1 / 7

Pattern Matching & Plan Generation: Application of Distributed GA’s

Pattern Matching & Plan Generation: Application of Distributed GA’s. Darrin Taylor, Sc.D. 21 st Century Technologies 11675 Jollyville Rd, #300, Austin TX, 78759 dtaylorz@21technologies.com. Example of template for panel members – “hot” topics. Title of topic: AOC Plan generation

soo
Download Presentation

Pattern Matching & Plan Generation: Application of Distributed GA’s

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Pattern Matching & Plan Generation:Application of Distributed GA’s Darrin Taylor, Sc.D. 21st Century Technologies 11675 Jollyville Rd, #300, Austin TX, 78759 dtaylorz@21technologies.com

  2. Example of template for panel members – “hot” topics • Title of topic: AOC Plan generation • Objective: Reduction of AOC Staffing Requirements. • Key Challenges: • Human Challenges • Reducing manpower • Gaining confidence in Automation • Our requirements are changing more rapidly. • Operational Challenges • Faster planning • Comprehensive Planning • Moving Decision making Lower • Technical Challenges • Combinatorial Optimization • Pattern Matching

  3. 21st Century Technologies ATO-Link Master Air Attack Planner Assigns planes and munitions to targets. Creates Missions Creates Packages Maximizes Resource utilization Maximizes PD metric Minimizes Fuel Consumption Generates notional Stick Route Accommodates Threat information Accommodates Mid Air Refueling Produces reports providing confidence in results. Managing of multiple plans TMODS Darpa - EELD Finds instances of Terrorist activity in large amounts of data Data represented as a graph Terrorist activity represented as a graph template Sub graph isomorphism Find instances of the template in the larger graph.

  4. Human Challenges ATO-Link Reducing Manpower Provide the Chief of Combat plans with Target Assignments The amount of data processed per person rises dramatically. Gaining confidence In Automation Provide metadata providing the reasoning behind the results Provide metrics for the results and compare them to results for similar processing Requirements Change. New requirements, and doctrine TMODS Reducing Manpower Provide the Analyst with potential avenues for further investigation The amount of data processed per person rises dramatically. Gaining confidence In Automation The templates provide the reasoning for finding a match Provide metrics (Graph Edit Distance) for the results to determine. Requirements Change Social Network Analysis provides a host of potential relevant metrics to use

  5. ATO-Link Operational Challenges Faster planning Compressing the Kill Chain Comprehensive Planning Making better use of existing resources. Planes, munition, fuel,airspace control Include traceability to EBO objectives Target Weaponeering Moving Decision Making Lower Planning is typically top down. There is a feedback loop by which the plan changes. UAV’s have flexibility in their COA Pilots may make plans and coordinate changes while in flight. Metrics- Measure the right thing Don’t maximize the number of targets hit. Endeavor to send all of the planes up fully loaded and return empty. Compute Smarter Don’t compute a solution when recognizing a problem that we already have the answer to will do. Future Research Automating the feed back of Battle Damage Assessment as much as possible.

  6. TMODS – Operational Challenges Faster Analysis Arose out of our Rome Labs IMPACT effort Link Analysis “Connecting the dots” Comprehensive Analysis Finding some instances may not be enough. Sometime you have to find all instances Great Application to ATO-Link Pattern matching technologies Can be used to ask the question:Have I seen this problem before? Metrics- Measure the right thing Currently using Graph Edit Distance Compute Smarter Don’t get stuck on one way to solve the problem. Our “merging matches” algorithm efficiently finds all of the matches Genetic Algorithm finds only some of the matches. Good at partial matches as well. Future Research Examine how SNA impacts link analysis technologies.

  7. Technical Challenges Optimization Time is an input into the planning process. Requirements Change: There are algorithms that are optimized for various sub-problems that are necessary in the planning process. Need for technologies robust in the face of changes in requirements. Pattern Matching & CBR Much of the computational investment takes place in advance of when it is needed. Key challenge is to recognize the problem when your confronted with it. The problem has high dimensionality Genetic Algorithms Currently: Planning is an optimization problem. Though we don’t have the time to get the “optimal plan”, and if we did , it would not survive execution anyway. Technology is robust in the face of requirements changes Future: Apply EELD Technology to ATO-Link When the planning time is scarce we use GA to do high dimensionality graph pattern matching against our plan representation. Based on our Darpa Link Analysis research.

More Related