance of high performance systems. Andreas Wierse in his paper entitled "The COVISE visualization system and its performance under different conditions" evaluated the performance of a specific software application. Robert Erbacher on the other hand described the performance limitations of specific existing hardware in his paper "Performance issues in a real-time true color data display". Steven Hackstadt discussed the DEVise system in his paper "Case study: applying scientific visualization to parallel performance visualization" and provided an example of a system using visualization tools to analyze the performance characteristics of parallel programs.

The use of supercomputers is now seen as a viable tool for aiding in the generation of visualization displays. Philip Chen provided an example of the use of supercomputers to generate climate simulations and perform data visualization.

Sonification

The use of sound as an additional medium for presenting data is becoming increasingly widespread. The work at the University of Massachusetts-Lowell's Institute for Visualization and Perception Research has shown that sound can be very effective as a medium for presenting additional information in addition to the data being displayed visually. Sound can also be used as a probe to determine if regions which appear nearly identical visually have different characteristics. These differences could be too minor to identify visually or be due to parameters not being visually displayed. Sound can also be used to support visually impaired users.

Matti Grohn and Tapio Takala discussed sonification of spatially located data sets. Their paper entitled "MagicMikes - Method for Spatial Sonification" extends previous sonification work with tools to aid in the identification of the spatial location of data attributes.

Steven Barass and Phil Robertson in their paper "Data exploration with sound using a perceptually linearized sound space" discussed the construction of a perceptual sound space as an aid in the mapping of data parameters to sound attributes. Their work is an attempt to formalize the theory but there are still many significant issues needing to be addressed.

Conclusion

This was an exciting conference that clearly spanned a wide interest. The range of ideas being developed and the variety of fields to which it is being applied make visual data exploration and analysis an extremely active and dynamic field. Very few areas in computer science are as closely tied to the needs of other fields. This requires that researchers in the field keep up to date with the latest techniques not only in their primary field but also many related fields. In order to fulfill the need for continued dissemination of knowledge in this area in a combined forum the Visual Data Exploration and Analysis conference will make its appearance again next year as part of the IS&T/SPIE symposium.