Transcript Document

Gheorghe Tecuci, Mihai Boicu, Dorin Marcu,
Bogdan Stanescu, Cristina Boicu, Marcel Barbulescu
Learning Agents Center
George Mason University
Computer Science Department Partners Day Symposium
May 4, 2004
Overview
Research Problem, Approach, and Application
Knowledge Representation, Reasoning, and Learning
Experiments of Agent Development and Use
Long Term Research Vision
Acknowledgements
How are agents built and why it is hard
Intelligent Agent
Domain
Expert
Knowledge
Engineer
Inference Engine
Dialog
Programming
Knowledge Base
Results
The knowledge engineer attempts to understand how the
subject matter expert reasons and solves problems and then
encodes the acquired expertise into the agent's knowledge
base.
This modeling and representation of expert’s knowledge is long,
painful and inefficient (known as the “knowledge acquisition
bottleneck”).
Research Problem and Approach
Research Problem: Elaborate a theory, methodology and family of
systems for the development of knowledge-base agents by subject
matter experts, with limited assistance from knowledge engineers.
Approach: Develop a learning agent that can be taught directly by
a subject matter expert while solving problems in cooperation.
1. Mixed-initiative
problem solving
The expert teaches
the agent how to perform
various tasks in a way that
resembles how the expert
would teach a person.
3. Multistrategy
learning
Interface
2. Teaching and
learning
The agent learns
from the expert,
building, verifying
and improving its
knowledge base
Problem
Solving
Learning
Ontology
+ Rules
Synergistic collaboration and transition to the USAWC
George Mason University - US Army War College
Students
319jw Case Studies in
developed
Center of Gravity Analysis scenarios
Identify the strategic COG candidates for the Sicily_1943 scenario
Which is an opposing force in the Sicily_1943 scenario?
Anglo_allies_1943
589jw Military Applications
of Artificial Intelligence
Identify the strategic COG candidates for Anglo_allies_1943
Is Anglo_allies_1943 a single member force or a multi-member force?
Anglo_allies_1943 is a multi-member force
Identify the strategic COG candidates for the Anglo_allies_1943 which is a multi-member force
What type of strategic COG candidates should I consider for a multi-member force?
I consider the candidates corresponding to the multi-member nature of the force
Identify the strategic COG candidates corresponding to the multi-member nature of the Anglo_allies_1943
What type of strategic COG candidates should I consider for the multi-member nature of the force?
I consider the relationships between the members of the force
I consider the type of operations being conducted by the members of the force
Identify the strategic COG candidates with respect to the type of operations being conducted by the members of the Anglo_allies_1943
Which is the primary force element that will conduct the campaign for Anglo_allies_1943?
Allied_forces_operations_Husky
Identify the strategic COG candidates with respect to the type of operations being conducted by Allied_forces_operations_Husky
Is Allied_forces_operations_Husky made up of a true single group or are there subgroups?
Allied_forces_operations_Husky is made up of several subgroups
Formalization of
the Center of Gravity
(COG) analysis process
Disciple
Artificial
Intelligence
Research
Knowledge bases and agent
development by subject matter
experts, using learning agent
technology. Experiments in the
USAWC courses.
Students
developed
agents
Use of Disciple in a
sequence of two joint
warfighting courses
Sample Domain: Center of Gravity Analysis
Centers of Gravity: Primary sources of moral or physical strength,
power or resistance of the opposing forces in a conflict.
Application to current war scenarios (e.g. War on terror, Iraq) with state and
non-state actors (e.g. Al Qaeda).
Identify COG candidates
Identify potential primary
sources of moral or physical
strength, power and
resistance from:
Government
Military
Test COG candidates
Test each identified COG
candidate to determine whether
it has all the necessary critical
capabilities:
Which are the critical
capabilities?
People
Are the critical requirements of
these capabilities satisfied?
Economy
If not, eliminate the candidate.
Alliances
If yes, do these capabilities
have any vulnerability?
Etc.
Problem Solving Approach: Task Reduction
A complex problem solving task is performed by:
T1
Q 1 S1
• successively reducing it to simpler tasks;
• finding the solutions of the simplest tasks;
A11 S11
…
T11a S11a T11bS11b
…
Q11b S11b
A1n S
1n
T1n
• successively composing these solutions until
the solution to the initial task is obtained.
…Test whether President Roosevelt is
President Roosevelt is a strategic COG
candidate that can be eliminated
a viable strategic COG candidate
Which are the critical capabilities that President Roosevelt should have to be a COG candidate?
A11b1 S11b1… A11bm S11bm
T11b1
T11bm
Knowledge Base
Object Ontology
Reduction Rules
Composition Rules
Does President Roosevelt have all
the necessary critical capabilities?
