Cheryl A. Geisler, Professor and Chair of Language, Literature and Communications

John E. Kolb, Dean of Computing and Information Services (Project Coordinator)

Brian Lonsway, J. Erik Jonsson Distinguished Visiting Assistant Professor, Architecture

Don L. Millard, Director of the Center for Integrated Electronics and Electronic Manufacturing

Mark S. Shephard, Professor Samuel A. Johnson '37 and Elizabeth C. Johnson Professor in Engineering and Director of the Scientific Computation Research Center (SCOREC)

William A. Wallace, Professor of Decision Sciences and Engineering Systems

Jack M. Wilson, Professor of Physics and Engineering Science and Dean of Undergraduate and Continuing Education

The Learning Continuum

• Uses of Intel Architecture Systems
• Numerically Intensive Computing
• Graphically Intensive Computing
• Communications Intensive Computing

Distributed Collaborative Learning Environments

• Studio Classes
• Troy and Pittsburgh Buildings: Studio Classrooms
• Pre-College Initiative
• Engineering Studio Classrooms
• Human Computer Interaction
• Electronic Arts

Simulation, Modeling, Visualization, and Parallel Processing

• Computer Graphics in the Engineering Curriculum
• High Performance Computing via Parallel Processing
• Numerically Intensive Computing (NIC) Cluster
• Modeling and Simulation of Interconnection Design
• Developing New Environments for Data Mining
• Technology of Architecture Advancement
• The Academy of Electronic Media and Interactive Electronic Media Development

Equipment, Budget, and Timing Summary

Conclusion

CIEEM was created in December 1994 through a merger of the Design and Manufacturing Institute, whose mission was to carry out industry-oriented research in design and manufacturing, and the Center for Integrated Electronics, known for its leadership in on-chip interconnect research. CIEEM now emphasizes all aspects of integrated electronics and electronics manufacturing from the device, to the chip, to the board, to the system levels. The Center's current activities range from basic/applied research and education to commercialization through partnerships with industry. CIEEM enhances Rensselaer's strong tradition of contribution to education and industry. Our purposes are many: to address industry's needs by establishing a closer collaboration with academia through specific semiconductor technologies and electronic manufacturing research initiatives; to develop world-class education programs; and to exchange insight and technology gained through research into applications in industry and government. CIEEM's activities range from basic and applied research and education in interconnections and interfaces to commercialization through partnerships with industry. A complement of about 50 faculty, 100 students, and 15 full time research staff conduct research activities incorporating projects for specific companies, as well as longer range programmatic efforts in fundamental areas of materials processes, design, fabrication and characterization related to electronics design and manufacturing. Research, development and technology deployment areas of CIEEM include, but are not limited to: semiconductor devices; advanced electronic materials and superconductors; electronics design, modeling and simulation; multilevel interconnect technologies; agile and environmentally conscious electronics manufacturing; IC fabrication and processing; electronics packaging; optoelectronic materials and optoelectronics; power electronics; and electronic media development and utilization. State-of-the-art facilities enhance research opportunities and include an 8500ft2 Class 100 microfabrication clean room with processing capabilities both for Si and III-V base devices/circuits and numerous state-of-the-art processing design, characterization and testing laboratories.

Graphically Intensive Computing

Rensselaer has a long-standing history with graphically intensive computing. In the late 1970's the Center for Interactive Computer Graphics (CICG), was established with then state of the art equipment and support from vendors such as IBM, PRIME, IMLAC, and Digital Equipment Corporation. Today, graphically intensive computing is not only a research interest, but also an integral part of everyday computing. For example, in partnership with IBM, Rensselaer is deploying the UNIX-based VideoCharger to serve video streams to Windows Intel desktops.

The Electrical, Computer and Systems Engineering (ECSE) department at Rensselaer has begun development of a model which involves the use of Web-based, interactive multimedia materials in conjunction with a Studio format of course delivery. Interactive Learning Modules (ILMs), developed using Macromedia's Director to produce multimedia materials delivered via Shockwave, are used in Studio classes by faculty to help explain concepts and demonstrate applications, and by students for specific learning exercises.

The Visualization Laboratory for Scientific Computing (VLSC) is provided by the SCOREC as a service to the Rensselaer's research community, using the application of 3D computer graphics to facilitate the solution of scientific and engineering problems. The visualization packages, which run on SGI UNIX workstations and servers, include Data Explorer Version-2 from IBM, IRIS Explorer from SGI, Visualization Toolkit from GE and SciAn from Florida State University.

Graphically intensive computing is the cornerstone of two new degrees at Rensselaer. The first is the Electronic Media Arts and Communication (EMAC) undergraduate major, the other the MFA of Electronic Arts. These are based on the new technology of computers and involve such things as multimedia, the Internet (in particular the web), and art forms. Graduates of these two programs can look forward to such careers as software and game design, Web design, and video and computer music composition. Classes for these two degrees are graphically intense, and require state-of-the-art technology. These degrees are pushing the boundaries of such technology and in some cases even creating new technologies. The School of Architecture is drastically revamping its curriculum to embed graphics-intensive applications (e.g., Softimage3D for behavior- and rule-based modeling and animation as well as for photo-accurate rendering).

Students at Rensselaer require graphically intensive computing, not only for their classes and class applications but also for their free time. Integral computing components of the undergraduate-computing environment include Pro/ENGINEER, Maple, and Matlab, for design, analysis and visualization. Web page design, required for some courses, is typically graphically intensive. The Rensselaer culture encourages students to push boundaries and explore all technology available for their use.

Over the last decade, Rensselaer has built a first rate network on campus and established high-speed connections to the Internet. FDDI rings provide high-speed interconnections between buildings and major centers, with a smattering of ATM deployed in select locations. Virtually all buildings, including dormitories are cabled with 10Mbps Ethernet, with 100Mbps Ethernet in some facilities. Rensselaer has a T3 connection to the Internet, and a sizable pool of high-speed modems and terminal servers for dial-up users.

Significant bandwidth is required to support infrastructure services such as Usenet news, over 150GB of files in the campus-wide AFS filesystem, accessible to virtually all UNIX and WinXX desktops, and the campus-wide tape backup system built upon IBM's ADSM. WWW servers function as applications and information servers for an increasing number of campus-wide functions, including the Library's online public access catalog, and the new Oracle-based Student Records System.

Distance learning initiatives such as RSVP (Rensselaer Satellite Video Program), the distance education function of Rensselaer's Office of Continuing Education and Distance Education, which initially provided instructional materials via satellite and video tapes, are now employing collaborative conferencing software such as Interactive Learning International Corporation's (ILinc) LearnLinc.

Rensselaer, building upon its successes with the creation of new learning environments, proposes to produce a system of interactive classrooms that addresses the continuum of the education process from secondary school to graduate and continuing education. These classrooms will use the collaborative technologies developed at Rensselaer (such as LearnLinc and the Collaborative Classroom work) to provide for the "Continuous Education" needs of a variety of audiences including:

• high school minority students aspiring to further study in technology, science, and engineering
• undergraduate students working in interactive learning environments both on campus and off
• graduate students in on campus programs
• graduate students who are in full time employment
• continuing education students at any stage of their career in industry, who need to upgrade their skills in specific areas. We will particularly focus on the skills required to integrate new computing and communication technologies across the Institute.

This Rensselaer continuum of "Continuous Education" pervades each of the sections below. We are asking Intel to partner with us in the creation of these new environments by providing the equipment required to create demonstration learning environments. We will take advantage of the advances in computing and communications pioneered by Intel and the advances in collaborative distributed classroom environments for which Rensselaer has won international recognition. Rensselaer is also prepared to match the Intel grant of equipment and technology with our own resources to acquire software tools, create software and instructional materials, and provide high quality facilities. This support is quite substantial as shown in the budget.

