|
"Unleashing Internet-scale data"
Digital media, pervasive sensing, web authoring, mobile computing, scientific and medical instruments, physical simulations, and virtual worlds are all delivering vast new datasets relating to every aspect of our lives. A growing fraction of these datasets are going unused or being underexploited due to the overwhelming scale of the data involved. Effective understanding and use of this new wealth of raw information poses one of the great challenges for the new century.
As the world grows increasingly data-rich, our interaction with such overwhelming problems has already gone from indirect and rare (supercomputing, traditional “big iron” applications) to daily (today’s web search) and will further expand to become continuous and non-stop (e.g., lifetime recording/retrieval, accurate machine perception, trusted
electronic advisors managing our news, activities and interruptions, and effortless automated language translation).
This theme aims to draw together and accelerate research into
1)5 Data-Rich applications, and
2)5 The processing, storage, transport, and management systems needed to support these applications.
Current Data-Rich Computing research efforts underway include:
-
Video Scene Understanding. Effective computer vision processing of high resolution video streams is beyond the capabilities of today’s platforms. One promising, recently developed algorithm for event detection in video sequences is over 1000X away from real-time performance on single workstation cores today. We are studying ways to parallelize and accelerate processing, storage, manipulation, and management of derived data for computer vision processing
of video streams—eventually aiming at real-time or faster than realtime performance that could enable new modes of human-computer interaction, such as sophisticated, markerless gesture recognition.
-
Collaborative Neighborhood Networks. The flood of data is not restricted to the enterprise. Our homes and apartments are becoming more data-rich as well, and in this emerging environment communication and transport layers are becoming barriers to new applications. Current broadband solutions have been shown to be limited both for fast upstream bulk transfers as well as real-time, peer-to-peer services. We are exploring the potential of wireless technologies to bridge this gap.
-
Cluster Provisioning and Management. Data-Rich systems will often be asked to scale effectively across machine boundaries and in the face of changing resources. Execution parameters, communication settings, and problem partitioning may need to adapt to network conditions, data placement, node failures, or hardware upgrades. We are developing an open, virtual-machine-based infrastructure based on ideas from cloud computing, to deliver an effective, resource-aware provisioning and management system for Data-Rich compute clusters. The goal is to make managing complex Data-Rich systems comparable in complexity to the management of today’s single-machine server environments.
Data-Rich applications will drive quality of life improvements, generate dramatic economic opportunities, and enable scientific breakthroughs, thereby producing new capabilities that are so compelling at a personal and professional level as to become an essential part of our daily lives—hence this theme’s connection to and place within the broader Essential Computing vision at Intel Research.
|
|
| |
| Who we are |
 |
|
Meet some of the researchers that drive our Essential Computing.
|
| |
|
|
