Theme
Analytics-driven enterprises have moved into the main stream, and traditional enterprises are increasingly using analytics-driven approaches to achieve their goals. Assimilating information across disparate domains (e.g., intelligence, economics, and science) provides a comprehensive view of complex systems enabling deep insights and improved decision making capabilities. The alignment of government, industry, and scientific analytic requirements is now creating a significantly larger market than for traditional HPC, making it possible for stakeholders to influence the development of commercial technologies. Government investment in advanced analytics capabilities could potentially be more broadly leveraged than any prior investment in computing; but only if we carefully rethink compute infrastructure, identify effective techniques, and benchmark solutions from the unique perspective of data analysis.
Patterns are emerging as a powerful medium to analyze petabyte data sets across a wide variety of fields, including social networks, biological systems, financial transactions, economic ecosystems, criminal activities, and cyber security. Patterns can succinctly capture the occurrence of events and the functionality of systems. The pattern to the left depicts the Denial-of-Service Smurf Attack. When annotated with auxiliary information, time stamps, geographical location, and confidence limits, the pattern can capture unfolding activities, evolution of the attack, the functionality and dependencies of component parts, and the uncertainty of outcomes. The pattern can be decomposed into smaller motifs that can be identified and tracked over time to follow an active attack and/or predict next events.
At this second CLSAC workshop, we will discuss using patterns to mine large data sets. Domain experts will present critical patterns from their problem spaces, and explain their structure and meaning. They will discuss which methods and technologies work well in practice, and which do not. We will consider the expressivity and productivity of different media for defining and transforming patterns, and tools for constructing ontologies and organizing data in meta-patterns. We will hear about high-performance, scalable methods and hardware designs for discovering patterns in petabyte data sets, and tracking their emergence, growth, and decay over time. On the last day, we will do deep dives on two real world problems. The workshop’s goals are to understand the versatility of patterns, define evaluation metrics, and identify technology gaps.
Organizing Committee:
Jim Ang (SNL)
John Feo (PNNL)
Rob Leland (SNL)
Dave Mountain (DoD)
Richard Murphy (Micron Inc)
Steve Pritchard (DoD)
June Weber (DoD)
Patterns are emerging as a powerful medium to analyze petabyte data sets across a wide variety of fields, including social networks, biological systems, financial transactions, economic ecosystems, criminal activities, and cyber security. Patterns can succinctly capture the occurrence of events and the functionality of systems. The pattern to the left depicts the Denial-of-Service Smurf Attack. When annotated with auxiliary information, time stamps, geographical location, and confidence limits, the pattern can capture unfolding activities, evolution of the attack, the functionality and dependencies of component parts, and the uncertainty of outcomes. The pattern can be decomposed into smaller motifs that can be identified and tracked over time to follow an active attack and/or predict next events.
At this second CLSAC workshop, we will discuss using patterns to mine large data sets. Domain experts will present critical patterns from their problem spaces, and explain their structure and meaning. They will discuss which methods and technologies work well in practice, and which do not. We will consider the expressivity and productivity of different media for defining and transforming patterns, and tools for constructing ontologies and organizing data in meta-patterns. We will hear about high-performance, scalable methods and hardware designs for discovering patterns in petabyte data sets, and tracking their emergence, growth, and decay over time. On the last day, we will do deep dives on two real world problems. The workshop’s goals are to understand the versatility of patterns, define evaluation metrics, and identify technology gaps.
Organizing Committee:
Jim Ang (SNL)
John Feo (PNNL)
Rob Leland (SNL)
Dave Mountain (DoD)
Richard Murphy (Micron Inc)
Steve Pritchard (DoD)
June Weber (DoD)

clsac13_agenda.pdf | |
File Size: | 312 kb |
File Type: |