Cognitive Information Processing (CIP) Shell
The Cognitive Information Processing Shell is a state-of-the-art intelligent system that is the foundational technology to fuse disparate data streams (text and video). It can find rare events and complicated patterns in task-critical real-time applications. The system was created by Dr. Supratik and Dr. Iyengar, and is also used and modified by other researchers.
The CIM Shell fuses disparate data streams like texts and videos to implement an interactive inspection and visualization system, which can be used to conduct real-time monitoring, analysis, and online diagnosis for industrial applications. Many areas such as manufacturing, agricultural, and oil production spaces could be redeployed remotely without consideration of critical events and to fix failures.
The system has been widely reported in many media outlets such as TechTarget[1], The Advocate[2] and Business Report[3].
CIP Shell Framework
The CIM shell is a CEP system that can analyze complex events and activities and adapt rapidly to evolving situations in a wide variety of environments. The system can find and react to complex patterns by storing and cross-referencing historical events. The system is nearly real-time by adjusting the parameters to adapt to environmental changes.
The system starts with two stages: define the goal, such as locating potential oil leaks; provide an initial set of trend-detection. Figure 1 shows that the system has 6 steps: Deploy sensors, Set rules, System execution, Check success, Update parameters and Human supervision. The 6 steps are run iteratively and continually improve the performance and accuracy.
Figure 1 is the overall framework of the CIM shell. The system can be used in myriad scenarios, such as soil and water management and video control in the presence of frame losses. The system has a capability of resolving 100,000 events and making a million inferences per second by implementing on a GPU-based cloud infrastructure.
CIM Shell can also be deployed in a distributed system by using a distributed storage schema and middleware, and the system's consistency can be ensured by a distributed agreement algorithm such as the Paxos algorithm. The agents can fuse disparate streaming data such as text and video to provide a consistent basis for decision-making. An interactive and intelligent sensing, inspection, and visualization system provides real-time monitoring and analysis.
There are three components of the system:
- Using a distributed storage provides persistent notification of events in the uncertain group communication environment. The storage system keeps the events as key-value pairs and uses Paxos (computer science) algorithm to ensure agreement.
- Crawlers running 24/7 on the cloud infrastructure for video and other types of data, and the related algorithm is a variant of Global Virtual Time algorithm which enables uniform interpretation of temporal events over a wide-area network.
- Distributed intelligent agents are generated manually or automatically and are used to fuse disparate streaming data such as text, video, and provide good decision-making ability. For example, one could use Secure Operations Language (SOL) [4] to generate a system that meets the requirements.
CEP systems can reason about causality, knowledge, belief, risk, and uncertainty and are able to make decisions under incomplete information. CIM Shell can discriminate effective seed information and modify the conditions as a human operator expects.
Applications
CIM Shell could have prevented an oil spill like that in the Gulf of Mexico. The system had been integrated with the Deepwater Horizon infrastructure; it would have interpreted prior maintenance incidents and concerns voiced by BP's engineers as a complex event and magnified the associated risks. It could advise operators remotely to close the rig for maintenance or even act autonomously [5].
The Online Diagnosis of Manufacturing Machines (ODMM) system, developed by SpotCheck Inc., licensed by AIlectric, and used by Novatec and ProphecySensorLytics to diagnose system failures in factories, with a Total Addressable Market (TAM) of $9.1B in 2014, which is predicted by ABIResearch to grow to over $24.7B by 2019; [6]
The DeepSAT Video and Image Analytics application, developed by LSU and used by the NASA Ames Research Center to provide decision support for satellite missions, with a TAM of over $41B; [6]
The DeepDrug system to shorten the developmental timeline of new drugs, developed by SynthLab, and ranked one of the top 10 systems worldwide competing for $3M in the IBM Watson AI XPrize competition.[7]
Impact and Market
The CIM Shell has been applied in many different areas and has great potential in more areas. Many institutions and companies have introduced the system and related algorithms, under the permission of AutoPredictiveCoding, LLC. According to Spotcheck, the Total Addressable Market for the technology is estimated to be over $24.7 billion by 2019 [8]
Highlighted applications[6]
- Foundational technology for Spotcheck Technology's online Diagnosis of Manufacturing Machines
- Technology incorporated into products sold by Novattec, Prophecy, SensorLytics, and others
- Used by NASA Ames Research for satellite imagery and decision support via NVidia's DeepSAT Video Analytics & Image program
- Used by USDA for decision support via the National Agricultural Imagery Project
References
- ↑ "Cognitive computing for all? Think about it". SearchBusinessAnalytics. Retrieved 2018-05-06.
- ↑ [email protected], TED GRIGGS |. "Baton Rouge team among Top 10 competing in $5 million IBM Watson AI XPRIZE contest". The Advocate. Retrieved 2018-05-06.
- ↑ "Baton Rouge software development team advances in IBM competition for $5M - Baton Rouge Business Report". Baton Rouge Business Report. 2017-12-06. Retrieved 2018-05-06.
- ↑ Bharadwaj, Ramesh; Mukhopadhyay, Supratik (2008). "A Formal Approach to Developing Reliable Event-Driven Service-Oriented Systems". 2008 32nd Annual IEEE International Computer Software and Applications Conference. IEEE. doi:10.1109/compsac.2008.87. ISBN 9780769532622.
- ↑ Iyengar, S.S.; Mukhopadhyay, Supratik; Steinmuller, Christopher; Li, Xin (2010-08). "Preventing Future Oil Spills with Software-Based Event Detection". Computer. 43 (8): 95–97. doi:10.1109/mc.2010.235. ISSN 0018-9162. Check date values in:
|date=(help) - ↑ 6.0 6.1 6.2 "Dr. S. S. Iyengar". people.cis.fiu.edu. Retrieved 2018-05-06.
- ↑ Advocate, The. "Lsu Scientists Develop New Efficiency Software | News | Communications of the ACM". cacm.acm.org. Retrieved 2018-05-06.
- ↑ "Maintenance Analytics to Generate $24.7 Billion in 2019, Driven by Predictive Maintenance and Internet of Things". www.abiresearch.com. Retrieved 2018-05-06.
This article "Cognitive Information Processing (CIP) Shell" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Cognitive Information Processing (CIP) Shell. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
