Report on Workshop on High Performance Computing and Communications for Grand Challenge Applications: Computer Vision, Speech and Natural Language Processing, and Artificial Intelligence

This paper reports the findings of the Workshop on High Performance Computing and Communications (HPCC) for Grand Challenge Applications: Computer Vision, Speech and Natural Language Processing (SNLP), and Artificial Intelligence (AI). The goals of the workshop are to identify applications, research problems, and designs of HPCC systems for supporting applications in these areas. In computer vision, we have identified the main scientific issues as machine learning, surface reconstruction, inverse optics and integration, model acquisition, and perception and action. Since vision algorithms operate in different levels of granularity, computers for supporting these algorithms need to be heterogeneous and modular. Advances in technology, new architectural concepts, and software design methods are essential for this area. In SNLP, we have identified issues in statistical analysis in corpus-based speech and language understanding, search strategies for language analysis, auditory and vocal-tract modeling, integration of multiple levels of speech and language analyses, and connectionist systems. Similar to algorithms in computer vision, algorithms in SNLP require high computational power, ranging from general purpose supercomputing to special purpose VLSI systems. As processing has various requirements, a hybrid heterogeneous computer system is the most desirable. In AI, important issues that need immediate attention include the development of efficient machine learning and heuristic search methods that can adapt to different architectural configurations, and the design and construction of scalable and verifiable knowledge bases, active memories, and artificial neural networks. A computer system for supporting AI applications is heterogeneous, requiring research in high-speed computer networks, mass storage and efficient retrieval methods, computer languages, and hardware and software design tools. Research in these areas is inherently multidisciplinary and will require active participation of researchers in device and networking technologies, signal processing, computer architecture, software engineering, and knowledge engineering. Besides extending current frontiers in research, an important aspect to be emphasized is the integration of existing components and results into working systems.