We are proposing to create:

1. Two distributed collaborative classrooms in the Troy building. Each classroom would be designed in the Studio Classroom format and would be equipped with 36 Intel computer systems. Each computer would be networked and would include the new Intel ProShare H.323 desktop video-conferencing system. These computers would be networked to WindowsNT servers. An external ISDN connection will allow these systems to be used in collaborative conferences with the other sites (high schools, universities, and corporations). The will use software from Intel business partner and Rensselaer affiliated company, ILinc, to enable the full conferencing capability of ProShare in the H.323 and H.320 environment.
2. A distributed collaborative design classroom in the Troy building that would be equipped as above with 20 computers and optimized for teaching topics in design, electronic media, and human computer interface issues and would also serve as a platform for research in the collaborative design process.
3. A management of technology classroom in the Pittsburgh building. This classroom would be used by the Lally school of Management and Technology in their on campus and off campus (distributed) programs. This would include support for the Executive MBA program, our cooperative programs with Chinese universities, and our courses delivered to 40 different multi-national corporations. These classrooms would also include networked Intel computers and the use of the ProShare desktop video conferencing system to facilitate interactions with other sites.
4. An Electronics and Instrumentation Studio classroom for Engineering. We request a server from Intel (Hewlett-Packard has agreed to provide the client IA-based computing and the electronic instrumentation for each student work area).
5. An Engineering Studio to support the classes in computing organization logic and design. For this class, we request 27 Intel computers. This class will move from a UNIX workstation focus to a PC focus.
6. Engineering CAD Studio classroom. Over 600 students each semester are introduced to Engineering Graphics and CAD. In the past, we have focused upon ProEngineer on UNIX workstations. We will move this course and other uses of the CAD program to WindowsNT-based Pentium II systems.
7. A Human Computer Interaction Studio will support both our rapidly growing corporate distance learning program and our world-renowned research in this area. Since the institution of the HCI certificate program for continuing education (at the graduate level) two years ago, hundreds of employees at cooperating corporations have participated.
8. The Electronic Media Arts and Communication (EMAC) facility will also be a dual use (research/education) facility and will support our new and rapidly growing program at the undergraduate level.

This need for "Continuous Education" comes at a time when universities and corporations are called upon to make significant increases in quality of the educational experience without the prospects for significant additional resources. At Rensselaer this has led to re-engineered courses, training, and curriculum that incorporate technology and often break the constraints of place and time. The old paradigm of on-site versus is distant learners is blurring rapidly as the increasing availability of network resources and collaborative software stimulates a convergence. Technologies such as desktop video conferencing promise the ability to reach anybody, anywhere, at nearly any time, and at much reduced cost. Realizing the promise of these technologies will take creativity and courage, as well as a deep understanding of the way we learn and retain information. Rensselaer has received a variety of awards for the creative solutions that we have developed for these environments.

Rensselaer Polytechnic Institute is in the midst of a full-scale, campus-wide effort to revolutionize undergraduate education and research, while simultaneously meeting significant fiscal challenges. For this effort we received one of the first awards in the National Science Foundation Institution-wide Reform Recognition Program.

A studio classroom operates on the principle that students learn more when they are actively thinking about and doing things related to the subject matter than when they are passively listening to and watching the instructor. This makes the focus of the studio class not the instructor but the student.

The studios are multi-focus rooms for 30-70 students. Students work in teams of two, with each team using an interactive learning station consisting of a networked PC. There are display screens in the front of the room, each capable of overhead, video or computer display for mini-lectures and demonstrations. Special lighting controls are used to direct lighting where the students are working, while providing good visibility for the displays. The PCs are connected with a local network, and a large video server is used to stream video directly to the students' desktops at 10 Mbps. The powerful new processors from Intel and the high performance capabilities of the IA equipment would be of great use in these new studio classrooms. Intel's technology could be present in the beginning of what will become a standard method of educating engineers, scientists, architects, psychologists, technical communicators, philosophers, managers, and economists.

When students face the front of the studio, they are seated at a table where they can take notes, consult reference materials, discuss issues, or perform design exercises. Students turn to face the rear of the facility for hands-on activities, where another table contains the PC, test and measurement equipment, and the prototyping materials. Software tools include discipline-specific applications such as PSpice, plus campus-wide applications such as ProEngineer, Maple, Matlab, Netscape, Word, Excel, etc.

The studio classes typically include 5 or 6 separate activities, so that the students are switching gears every 20-30 minutes to maintain interest and motivation. Activities typically include mini-lectures (no more than 15-20 minutes), pencil and paper exercises, interaction with Web-based materials, and, when applicable, design, simulation, and experimental measurements. With the exception of the paper and pencil exercises the rest of the class time involves computers, including the mini-lectures in which the Professor is likely to use the computer to illustrate and highlight points and information. The overall time spent lecturing to students has been reduced by about one half from the traditional version, with a corresponding increase in hands-on activities.

These classrooms often involve graphically intensive and communications intensive computing (e.g., Web-based materials and graphic electronic demonstrations). The studio formats have commingled electronic and non-electronic (verbal and non-verbal) communications in both on-site and distance formats. Students have the ability to raise their hand electronically, take over or lead discussions, ask questions, or make points all in an electronic forum. Intel's equipment would be of great value in these classrooms, allowing the graphically and communications intensive uses of electronic equipment to run smoothly. Such use would be a great opportunity for both Rensselaer and Intel to develop new uses for IA equipment as well as extend the present capabilities.

The Troy Building will contain three studio style classrooms. One of these is called the Collaborative Classroom; it is a studio "classroom-in-the-round" (see figure of the three classrooms in the Troy Building). Intended to be a showcase for interactive learning, it will be equipped with a computer projection system, several electronic "white" boards that store information written on them directly to the computer, and tables for student work groups of six. This room will be a showcase of Rensselaer's past, present, and future with etched glass panels displaying scenes from Rensselaer's history (the Brooklyn Bridge, the Ferris wheel, etc.). Two additional interactive classrooms capable of seating 60 students flank the classroom-in-the-round. The entire building will be wired for modern and future data communication. Designed with extensive campus input, the state-of-the-art teaching tools include visual aids, strategically placed computer jacks, and overhead projection systems for multimedia presentations. The rooms also will be completely wired for distance learning using the Intel ProShare system.

The Collaborative Classroom accommodates the needs of intellectual teamwork by providing teams of users with the technological support required to generate, coordinate, and refine the joint activity required by multidisciplinary design projects. This classroom would provide across-the-table seating for up to forty-two students at technology-enhanced conference tables. At each table, shared high-end workstations, Team Systems, will serve teams of students via large-screen color monitors buried in the table and shared sets of keyboards and mice. A high-speed network will connect the Team Systems via a gateway to the campus and worldwide network. Ethernet ports at each of the tables will allow students to link their Individual Systems- laptop computers brought to class. At the front of the room, an Instructor System will consist of a high-end workstation and input devices, a slightly smaller color monitor, and a high-end projection system. The addition of video-conferencing capabilities in the Studio Classrooms, will permit interactions between the class and teams of off-site students.

Rensselaer will apply knowledge learned from the collaborative studio style classrooms to distance learning. In the next phase of collaboration education, for which we seek Intel support, we plan to extend these innovations in collaborative computing in two ways. We will extend the WindowsNT-based Collaboration Network, now being developed for use in the Collaborative Classroom, to support collaborations within teams consisting of both on- and off-campus students. Currently, membership in distance teams must be limited to a single site. Building on the development work required to support distant teaming, we will incorporate image reconstruction techniques into the Collaboration Network. Collaborative work often centers around not only shared files and applications, but also the sharing of material artifacts. Through image reconstruction techniques, we plan to support the sharing of these artifacts.

