|
Symbolic (Artificial Intelligence)
and Net-Centric Computing
In addition to experience in such numerically rich problem domains as those described above, he also has developed solutions to other problems that required new computational models, new data structures, better search and ordering algorithms, and new object-theoretic methods to develop tight, efficient—yet readily extensible and adaptable—implementations.
Involvement with this class of abstract problem-solving development—symbolic versus numerical computing—began with the distributed-AI that characterized the Pilot’s Associate at Lockheed, and led to other distributed AI-based development efforts at Unisys, Contel, and GTE.
- At ISR, developed neural network-based strategies and methods for the verification and validation of OLNN’s (online learning neural networks). Results published in “Certification of On-Line Learning Neural Networks,” ASC 2003, July 2003.
- At ISR, developed CONOPS for the implementation of a sensor machine-to-machine (M2M) capability that is to be GIG-based (Global Information Grid). The underlying approach involved leveraging of Internet telecom technologies (SIP, SIMPLE, etc.) that GIG development already has committed to employ.
- At GTE, collaborated with Bellcore in the development of their ISCP (Intelligent Service Control Point) SLEE platform, service creation environment, etc. based on a proprietary graphic UML-like language (a la, flow-chart programming) to better enable the non-programmer to symbolically capture the requirements of the service, whether by the customer (carrier), a third-party provider, or the end-user.
- At GTE, provided technical oversight—from the system specification stage through testing—of GTE requested enhancements of the ISCP that were formally tested including being used to prototype advanced services in the GTE AIN testbed—an advanced testing facility already discussed above.
- At GTE, developed conceptual design, and provided technical direction to DGM&S (now Ulticom) for the contracted development of commercialized AIN Service Creation and Service Logic Execution Environments—SCE and SLEE, built around such cutting-edge technologies as Versant’s OODBMS and implementing the customer-centered symbolic reasoning concepts developed during his work on the High-Value Platform effort.
- At GTE, presented papers—both internally, and later professionally at conferences—on the proposed design, supporting symbolic programming principles, etc.—such as “AIN System Development—A Customer-Centered Approach,” TINA-95, Feb. 1995.
- At GTE, served as technical liaison to several standards-body organizations—influencing strategic direction, functional requirements, API’s and specifications, testing and validation requirements, etc. For example, worked closely with the development of the OSGI specification—expressed using JAVA.
- At Unisys and Contel, developed symbolic AI-based computing environments in
Lisp, ADA, and C that prefigured many of the object-oriented features now standard in such languages as Java. For example, the implementations incorporated the ability to perform JIT—just in time—late binding, late compilation, as well as self-introspection capabilities (a la, Java’s reflection mechanism) required for an intelligent system or application to explain itself to other intelligent systems.
- The AI-based systems were developed to provide SDK toolsets and environments
—including inspector-editor-debugger capability, etc.—for use by other R&D projects and eventually in support of commercial contracts. Consequently, they were subjected to extensive and rigorous testing and evaluation from their beginning conceptual design.
- At Lockheed, integrated cutting-edge distributed AI methods and techniques—
e.g., the Negotiation Metaphor of Randall Davis, and the Functionally Accurate Cooperative Systems of Victor Lesser, with new man-machine cybernetic principles—e.g., those of Dr. William Rouse of Search Technology, together with data fusion systems—e.g., the Blackboard paradigm of the HEARSAY project, to develop a winning proposal for the multi-million dollar Pilot’s Associate DARPA project.
|