The necessary critical capabilities are: be protected, stay informed, communicate,
be influential, be a driving force, have support and be irreplaceable
Test whether President
Roosevelt has the critical
capability to be protected
President Roosevelt has the critical capability to be protected. President Roosevelt is
protected by US Service 1943 which has no significant vulnerability
Test whether President
Roosevelt has the critical
capability to stay informed
President Roosevelt has the critical capability to stay informed. President Roosevelt
receives essential intelligence from intelligence agencies which have no significant
vulnerability
Test whether President
Roosevelt has the critical
capability to communicate
President Roosevelt has the critical capability to communicate through executive orders,
through military orders, and through the Mass Media of US 1943. These communication
means have no significant vulnerabilities
Test whether President
Roosevelt has the critical
capability to be influential
President Roosevelt has the critical capability to be influential because he is the head of the
government of US 1943, the commander in chief of the military of US 1943, and is a trusted
leader who can use the Mass Media of US 1943. These influence means have no
significant vulnerabilities.
Test whether President
Roosevelt has the critical
capability to be a driving force
President Roosevelt has the critical capability to be a driving force. The main reason for
President Roosevelt to pursue the goal of unconditional surrender of European Axis is
“preventing separate peace by the members of the Allied Forces”. Also, “the western
democratic values” provides President Roosevelt with determination to persevere in this
goal. There is no significant vulnerability in the reason and determination.
Test whether President
Roosevelt has the critical
capability to have support
President Roosevelt has the critical capability to have support because he is the head of a
democratic government with a history of good decisions, a trusted commander in chief of
the military, and the people are willing to make sacrifices for unconditional surrender of
European Axis. The means to secure continuous support have no significant vulnerability.
President Roosevelt does not have the critical capability to be irreplaceable. US 1943 would
No.
Problem Solving and Learning
force
We need to
Identify and test a strategic COG candidate
corresponding to a member of the Allied_Forces_1943
Which is a member of Allied_Forces_1943?
multi member force
multi group force
single member force
multi state force
single group force
...
US_1943
Therefore we need to
Identify and test a strategic
COG candidate for US_1943
EXAMPLE OF
REASONING
STEP
...
multi state alliance
multi state coalition
...
dominant partner
multi state alliance
equal partners
multi state alliance
Allied Forces 1943
LEARNED RULE
IF
Identify and test a strategic COG candidate
corresponding to a member of the ?O1
Question
Which is a member of ?O1 ?
Answer
?O2
INFORMAL STRUCTURE
THEN
Identify and test a strategic COG candidate
for ?O2
single state force
ONTOLOGY
FRAGMENT
has as member
US 1943
IF
Identify and test a strategic COG candidate corresponding
to a member of a force
FORMAL STRUCTURE
The force is ?O1
Plausible Upper Bound Condition
?O1
is multi_member_force
has_as_member ?O2
?O2
is force
Plausible Lower Bound Condition
?O1
is equal_partners_multi_state_alliance
has_as_member ?O2
?O2
is single_state_force
THEN
Identify and test a strategic COG candidate for a force
The force is ?O2
Use of Disciple at the US Army War College
319jw Case Studies in Center of Gravity Analysis
Disciple helps the students to perform a center
Disciple was taught based on the expertise of
of gravity analysis of an assigned war scenario.
Prof. Comello in center of gravity analysis.
Teaching
Learning
Disciple
Agent KB
Problem
solving
Global evaluations of Disciple by officers from the Spring 03 course
Strongly
Agree
Agree
Neutral
Disagree
8 Disciple should be used in
7 future versions of this course
6
5
4
3
2
1
0
Strongly
Disagree
Strongly
Agree
Agree
Neutral
Disagree
8Disciple helped me to learn to
7 perform a strategic COG
6 analysis of a scenario
5
4
3
2
1
0
Strongly
Disagree
Strongly
Agree
Agree
Neutral
Disagree
Strongly
Disagree
8 The use of Disciple is an
7
assignment
that is well suited to
6the course's learning objectives
5
4
3
2
1
0
Use of Disciple at the US Army War College
589jw Military Applications of Artificial Intelligence course
Students teach
Disciple their COG
analysis expertise,
using sample
scenarios (Iraq 2003,
War on terror 2003,
Arab-Israeli 1973)
Students test
the trained
Disciple agent
based on a
new scenario
(North Korea
2003)
Global evaluations of Disciple by officers during three experiments
I think that a subject matter expert can use Disciple to build an agent,
with limited assistance from a knowledge engineer
Strongly
Agree
Agree
Neutral
Spring 2003
COG testing based on
critical capabilities
Disagree
9
8
7
6
5
4
3
2
1
0
Strongly
Disagree
Strongly
Agree
Agree
Neutral
Spring 2002
COG identification
and testing
Disagree
9
8
7
6
5
4
3
2
1
0
Strongly
Disagree
Strongly
Agree
Agree
Neutral
Disagree
Spring 2001
COG identification
Strongly
Disagree
9
8
7
6
5
4
3
2
1
0
Parallel development and merging of knowledge bases
Initial KB
Domain analysis and ontology
development (KE+SME)
Parallel KB development
(SME assisted by KE)
DISCIPLE-COG
Team 1
stay informed
be irreplaceable
5 features
10 tasks
10 rules
Knowledge
Engineer (KE)
All subject matter
experts (SME)
37 acquired concepts and
Extended KB features for COG testing
DISCIPLE-COG
Team 2
communicate
14 tasks
14 rules
DISCIPLE-COG
Team 3
be influential
2 features
19 tasks
19 rules
KB merging (KE)
Unified 2 features
Deleted 4 rules
Refined 12 rules
Final KB:
+9 features  478 concepts and features
+105 tasks 134 tasks
+95 rules 113 rules
Correctness = 98.15%
432 concepts and features, 29 tasks, 18 rules
For COG identification for leaders
Training scenarios:
Iraq 2003
Arab-Israeli 1973
War on Terror 2003
DISCIPLE-COG
Team 4
have support
35 tasks
33 rules
DISCIPLE-COG
Team 5
be protected
be driving force
3 features
24 tasks
23 rules
Learned features, tasks, rules
Integrated KB
DISCIPLE-COG
5h 28min average training time / team
3.53 average rule learning rate / team
COG identification and testing (leaders)
Testing scenario:
North Korea 2003
Other Disciple agents
Disciple-WA (1997-1998): Estimates the
best plan of working around damage to a
transportation infrastructure, such as a
damaged bridge or road.