Distance learning programs have often been viewed as ineffective because of the lack of interactivity. The dominant media used today includes the synchronous (live or real time) modes of satellite video and desktop video conferencing, or the asynchronous (on-demand) modes of videotape, print material, CD-ROM, email, or Web courses. Rensselaer presently educates about 900 students per semester in interactive distance learning formats, and they are among the leaders in research on this format. Rensselaer wants to blend distance learning with campus learning, to provide distance learning students the same learner-centered studio experience as on-campus students. Using Intel equipment, we plan to develop a WindowsNT-based satellite conference room, the Interactive Seminar Room, that will demonstrate the potential for students at distant sites to be incorporated into on-campus teams and combine both the synchronous (live) and asynchronous (on-demand) modes.

Personnel support for this development work has already been provided by the National Science Foundation and NYNEX. The department of Language, Literature and Communication will provide the space and facilities renovation required for the Interactive Seminar Room. This project will draw on the high-performance capabilities of Intel-based computers, taking them, and Rensselaer, into new areas of education classroom, and educational, design.

We are extending these models into the Lally School of Management & Technology. The Lally School is executing an ambitious plan to extend their activities through on-campus education, off-campus education, and new research ventures. Our management program is the largest segment of the Rensselaer offerings of distributed education to corporations. This program has been growing over the past twelve years and the Lally School taught the majority of the 900 students enrolled this past spring. There is also a rapidly growing program with several universities in China. Students divide their time between China and Troy. This summer, Rensselaer will completely renovate the Pittsburgh Building for the Lally School of Management and Technology. We will then, in partnership with Intel, create a Distributed Collaborative learning Environment for Management and Technology.

The Pre-College Initiative plans to use the Studio Style Classrooms and new learning methods to extend Rensselaer to the K-12 community. In so doing, Rensselaer will not only continue with its tradition of outreach and community service, but also increase its recruitment of the most talented high school youngsters.

Despite a wave of reform ignited by "A Nation at Risk," too many of our students come to us without the requisite skills and knowledge necessary to meet the high academic standards and demands we make of them. In addition, far fewer high school students are opting for careers in mathematics, science and technology, and we believe that part of this can be attributed to a lack of stimulating and interesting secondary school classroom experiences. We also believe that the hands-on, learning-by-doing methods that we have developed over the past decade can revitalize the secondary school curriculum just as they have revitalized education on our own campus.

Rensselaer's challenge, therefore, is to enhance the educational experience for as many high school youngsters as possible, while at the same time assuring transfer of effective, innovative pedagogical techniques to classroom teachers. We are proposing to do this with technology acquired from Intel and matching support from several other sources for the teacher enhancement effort and the curriculum development work.

Rensselaer is proposing a three-year pilot project whose objectives are to:

• Help families, schools and communities understand how to use technology by educating teachers in classroom use of new technologies and interactive learning, and producing a cadre of teachers who can be change agents and mentors in their respective school districts;
• Ensure maximum curriculum options for high school students, despite shrinking budgets and a shortage of qualified teachers, by delivering college-level educational experiences to students using our new technologies and pedagogies;
• Transfer the knowledge and expertise that Rensselaer has developed in student-centered learning environments in order to create a "virtual studio classroom" at a distance that will enhance learning and serve as a national model both for university/school collaborations, and for future distance learning efforts of all kinds;
• Increase the pool of qualified students for Rensselaer and other universities, while developing alternate sources of income.

During the first year of the pilot we will bring to campus, for a one-year sabbatical, mathematics and science teachers who will spend their time both as "students" and as curriculum developers. Specifically, they will: learn about Rensselaer's new instructional technologies and teaching pedagogies; help develop college level curriculum materials and courses for use in their school districts during the second and third years of the project, which will be consistent with the new national standards and state frameworks; be involved in the development of state-of-the art classrooms at their school districts; and help develop tools to assess and evaluate the educational impact of these pilots. During the first year of the project we will develop two state-of-the-art classrooms at Rensselaer which we would use for 3 main purposes: for on-site training for the visiting teachers; as test areas for delivery of our studio-based courses, and as the points of origin for finished courses delivered to participating school districts. During the second year, the above courses will be piloted at the participating school districts through web-based experiences and long-distance technologies. The teachers who were on sabbatical at Rensselaer during the first year will serve as on-site school instructors overseeing both the delivery of the courses and their revision. During the third year, the revised courses will again be offered as a final pilot at the participating school districts through web-based experiences and long-distance technologies. These courses will undergo their final modification for use on an ongoing basis at the participating school districts. In addition, Rensselaer will begin the process of scaling up the pilot to other school districts, using the final revised courses as the product to be delivered.

During all three years of the pilot, an effort will be mounted to evaluate and assess these activities. We expect that the lessons learned in this pilot will serve as an important assessment of web-based and long-distance learning in general. This pilot will:

• Provide a proof of concept for long distance delivery of courses to high school students;
• Produce a large number of teachers across the country who will be networked with each other and with college educators, who will be capable of using this network to the best advantage for themselves and their students, and who will serve as change agents and lead teachers in their respective school districts
• Result in a collection of web-based materials and resources that teachers can easily access
• Allow ongoing interaction and communication amongst Rensselaer and all participating school districts utilizing the same web-based technologies developed as part of the pilot program
• Develop a set of assessments and/or assessment standards that will be useful for future long distance courses for students
• Create national recognition for Rensselaer, and enhance undergraduate recruitment
• Produce a substantial income stream for Rensselaer through tuition charged for these courses and through summer and academic year workshops to be offered to secondary school teachers (Rensselaer is presently negotiating with the College Board to be the sole provider in the northeast of enrichment experiences for teachers of high school advanced placement mathematics, science and technology courses)

Intel's involvement in such a venture would be highly visible and Intel's equipment would benefit not just the Rensselaer campus, but also the nation's high school students. Use of Intel's equipment would encourage migration of the equipment used in this program and other involved in it to IA based equipment. It would also give the high school students and teachers knowledge of what Intel equipment can do, perhaps encouraging personal and volume purchasing of Intel equipment. The program would use Intel equipment in unique ways demonstrating Intel's equipment capabilities in these special areas.

Engineering Studio Classrooms

Rensselaer, world-renowned for its engineering, has discovered that engineering programs are faced with the triple challenges of: 1) declining percentages of high school students choosing engineering as a collegiate major, 2) increasing diversity of those students who do choose to major in engineering, both in cultural background and high school preparation levels, and 3) continually increasing costs while revenues remain essentially flat. These challenges are exacerbated by the traditional pedagogy used for engineering education, which has remained relatively unchanged for several decades, and by the state of engineering educational facilities, particularly those used to develop professional practice skills, which are woefully out of date at most schools, despite the rapid changes in technology. If Rensselaer takes advantage of technological changes, a new model of engineering education will come into existence -- one that would meet the triple challenges of preparing engineers for tomorrow. With help from Intel, Rensselaer will well establish such a model, perhaps making it the norm for engineering.

The Electrical, Computer and Systems Engineering (ECSE) department at Rensselaer has begun development of a model which involves the use of Web-based, interactive multimedia materials in conjunction with a Studio format of course delivery. Interactive learning modules (ILMs) are used in Studio classes by faculty to help explain concepts and demonstrate applications and by students for specific learning exercises. Outside of class, the ILMs are available anytime, anywhere over the Internet, using either public or private computing resources. They are used by students for in-depth exploration of concepts and interactive practice with design and problem-solving activities.

Our approach is to create highly interactive classroom facilities that have state-of-the art information technology for display, simulation and network access, and appropriate disciplinary-specific technology to provide meaningful hands-on experimentation (images of which may be viewed at http://www.ecse.rpi.edu). Rensselaer has successfully pioneered this approach in introductory science courses and the current work represents the first Studio implementation in a discipline-specific engineering program and the first direct educational use of the Web-based ILMs. Our first venture has been a Circuits/Electronics Studio that has been in operation since September 1996. Two additional facilities are currently under development and will be in operation for the start of the 1997/98 academic year.

This facility is designed to hold 44 students working in teams of two, with each team using an interactive learning station consisting of a networked PC with a GPIB interface to electrical test and measurement equipment (oscilloscope, function generator, multimeter and power supplies), plus prototyping materials with which to create actual circuits. In this engineering studio, the class time includes: mini-lectures (no more than 15-20 minutes), pencil and paper exercises, interaction with Web-based, design, simulation, circuit construction and experimental measurements.

The ILMs are an integral part of the proposed engineering classrooms. The basic concept of the ILMs is to develop easy to use, highly interactive materials which simultaneously stimulate multiple senses. The student is first presented with the context of the material and challenged to explore successive layers of the material to gain a full understanding of the basic principles and concepts. Additionally, the student is encouraged to play in a highly interactive, highly functional design space where he/she can get immediate answers to "what if" questions and develop the problem-solving and design skills that such a valuable component of an engineering education. Students are also encouraged to use the ILMs outside of class for further understanding and practice, although we have not yet introduced required outside use of the ILMs. Outside use, judged by the activity of our servers, is already high. Currently available ILMs may be sampled at the following URL: http://www.academy.rpi.edu/.

Our objective is to create interactive materials that combine graphics, animations, audio/video, modeling and simulation in an enticing environment that allows the user to understand how electronic circuits operate, how they are designed, how they are manufactured and how they are utilized in real-life applications. The ILMs pick up where most textbooks leave off -- via the use of multimedia -- to additionally allow the user to explore and understand the dynamic nature of how electronic circuits are designed, manufactured and applied. For example, the capability to see and hear the result of how an amplifier increases the amplitude and volume on an input signal provides the user with a much more powerful understanding of the circuit operation. The hope is that this integrated presentation approach will ultimately generate a much greater impact than that which single-dimensional books or static web pages can accomplish. High-end implementation of Intel's equipment can be employed in all levels of creating and utilization of the ILMs.

Future activity will include ILMs on topics that dig deeper into the elements of design that are associated with cutting-edge products (e.g., greater than 250 MHz timing signals on a computer motherboard) along with an embellishment of the current ILMs. All of these will be aided with Intel's equipment, taking education and computer uses in education to new levels.

The early results of the combined ILM/Studio approach has been so encouraging that the ECSE faculty have voted to move all 8 of our introductory courses (typical enrollment is 80-150 per term) in Electrical Engineering and Computer and Systems Engineering to this format. Concurrently, several required laboratory courses will be eliminated from the curricula. These results show that Rensselaer is on the right path for a new approach to engineering education.

Rensselaer's national leadership in technology is grounded in the mission of educating professionals to use information technologies to support product development in a corporate environment. Historically, we began with a highly successful on-campus professional masters degree program aimed at entry-level professionals. A new Professional Masters Certificate Program in Human-Computer Interaction marks our entry into the emerging market of working professionals looking for career advancement.

Changes within the field of technical communication have created a need for advanced professional education. As technology has provided an increasing number of information design options, professionals educated for an exclusively-text-based culture face new challenges in developing the conceptual framework and mastering the tools necessary to produce good technical communication. Ten years ago, product information meant a paper manual. Today product information comes in a wide array of information formats ranging from paper to CD-ROM and now Web-based hypertexts. Information development tools provide technical communicators many options. No longer limited to producing linear text, technical communicators can integrate text with two- and three-dimensional graphics and link them non-linearly in a variety of ways. This increase in technical design options has resulted in the educational need for which the new Professional Masters Program in Human-Computer Interaction was developed.

With Intel equipment, we plan to extend our Laboratory in Human-Computer Interaction to support the design and modeling of human-computer interactions. Drawing on groundbreaking work in physical performance and innovative curriculum in three-dimensional communications, we plan to use the high-performance capabilities of the WindowsNT platform to integrate three-dimensional models into the design of human-computer interactions.

As part of its teaching and research program in human-computer interaction, Rensselaer's Department of Language, Literature, and Communication has established the Human-Computer Interaction Laboratory, a state-of-the-art usability testing laboratory. The laboratory is used as a test bed for funded and experimental research and for student projects on usability of interfaces, including both hardware and software. The laboratory is currently set up with equipment to monitor and record tests, but it is notably weak in equipment on which to develop and run test scenarios.

An HCI lab with Intel computers will be useful for teaching and research purposes. Since Intel processors and video compression are extremely fast, faculty and students will develop and test interactive multimedia interfaces without having their test results tainted by a slow delivery system. We would be able to run tests on the usability of various interface styles and formats and also tests comparing the same interfaces with different processor speeds. Intel has been a leader in video compression/decompression software since they purchased GE's DVI technology. Future applications will use more video, and the HCI laboratory can help to demonstrate that Intel equipment is a logical choice for testing new multimedia interfaces because of their work with video compression.

Electronic Arts at Rensselaer has a national and international reputation for its educational program in multiple media. Our offerings consist of EMAC (Electronic Media, Arts & Communications), an exciting new degree offered jointly by the Department of Language, Literature & Communication and the Department of the Arts, and an MFA in Electronic Arts. The undergraduate program brings in up to 50 students a year for a four-year undergraduate program. The MFA program admits 12 students annually for a two- to three-year program.

EMAC has just completed its first year. Though we have no graduates yet, initial reaction from industry representatives has been extraordinarily positive, and many of our current students are already being pursued for internships and Co-ops. The MFA program, which just completed its sixth year, has seen all of its entire alumni move on to careers directly related to their studies. This includes university level teaching, working in or directing academic or commercial studios, creative direction of advertising and Internet projects, and work for corporations like Disney and Voyager. In addition, our alumni have won major awards from the Rockefeller Foundation and the New York State Council on the Arts, and have exhibited their electronic art work world-wide.

Since the inception of the MFA program in 1991, electronic arts courses at Rensselaer have focused on the use of Macintosh-based hardware, in keeping with industry standards in the entertainment, music and graphics industries. However, as we move toward the new generation of hardware and software, it is apparent that we need to move our studio instruction to a fully multi-platform model, in order to best prepare our students for the commercial and creative professional worlds they will enter when they graduate.

In order to do this effectively, we will approach the goal in several stages. In the first phase (Spring and Fall, 98), we will provide electronic arts faculty with powerful IA machines optimized for music, video and graphics applications and plan initial cross-platform studios. In the second phase (Spring 99), we will install IA machines in two selected studios and teach a vanguard of two or three studio classes in the cross-platform model. Finally, in the third phase (Fall 99- Spring 2000) we will expand our facilities and our curriculum to be entirely multi-platform.

With the technological changes of the past three years the applications of the technology in research and education have also changed. Everything is faster and more capable and new software and hardware, perfect for educational and research applications such as simulation, modeling, visualization and parallel processing, have been improved or have come into being. These applications are ever more graphically and computationally intense, and keeping current is expensive. Rensselaer proposes to use IA-based equipment to create cost-effect methods to keep up-to-date state-of-the-art computing applications at Rensselaer. The computationally intense Data Mining, Parallel Processing, and NIC along with the graphically and computationally intensive, Pro/ENGINEER, Interconnection Design, Architecture studio305, and Interactive Electronic Media Development would perform to the highest levels of Intel's capabilities while pushing frontiers and boundaries.