Demonstrated that a knowledge engineer
can use Disciple to rapidly build and update
a knowledge base capturing knowledge from
military engineering manuals and a set of
sample solutions provided by a subject
matter expert.
Disciple-COA (1998-1999): Identifies
strengths and weaknesses in a Course of
Action, based on the principles of war and
the tenets of army operations.
Mission:
BLUE-BRIGADE2 attacks to penetrate RED-MECH-REGIMENT2 at 130600 Aug in order to enable the completion of seize
OBJ-SLAM by BLUE-ARMOR-BRIGADE1.
Close:
BLUE-TASK-FORCE1, a balanced task force (MAIN EFFORT) attacks to penetrate RED-MECH-COMPANY4, then clears
RED-TANK-COMPANY2 in order to enable the completion of seize OBJ-SLAM by BLUE-ARMOR-BRIGADE1.
BLUE-TASK-FORCE2, a balanced task force (SUPPORTING EFFORT 1) attacks to fix RED-MECH-COMPANY1 and REDMECH-COMPANY2 and RED-MECH-COMPANY3 in order to prevent RED-MECH-COMPANY1 and RED-MECHCOMPANY2 and RED-MECH-COMPANY3 from interfering with conducts of the MAIN-EFFORT1, then clears REDMECH-COMPANY1 and RED-MECH-COMPANY2 and RED-MECH-COMPANY3 and RED-TANK-COMPANY1.
…
Reserve:
The reserve, BLUE-MECH-COMPANY8, a mechanized infantry company, follows Main Effort, and is prepared to reinforce )
MAIN-EFFORT1.
Security:
SUPPORTING-EFFORT1 destroys RED-CSOP1 prior to begin moving across PL-AMBER by MAIN-EFFORT1 in order to
prevent RED-MECH-REGIMENT2 from observing MAIN-EFFORT1.
…
Deep:
Deep operations will destroy RED-TANK-COMPANY1 and RED-TANK-COMPANY2 and RED-TANK-COMPANY3.
Rear:
BLUE-MECH-PLT1, a mechanized infantry platoon secures the brigade support area.
Demonstrated the generality of its learning
methods that used an object ontology created
by another group (TFS/Cycorp).
Demonstrated that a knowledge engineer and a
subject matter expert can jointly teach Disciple.
Fires:
Fires will suppress RED-MECH-COMPANY1 and RED-MECH-COMPANY2 and RED-MECH-COMPANY3 and REDMECH-COMPANY4 and RED-MECH-COMPANY5 and RED-MECH-COMPANY6.
End State: At the conclusion of this operation, BLUE-BRIGADE2 will enable accomplishing conducts forward passage of lines through
BLUE-BRIGADE2 by BLUE-ARMOR-BRIGADE1.
MAIN-EFFORT1 will complete to clear RED-MECH-COMPANY4 and RED-TANK-COMPANY2.
SUPPORTING-EFFORT1 will complete to clear RED-MECH-COMPANY1 and RED-MECH-COMPANY2 and RED-MECHCOMPANY3 and RED-TANK-COMPANY1.
SUPPORGING-EFFORT2 will complete to clear RED-MECH-COMPANY5 and RED-MECH-COMPANY6 and RED-TANKCOMPANY3.
Disciple’s vision on the future of software development
Mainframe
Computers
Personal
Computers
Learning
Agents
Software systems
developed and used by
persons that are not
computer experts
Software systems developed
by computer experts
and used by persons that
are not computer experts
Software systems
developed and used
by computer experts
Vision on the use of Disciple in Education
teaches
Disciple
Agent KB
teaches
Disciple
Agent KB
teaches
…
The expert/teacher teaches Disciple
through examples and explanations,
in a way that is similar to how the
expert would teach a student.
Disciple
Agent KB
Disciple
Agent KB
teaches
Disciple tutors the student in a
way that is similar to how the
expert/teacher has taught it.
Acknowledgements
This research was sponsored by the Defense Advanced
Research Projects Agency, Air Force Research
Laboratory, Air Force Material Command, USAF under
agreement number F30602-00-2-0546, by the Air Force
Office of Scientific Research under grant number
F49620-00-1-0072 and by the US Army War College.