Computer graphics has been a part of the engineering curriculum at Rensselaer since the late 1970s when CAD packages were first used in the freshman graphics course to supplement the classes on hand sketching. Solid modeling capabilities were added in the spring of 1991 with the inclusion of Pro/ENGINEER (instead of CADAM) in the freshman graphics course. With the introduction of Pro/ENGINEER, it was quickly realized that solid modeling should be the basis of the freshman graphics course with hand sketching used to supplement the solid modeling exercises. This laid the foundation for the present graphics course, Engineering Graphics & Computer Aided Design (EG&CAD).

EG&CAD is a first year engineering course that teaches the fundamentals of solid modeling and engineering documentation. The course has five objectives for the students:

1. Create and modify engineering parts with a variety of construction methods. In addition, introduce the engineering capabilities of solid models with dimension relations. (18 Hours)
2. Create detailed engineering drawings of engineering parts with principal orthogonal views, cross sectional views, and auxiliary views. (10 Hours)
3. Create hand sketches of engineering parts with both isometric and orthographic projections. (5 Hours)
4. Create and document assemblies of engineering parts. (3 Hours)
5. Create a final project incorporating all of the above. (6 Hours)

The course is taught using Pro/ENGINEER in a laboratory environment. At present, Pro/ENGINEER version 16 is used; the course will be taught with Pro/ENGINEER version 18 starting in the second summer 1997 session. Pro/ENGINEER was selected in 1991 due to its capabilities, performance, and system requirements (size of data files). The course consists of 14 sessions, three hours in length. The sessions are held in computer rooms containing 25-30 workstations running UNIX with AFS. The first 12 sessions consist of a lecture approximately 50 minutes long followed by two assignments that incorporate the lecture material as well as material from previous lectures. The last two sessions are for students to work on their final project. The laboratory time is led by two graduate students and an undergraduate teaching assistant providing a student/teacher ratio of 8/1.

The course consists of Pro/ENGINEER training files. Students begin the session by loading Pro/ENGINEER and running the tutorial. In addition, the instructor has a computer attached to an overhead monitor. The training files consist of text pages to present theory followed by demonstrations that apply the theory. The demonstrations also show how Pro/ENGINEER is used (menu picks, etc.). The training files are stored in an AFS locker and can be accessed by all Rensselaer students and faculty (anyone with a *.rpi.edu account). Using AFS, multiple students can access the lecture material and changes made to the lecture material are made to one set of files and instantly available to the students. Finally, there is a textbook containing the lecture and laboratory materials.

EG&CAD is not intended to complete the students' knowledge of solid modeling. Solid modeling and documentation are included in design courses with the intention that the student will use Pro/ENGINEER for each of the four years. In addition, special courses such as Advanced Topics in Solid Modeling and Advanced Manufacturing Laboratory have large solid modeling requirements.

Future enhancements include upgrading the Pro/ENGINEER software and increasing the number of available platforms. Pro/ENGINEER is released twice a year. At Rensselaer, the even number versions are installed as these versions are shipped in the spring, giving the summer to update files, the manual, and test the new version with 10-15 students. At present, Rensselaer can run Pro/ENGINEER on SGI and IBM workstations with UNIX and AFS. Rensselaer has obtained a contract to allow the use of PC platforms using Windows95 or WindowsNT. This is an exciting development in that a PC based platform allows for students to have more control of their data. In addition, students will have the opportunity to purchase a student version of Pro/ENGINEER giving them the ability to use the software at home. At present, we have tested Pro/ENGINEER version 17 on a Windows95 platform and found that we can access the training files and move data easily from the PC to UNIX/AFS and back if desired. Our goal is to have one section of EG&CAD taught on a PC platform for the fall 1997 semester. If successful, the PC platform will be expanded to include interactive lectures using existing software. Given the price performance of PC hardware over a workstation, the PC platform has the potential of becoming the preferred platform for solid model creation.

Together Rensselaer and Intel can test heavy engineering applications, such as Pro/ENGINEER, on a Windows platform, and eventually move the UNIX application over to a Windows based platform. As Rensselaer moves such applications to WindowsNT it is likely many other colleges and universities will follow suit, along with industry.

Over the last three years the Rensselaer campus has experienced an explosive growth in the development and application of high-performance computing and visualization in research and educational activities. The computational needs of these activities have been met through major investments in high performance computing including a parallel computer, workstation clusters for numerically intensive computing, a high-performance visualization laboratory, and distributed workstations. Rensselaer's strong interdisciplinary research programs provide the ideal environment to advance simulation capabilities for important new physical phenomenon and to apply those procedures to advance the science of the application area. One of the major research directions in the Scientific Computation Research Center (SCOREC) is the development of parallel algorithms for adaptive scientific computation. Some of our current applications are in the areas of automatic mesh generation, soft tissue analysis, material processing, rotorcraft aeroelasticity and multiscale failure analysis for composites. All of these applications are very demanding, not only in terms of sheer computational requirements but also in terms of system resources such as memory and I/O. The only practical way to perform many of these computations is in a parallel environment. Over the past several years we have developed a variety of tools to allow both data and computation to be distributed in a parallel environment in a manner which is independent of the details of how the parallel system is constructed.

Rensselaer currently has a scientific computation infrastructure that includes a 36 processor IBM SP-2 parallel computer, a Visualization Laboratory for Scientific Computation (VLSC), and a large number of high-performance workstations. Currently most of the SCOREC work is done on the SP-2. This distributed computing environment is used by almost every major research group on campus, and Rensselaer has become a leader in the development and application of parallel scientific computation techniques. However, usage is already near saturation and productivity is constrained by the speed of the existing campus network.

Many of our industrial and government sponsors have shown interest in, and supported, this distributed computing environment. However, they have also stated that this would be even more beneficial to them if it was possible to do parallel adaptive computations on networked workstations instead of only a special purpose parallel machine such as the SP-2. We believe that it is possible to do this effectively using the tools that we have developed, if the network connecting the workstations is of sufficiently high performance. Intel could join the exploration of this frontier, through which new uses and boundaries for Intel's equipment will be discovered.

For real world problems, communication bandwidth and latency dictate how effectively a computational environment can be used for parallel computation. One great benefit of the SP-2 is its high-speed switched communication between nodes of the machine. We know that the speed and latency of 10 Mbs Ethernet makes it very difficult to get sufficient speed from an adaptive parallel computation, thus making the "network of workstations as a parallel machine" idea ineffective. We believe that a higher speed network, such as one based on ATM, connecting a group of workstations is likely to allow practical parallel computations to be performed. Such a network would be a very cost-effective solution for companies looking to take advantage of parallel processing.

We currently have two Bay Networks switches to support the communications, and we are planning to link UNIX-based SMPs from SUN and IBM into a parallel cluster on which we will tune our current codes to take better advantage of the SMPs shared-memory. We propose to work with Intel on the development of a WindowsNT based cluster consisting of 6 four-processor Pentium based systems. The efforts to be carried out would include:

• Convert the SCOREC parallel adaptive analysis tools to a WindowsNT environment. Since our current parallel tools use MPI, we can take direct advantage of the Intel supported work at Mississippi State University on the development of a version of MPI for WindowsNT.
• Tune software to work effectively in that environment. These efforts will first focus on the conversion of the SMP operations to take better advantage of the shared-memory environment of the SMP than is possible with MPI.
• Perform simulations on the WindowsNT parallel cluster and document capabilities of the system. High-performance distributed computing is key to the development of the next generation of instructional delivery systems. Although the computational demands are different from those of large scale research simulations, distributed multimedia-based delivery systems are just as demanding from a data throughput viewpoint, and can be even more demanding from the viewpoint of supporting interactive responses to a large number of users. The proposed instrumentation will allow Rensselaer to move to the next level in high-performance computing and communications in Rensselaer's research programs while utilizing Intel's equipment. The proposed program will then:
• Apply distributed parallel computing to support research in the areas of materials processing, multiscale failure analysis, multiphase flows.
• Develop and employ virtual reality techniques to investigate the results of large-scale simulations and to interactively steer the simulations.
• Research, develop, and evaluate student mobile computing for delivery of anytime, anywhere instruction.
• Further research and develop advanced networking.

The deployment of an ATM-based network would support all facets of this program.

Demonstrating such a system to our sponsors would be beneficial for Intel since it would give companies an additional concrete reason do parallel processing in a WindowsNT environment. Many of our government sponsors, such as NASA, NRL, NSF, ARO, ONR, ARPA, LBL, and WPAFB, and our industrial sponsors, such as CTC, GM, GE, IBM, Martin Marietta, MSC and Schlumberger, visit us on a regular basis and will see the system in operation first hand.

The proposed project focuses on research, which relies on a highly distributed, increasingly parallel, computing environment, interconnected by a high-speed ATM network. Parallel processing remains a cornerstone of the proposed activities due to the computational demands of simulations needed in many fields of science and engineering. We will continue to rely on our centralized parallel computer server, but also focus on performing parallel processing in a distributed heterogeneous environment, taking advantage of symmetric multiprocessor WindowsNT workstations which represent a cost effective method to support large scale calculations. The development and use of visualization will be extended to full immersion virtual reality, which will be used to interactively steer large-scale distributed parallel simulations. Educational research on student mobile computing for delivery of anytime, anywhere instruction will be performed, forming a much closer link between research and education at Rensselaer. Finally, research specifically in advanced networking will be conducted. An Advanced Network Testbed forms the backbone on which all these research projects rely.

The requested computing and networking equipment is central to the research programs of six of Rensselaer's research centers and two of its largest academic departments. The specific centers are the Center for Composite Materials and Structures, the Center for Image Processing Research, the Center for Integrated Electronics and Electronics Manufacturing, the Center for Multiphase Research, the International Center for Multimedia in Education, and the Scientific Computation Research Center. The faculty in these centers represent 13 academic departments in the Schools of Science and Engineering. The two academic departments with specific focus on the networking developments are Computer Science, and Electrical, Computer and Systems Engineering. Other departments may, with the proposed program, also find uses for high performance computing, further integrating the Institute via parallel computing.

To support these research and educational activities we are requesting equipment and support to provide cost effective distributed parallel processing based on ATM-attached SMP nodes. Using such Rensselaer plans to enter the next level of high-performance parallel computing based on a new method we have developed. Intel could work with us on creating this method of cost effective distributed parallel processing in which we will utilize the IA equipment in new ways and show the world what Intel equipment and Rensselaer ideas can do.

At Rensselaer, numerically intensive computing is not an ancillary activity practiced by a handful of faculty members, but a fundamental research activity engaging many faculty members and graduate students. A cost-effective solution is to provide a Numerically Intensive Computing (NIC) service using off-the-shelf components, especially WindowsNT workstations using Intel processors. This would allow less expensive desktop computers to be purchased for faculty while filling the need for intense numerical computations.

Rensselaer already has a NIC service based on UNIX. However Rensselaer would like to replace the expensive outdated UNIX workstation to better fill the numerical computing need of our faculty. There are many goals for the WindowsNT based NIC, some of which are similar to the UNIX-based system. One goal is to provide a readily accessible development environment for NIC users, which facilitates the transition from test to production runs. The service would be integrated with the Rensselaer computing environment - authentication using RCS userIDs, access to the campus central file system space and printing. Distributed Computing Environment (DCE) will provide authentication. It would use a distributed batch queuing system to submit and schedule NIC jobs. Another goal would be to have DQS provide the Batch Queuing system. The features of this queuing system will be used to maximize use of the NIC hosts for both parallel and serial batch jobs. The NIC would use a commonly accessible filesystem to minimize data movement between systems. The Distributed File System (DFS) and SAMBA will provide access to the campus central file system. Access to PC file space will be provided by Windows networking. It would also provide both serial and parallel NIC environments. MPICH will be used for parallel programming message passing. The final goal is to provide user training in such areas as FORTRAN, parallel programming, and code optimization. Rensselaer's traditional training courses will be augmented with information on optimization in the Intel/NT environment.

We would make this enabling technology available to departments wishing to provide their own NIC capability, an exciting side-effect of using "commodity" desktop hardware. Using DQS, DFS, and/or MIPCH to implement a windfall system, departments will build a NIC environment instead of, or in addition to the campus-wide service.

Rensselaer has recently has finished negotiations for a campus-wide licensing agreement with The Numerical Algorithms Group, Inc., for the NAG package, which includes a FORTRAN compiler and library that run across all platforms (Win95, WinNT, UNIX), as well as the IRIS Explorer, which can be used for data visualization and visually controlling processing. Axiom and Maple are also available for solving large symbolic problems. Other compute intensive applications will be available include:

• SAS for data mining
• FIDAP for fluid dynamics work
• AMPL for optimization
• MatLab for numeric computation
• Abaqus
• Patran

Rensselaer is willing to partner with Intel to provide equipment and services for the NIC clusters. We will provide the Networking infrastructure, standard NIC and WindowsNT training, programming staff to port/integrate DQS, DCE, MPICH, and assessment of the relative merits of the WindowsNT/Intel NIC environment vs. our current UNIX-based NIC services.

Rensselaer sees the switch from a UNIX-based NIC to a WindowsNT-based NIC on Intel equipment as high end use of Intel's processors and beneficial demonstration of the equipment's performance capabilities. Other schools with similar needs may follow Rensselaer with proof this is a cost effective and efficient method of desktop numerically intense computations. It will also further familiarize the faculty and some students with Intel equipment and uses and likely encouraging them to purchase Intel for personal or other work related uses.

Interconnection limitations will severely constrain the performance of future generations of microelectronic chips and systems. Accurate modeling is essential to determine the necessary tradeoffs in materials, processing, and design. In particular, modeling will provide a framework for estimating the performance and reliability of present and proposed interconnection designs and processes. The Center for Integrated Electronics and Electronics Manufacturing (CIEEM) has a coordinated effort underway in interconnect modeling and simulation that leverages almost 20 years of Rensselaer's microelectronics materials research and industrial experience. E-CAD tools are being developed at Rensselaer in collaboration with industry, which link performance to interconnection materials and system architecture. Work in CIEEM is focused on developing models and simulators that address advanced interconnection technologies with higher levels of integration and performance, lower power, increased reliability, and reduced cost. Our CAD tool development incorporates highly efficient estimators of interconnection capacitance, resistance, and inductance (developed at Rensselaer) as well as our design knowledge from the development of fast RISC architectures. These activities additionally include on- and off-chip electrical and optical interconnection optimization. Both reliability and performance now need to be modeled, but typically require high performance numerically intensive workstations to utilize the results in a timely manner. The IA-based workstations will allow for reliability modeling and simulation of thermomechanical stresses in complex interconnection structures to be performed, and will provide accurate predictions of both the global behavior and local field parameters in the critical regions.

Our research is intended to develop predictive models based on physical, electrical and chemical principles, verifiable by experimental results obtained in various research activities. Our approach to interconnect modeling and simulation address three key limits which interconnections impose on the performance of electronics - wiring space, critical path lengths, and signal degradation. Models of materials properties and their dependence on various process parameters (e.g. materials reliability issues such as EM and SM) require tremendous amounts of compute power to completely model and simulate interconnect performance. We propose to utilize efficient numerical procedures developed at Rensselaer for extracting interconnection resistance (R), capacitance (C), and inductance (L), in order to accurately simulate signal degradation. The key physical limits to future digital systems imposed by interconnection technology and architecture will be identified and innovative techniques for approaching these limits and revealing promising opportunities that enable continued progress of digital systems towards higher levels of integration, higher performance, lower increased reliability, and reduced cost will be developed. The research results from the equipment requested in this proposal would not only lead to an optimization of the interconnection technology, but also provide stand-alone models that can be used in related research and development areas.

There are four key interconnection-related processes: material deposition; material removal by etching to create patterns; planarization and annealings to which interconnections are subjected. Materials to be concerned with are metals, diffusion barriers, and dielectrics. Realistic process models, verifiable by actual experimental results, are key to achieving a low-cost and reliable process. We propose to research various ways to model the processes, knowing that some of the above-mentioned processes are too complicated with too many variables, e.g. CMP, which is listed with over 40 independent and dependent variables. This complexity is a perfect example of how the acquisition of the proposed equipment will greatly accelerate process understanding and implementation. Our goal is to break up a complicated process into smaller units of physical/chemical models that are adaptable in various applications. These unit models will then be combined to characterize and simulate the complete process with verifiable results using the cluster of high-horsepower Intel workstations. We have developed a number of algorithms and generic tools that can be used for deposition, including modeling of step coverage, via or trench fills and in some cases for etching.

• A new adaptive-mesh finite element code, which can be, used for modeling complex fluid or solid reaction-diffusion systems with irregular boundaries.
• A new free-boundary code that enables one to obtain exact solutions of problems in which the boundary moves in a complex manner and its local position is a time-dependent part of the solution for the problem.
• A new free-boundary code that enables one to study feature and reactor-scale models for vapor-deposition of low K polymer films in which growth occurs by solid-phase diffusion with bulk rather than surface reactions.
• A new symmetric CVD reactor-scale physical/chemical 3-D model code that handles multiple nonlinear gas phase- and surface-reactions and can be applied to thermal/CVD processes for both dielectric and metallic films.
• A new transition-region theory of diffusion for the CVD of metals and dielectrics in vias and trenches, in which both particle-particle and particle-film collisions are important, and
• A new approach to PVD modeling based on kinetic theory, which simplifies calculations for complex-surface geometries.

A simple and effective method of conjugating, or combining, these approaches has been recently been developed which employs the basic concepts of matched asymptotic expansions, and enables one to include the effect of process tool dynamics on feature scale structures. The proposed workstations will again be used to both execute these codes and provide real-time process simulation and modeling ability.

The past three years have seen the emergence of data mining (DM) as one of the most relevant new topics in information technology. The basic problem in DM is how to extract information from the huge amount of data that gets amassed using computer technology. Inherently cross-disciplinary, DM integrates theory and practice from statistics, mathematics, decision sciences, artificial and computational Intelligence, and database management and then applies them to practical applications in business, science, and engineering. Rensselaer has developed a group, the Rensselaer Knowledge Discovery and Data Mining Group, to make Rensselaer, a recognized leader in the industrial application of DM, a world-renowned center of research and education in data mining. Using our pool of expertise in diverse disciplines, we have already begun developing cutting-edge DM technologies. Our goal is to build relationships with industry and government to gain expertise and visibility in the development and use of our technologies on important DM applications. Since our formation of the group in 1995, we have made significant progress, illustrated by a long list of publications, which will be provided on request, and three doctoral theses. We have met with and/or received problems and data from such companies as: AT&T, GE, the US Forest Service, Chase Manhattan Bank, Wells Fargo Bank, Citicorp, Equifax, and the New York State Department of Health. The problems and data sets range from Nielsen ratings analysis, to tree mortality, to credit card fraud detection, and have been used for both research and Rensselaer courses. Not only would DM utilize Intel's new powerful processors in highly visible ways but would allow Intel to become part of Rensselaer's cutting-edge world renowned center of research and education in data mining. Such a partnership would be advantageous to both Intel and Rensselaer.

The interdisciplinary nature of DM allows us to build on existing faculty expertise across schools. It integrates the best of traditional statistical analysis with nonparametric methods from artificial Intelligence, operations research, and mathematics. It currently includes collaboration among the Mathematics, Computer Science, and Decision Sciences and Engineering Systems departments. The infrastructure for data mining can be shared with other proposed initiatives in the schools of Management and Science (e.g. finance-related application). By developing partnerships with industry, the group will generate net revenue and become a world-renowned center for DM and its applications. In addition, the work supported by this center will lay a valuable foundation for support of both existing courses and a future information technology curriculum.

Current work in data mining at Rensselaer involves database marketing, improving marketing yields by targeting customers using metrics developed by mining prior customer data. Our approach is to combine and integrate neural networks, decision trees, statistical discrimination analysis, and mathematical programming, into hybrid configurations appropriate to the problem domain. The three involved departments bring research expertise in many different areas applicable to data mining including: mathematical programming, machine learning, computational Intelligence, neural networks, statistics, operations research, decision analysis, data visualization, database management, data mediation, and time series analysis. These areas of expertise enrich Rensselaer's data mining and open a wide range of possibilities for the future of data mining.

The key difficulty in accomplishing a successful data mining operation is the massive quantity and irregular quality of the data sets typical for data mining operations. Large amounts of computer storage space are required. Data needs to be cleaned and different data formats have to be accommodated in order to use a range of different software platforms. Industry expects us to already have an infrastructure in place for DM applications. All this would demonstrate the capabilities of IA-based systems with high-end usage. Ultimately all this work will be supported by industry. But to get started immediately we first have to provide an infrastructure.

The goal of the proposed effort is to provide the necessary infrastructure to make our industrial contacts more concrete and productive. We need to be able to make a rapid and effective response when we first obtain industrial problems. To achieve this we need three things: computational support, student support, and matching funds. Intel will provide us with the first and thus help us obtain our other resources. The objectives to be accomplished are as follows:

• Develop an infrastructure at Rensselaer capable of processing and analyzing massive data sets of at least 100 MB.
• Propose active relationships with several companies including Chase Manhattan Bank and GE. To do this we must establish concrete partnerships with industry to implement data mining applications. This may involve industrial support of faculty visits, faculty time, student internships, students support, and equipment.
• Pursue at least one of the government-sponsored funding opportunities for joint research between industry and academia such as NSF Grant Opportunities for Academic Liaison with Industry and NSF University-Industry Cooperative Research Programs.
• Establish, after one year, a concrete cooperative effort between Rensselaer and at least two separate industrial firms, to have students doing extensive on site collaboration with industry, and having eight graduate students doing research on DM related activities.
• Develop a strong presence in the knowledge discovery and data mining research community through publication of research papers, presentation at conferences, and organization of conference sessions. Specific targets include the special issue of INFORMS Journal on Computing on DM, the International Conference on Knowledge Discovery and Data Mining, and the INFORMS Conference. In addition, our work will be disseminated via a comprehensive web page.
• Construct and support a library of real-world data available campus-wide for research in DM and for student projects. This material will be a great resource for curricular enhancements of course directly or indirectly related to DM.
• The establishment of a web site with a goal of 10,000 visits after one year. A preliminary web site, set up in the context of a data mining course, resulted in over 1000 visits from December through February (www.rpi.edu/~vanepa2).

The proposed initiative of this proposal is to make Rensselaer, a recognized leader in the industrial application of DM, into a world-renowned center of research and education in data mining. The work on this project will directly benefit at least 6 faculty members and an estimated 250 students via course offerings, Ph.D. research, and industrial partnerships. Once established it is certain more faculty, departments, and students will benefit from the data mining center, some in ways we might not yet even consider.

Data mining at Rensselaer is an Institute collaboration based on technology. At the Data Mining Center different departments and schools are drawn together by data mining needs and interests. With work this center will become the recognized leader for data mining worldwide and set standards for the future of data mining. The use of Intel technology and equipment will further improve the center. The DM center will take full advantage of IA based equipment, demonstrating and maybe even pushing the performance capabilities, and will do this in a highly visible setting. Not only will industry and other colleges and universities turn to Rensselaer as the leader of DM, but they will also imitate Rensselaer's methods. As part of the center, Intel may soon be approached regarding the use of IA-based equipment in data mining. Together Rensselaer and Intel can create the cutting-edge, world-renowned Center of Data Mining Research and Education at Rensselaer.

The School of Architecture is embarking on a complete restructuring of its design curriculum.. This effort embraces the most advanced technologies accessible for digital/non-digital (multi-modal) collaboration, interactive and photo-realistic visualizations, large-scale rapid-prototyping, three-dimensional scanning, and advanced communication of building construction information. A complete a renovation is planned for of one of our design studios to support these applications. The IA-based platform would place us in a position to complete our cutting-edge development, as it would integrate the most advanced Intel-based tools into this environment.

The School of Architecture has had many recent projects and investments including the Lab for Informatics and Architecture. Created in the summer of 1996, it is an advanced research lab filled with high-end Silicon Graphics and NeTpower Pentium-Pro visualization equipment to support advanced research and design projects. Another project is the development of new post-professional degree in Informatics and Architecture (1997-98) along with the planning of advanced technology-intensive degree to be offered beginning in the 1998-1998 academic year, one of three new degrees to be offered. We are planning for the new Virtual Studio Design facility (fall 1998), a cutting-edge research facility for interactive virtual sets, incorporating real-time video consisting of live-action, computer-generated sets, and human-controlled, computer-generated actors.

Studio305 is another example of the investment in architecture advancement at Rensselaer. The restructuring of the studio has one important constraint: all technological additions must be seen as infrastructure, rather than workstation-based. The studio will include high-speed networking connectivity and file- and web-server access. Integrating built-in large format monitors and networked projection and collaboration tools, we are intending to encourage users to come prepared to connect, with their own personalized equipment, from laptops to the most advanced UNIX or WindowsNT based workstations.

The studio will integrate into this network:

• 3-D and large-format 2-D scanning devices
• selective laser-sintering rapid-prototyping equipment
• interactive, networked projection systems
• professional quality video editing hardware and software
• advanced web design and management software
• multi-processor equipment for photo-realistic and interactive rendering.

The proposal would locate, within this network, a multi-processor, or multiple multi-processor rendering servers (to attain 32 processors) that will be accessed in a distributed network setting. This server would be able to support advanced virtual-reality based interactive rendering and multi-processor rendering for photo-realistic animations. By allowing students to create interactive VR-based presentations of complex, textured 3-D models, and by allowing the turn-around process for rendering long animations to shift from one week to a few hours, we will be revolutionizing the digital design process. Typically, this level of equipment has only been accessible to commercial animation companies. Even the profession of architecture has not seen equipment of this sort dedicated to its own interests. We stand firmly resolved to overcome this barrier, bringing such technology not only to the aid of final productions within architecture, but also to the iterative process of design. We will seamlessly integrate such technology into the multi-modal design network to overcome these barriers in the profession and in the firms in which our graduates are employed.

The Academy of Electronic Media (AEM) and Interactive Electronic Media Development

The Academy of Electronic Media (AEM) aids in the development and utilization of multimedia materials specifically tailored to simulation/visualization research and technical education, and to demonstrate their effectiveness in academic, industrial and commercial environments. Compelling interactive learning modules (ILMs), design/analysis/modeling tools, challenging simulation exercises and games, tutorials and case studies, and new avenues for learning are being developed to allow users to explore materials associated with learning about and applying science and technology. We intend to utilize these in a continuum of high school, undergraduate and graduate level curricula, corporate training programs and, ultimately, throughout life (providing web access tools to facilitate anywhere at anytime access).

Specifically, AEM's NSF-sponsored Interactive Learning Modules (ILM) project has created a wide variety of multimedia educational applications over the past 3 years. These modules are being used in the studio classroom environment as well as being deployed over the WWW. An Internet available interface to access, control and utilize an x-ray laminograpy system has been developed and implemented for use by those interested in inspecting electronic modules and assemblies. AEM has also been involved in interactive learning technology development involving sound, video, animation, complex 3D graphics, and numerous uses of QuickTime Virtual Reality (QTVR).

A wealth of expertise has been accumulated by the AEM developers-- faculty, staff, graduate and undergraduate students--in interactive authoring, programming, and in the preparation and incorporation of media elements. However, there are tools that are available for authoring that would greatly enhance the effectiveness of our modules but which are largely inaccessible to our use, by virtue of the processing speed barrier imposed. Building, rendering, and real-time motion of 3-D objects and environments is so painfully slow on a standard PC that it makes any serious design of true three-dimensional worlds impractical. This design limitation curtails the depth dimension of materials and hence the degree of immersion (impact and memorability) for the student--the engrossing quality that high-quality commercial games and edu-tainment are rarely without. We are in the position of having both a wealth of valuable educational content and a cadre of experienced developers and not being able to make the move (literally) to the next dimension, where we can use CAD-CAM tools, world-building tools (like Bryce), and even the simpler 3-D modeling programs (e.g. Extreme 3D).

The proposed acquisition of the Intel machines will also serve AEM's need for robust, multipurpose WWW servers, on which to coordinate on-line interactive multimedia materials for course delivery and use in virtual classrooms of 150 or more students (30-50 locally -- and 100 or more remotely) in conference or class at a time. The exploration and creation of the z-axis, of virtual depth, is one of the most important aspects of computer aided and mediated communication and learning. We are at crest of a rhetorical revolution, thinking into-through-and learning from interaction with the computer.

Rensselaer is requesting $2,981,842 in equipment support from Intel. Rensselaer is matching that amount with $3,290,530 worth of software acquisitions, training, networking, space renovation and other miscellaneous expenses (see table below). A detailed budget, which includes an overall summary, a project by project summary and detailed configurations, is provided in Appendix B.

Through our interactions with Intel personnel, we have been told that most of the equipment will be available for delivery before the end of calendar year 1997. This is compatible with our timing to begin deployment of equipment in late 1997. However, there are some projects, for example the Pittsburgh Building where we will be constrained by the timing of the renovation work (Pittsburgh is currently scheduled to be completed by Fall of 1998). We expect that further interactions are necessary, between Intel and Rensselaer, to more precisely define this timeline.

Rensselaer educates the leaders of tomorrow for technologically based careers. We celebrate discovery, and the responsible application of technology, to create knowledge and global prosperity.

With Intel as a partner Rensselaer will continue to change current methods of education and research, while bringing cutting edge technology into both arenas. Specifically we will:

• Create new learning modules and learning spaces and demonstrate on a large scale the new Rensselaer model for on-site studio and distance learning.
• Find new ways to use the IA systems in our most demanding education and research environments.
• Produce new software tools for research in parallel processing, data mining and numerically intensive environments.
• Expand the uses, and platforms, of computing in areas such as electronic media and arts and architecture and distance learning.
• Understand better the interactions of the human and computer.

Rensselaer and Intel are well poised to establish an internationally prominent demonstration of the power of IA computing for the students who attend Rensselaer, the thousands of visitors who come to our campus each year and our colleagues in Higher Education who constantly observe Rensselaer innovations in education and research.