Ontologies for Reasoning about Failures in AI Systems Matthew D. Schmill, Tim Oates, Dean Wright University of Maryland Baltimore County Don Perlis, Shomir.

Slides:



Advertisements
Similar presentations
The Teacher Work Sample
Advertisements

Heuristic Search techniques
Level 1 Recall Recall of a fact, information, or procedure. Level 2 Skill/Concept Use information or conceptual knowledge, two or more steps, etc. Level.
Improving System Safety through Agent-Supported User/System Interfaces: Effects of Operator Behavior Model Charles SANTONI & Jean-Marc MERCANTINI (LSIS)
Bridgette Parsons Megan Tarter Eva Millan, Tomasz Loboda, Jose Luis Perez-de-la-Cruz Bayesian Networks for Student Model Engineering.
Chapter 19: Network Management Business Data Communications, 5e.
The Network of Dynamic Learning Communities C 107 F N Increasing Rigor February 5, 2011.
Learning skills for the 21 st century. Learning skills for the 21 st century Being a student in the 21 st century Enlarge image Video.
Display of Information for Time-Critical Decision Making Eric Horvitz Decision Theory Group Microsoft Research Redmond, Washington 98025
Developing Learning Cycles. Insights from Science of Improvement Understand interdependencies in the components of the system where the changes are being.
Learning skills for the 21 st century. Learning skills for the 21 st century Being a student in the 21 st century Enlarge image Video.
1 Reinforcement Learning Introduction & Passive Learning Alan Fern * Based in part on slides by Daniel Weld.
Yiannis Demiris and Anthony Dearden By James Gilbert.
The Comparison of the Software Cost Estimating Methods
Chapter 19: Network Management Business Data Communications, 4e.
1 © 1998 HRL Laboratories, LLC. All Rights Reserved Construction of Bayesian Networks for Diagnostics K. Wojtek Przytula: HRL Laboratories & Don Thompson:
Knowledge Acquisitioning. Definition The transfer and transformation of potential problem solving expertise from some knowledge source to a program.
1 ITC242 – Introduction to Data Communications Week 12 Topic 18 Chapter 19 Network Management.
Why Deep Blue Can’t Play Checkers (and why today’s smart systems aren’t smart) Mike Anderson, Tim Oates, Don Perlis University of Maryland
Experiences with an Architecture for Intelligent Reactive Agents By R. Peter Bonasso, R. James Firby, Erann Gat, David Kortenkamp, David P Miller, Marc.
Constructivism Constructivism — particularly in its "social" forms — suggests that the learner is much more actively involved in a joint enterprise with.
Integrating POMDP and RL for a Two Layer Simulated Robot Architecture Presented by Alp Sardağ.
The Importance of Architecture for Achieving Human-level AI John Laird University of Michigan June 17, th Soar Workshop
MICHAEL T. COX UMIACS, UNIVERSITY OF MARYLAND, COLLEGE PARK Toward an Integrated Metacognitive Architecture Cox – 8 July 2011.
Metacognition Helping students to self-regulate. Definitions  Metacognition - literally “beyond knowing”, knowing what one knows and doesn’t know - promoting.
SUNITA RAI PRINCIPAL KV AJNI
The Metacognitive Loop and the Problem of Brittleness Michael L. Anderson University of Maryland Joint work with Don Perlis, Tim Oates,
Science Inquiry Minds-on Hands-on.
Metareasoning: An Overview of Select Systems Joshua Jones 3/4/2011.
Beverly Park Woolf University of Massachusetts/Amherst U.S.A
Shonda Kuiper Grinnell College April 27 th, 2010.
What should teachers do in order to maximize learning outcomes for their students?
Exploring Design Innovation: The AI Method and Some Results Ashok Goel Georgia Tech May 18, 2006.
© Grant Thornton | | | | | Guidance on Monitoring Internal Control Systems COSO Monitoring Project Update FEI - CFIT Meeting September 25, 2008.
SLB /04/07 Thinking and Communicating “The Spiritual Life is Thinking!” (R.B. Thieme, Jr.)
11 C H A P T E R Artificial Intelligence and Expert Systems.
Computer Science CPSC 322 Lecture 3 AI Applications 1.
T 7.0 Chapter 7: Questioning for Inquiry Chapter 7: Questioning for Inquiry Central concepts:  Questioning stimulates and guides inquiry  Teachers use.
Welcome to AP Biology Mr. Levine Ext. # 2317.
SCANS Skills Competencies &Personal Qualities. What Is SCANS Skills???  Secretary's Commission on Achieving Necessary Skills (SCANS) - appointed by the.
Synthetic Cognitive Agent Situational Awareness Components Sanford T. Freedman and Julie A. Adams Department of Electrical Engineering and Computer Science.
Socialization and Self Esteem
Systems Analysis and Design in a Changing World, Fourth Edition
University of Windsor School of Computer Science Topics in Artificial Intelligence Fall 2008 Sept 11, 2008.
Transfer Like a Champ! By Michelle Brazeal. Transfer Training Why do we teach?
Advanced Decision Architectures Collaborative Technology Alliance An Interactive Decision Support Architecture for Visualizing Robust Solutions in High-Risk.
Curiosity-Driven Exploration with Planning Trajectories Tyler Streeter PhD Student, Human Computer Interaction Iowa State University
Project-Based Learning (PBL) Vivene Robinson.
Effective Teaching Strategies Day 2
How People Learn – Brain, Mind, Experience, and School (Bransford, Brown, & Cocking, 1999) Three core principles 1: If their (students) initial understanding.
Learning and Acting with Bayes Nets Chapter 20.. Page 2 === A Network and a Training Data.
© 2013 UNIVERSITY OF PITTSBURGH Supporting Rigorous Mathematics Teaching and Learning Engaging In and Analyzing Teaching and Learning Tennessee Department.
Performance Task Overview Introduction This training module answers the following questions: –What is a performance task? –What is a Classroom Activity?
Artificial Intelligence: Research and Collaborative Possibilities a presentation by: Dr. Ernest L. McDuffie, Assistant Professor Department of Computer.
OBJECTIVE 1.01: UNDERSTAND INDIVIDUAL CHARACTERISTICS/ TRAITS, INTERESTS/PREFERENCES, ABILITY LEVELS, SKILL ACQUISITION, TALENTS/APTITUDES, LEARNING STYLES.
Learning for Physically Diverse Robot Teams Robot Teams - Chapter 7 CS8803 Autonomous Multi-Robot Systems 10/3/02.
UCLA Graduate School of Education & Information Studies National Center for Research on Evaluation, Standards, and Student Testing V4, 1/18/07 Research.
#1 Make sense of problems and persevere in solving them How would you describe the problem in your own words? How would you describe what you are trying.
Kansas State University Department of Computing and Information Sciences CIS 830: Advanced Topics in Artificial Intelligence Wednesday, January 26, 2000.
Investigate Plan Design Create Evaluate (Test it to objective evaluation at each stage of the design cycle) state – describe - explain the problem some.
Tier III Preparing for First Meeting. Making the Decision  When making the decision to move to Tier III, all those involve with the implementation of.
Chapter 19: Network Management
Using Cognitive Science To Inform Instructional Design
Enabling Team Supervisory Control for Teams of Unmanned Vehicles
Grade 6 Outdoor School Program Curriculum Map
Announcements Homework 3 due today (grace period through Friday)
Lecturette 1: Leveraging Change through Strategic Planning
Lecturette 1: Leveraging Change through Strategic Planning
Mike Anderson, Darsana P. Josyula, Don Perlis
Why Deep Blue Can’t Play Checkers
Presentation transcript:

Ontologies for Reasoning about Failures in AI Systems Matthew D. Schmill, Tim Oates, Dean Wright University of Maryland Baltimore County Don Perlis, Shomir Wilson, Scott Fults University of Maryland Michael Anderson Franklin & Marshall College Darsana Josyula Bowie State University

Brittleness Brittleness is the propensity of an agent to perform poorly or fail outright in the face of unanticipated changes

People: not very brittle

(this guy is juggling chainsaws in a ring of fire)

AI Systems: maybe just a bit on the brittle side the complexities of real-world environments are difficult to account for in advance organization and integration of multiple, varied cognitive components a challenging task

Failures and Self-Ignorance

Perturbation Tolerant Systems A perturbation is any unanticipated change, either in the world or in the system itself, that impacts an agent’s performance. Perturbation tolerance is the ability of a system to quickly recover from perturbations. How can we endow AI systems with human-like perturbation tolerance?

Intuition Based on observations in human problem solving Generic formula for perturbation tolerance: 1.notice something is different 2.assess the situation 3.decide how to –React –Adapt

The MetaCognitive Loop An architecture for perturbation tolerance Allows a system to declare expectations MCL continuously monitors expectations and notices when they are violated assesses the cause of the violation guides the host system to an appropriate response

Prior Work MCL as a tightly coupled system component –human-computer dialog (ALFRED) –reinforcement learning (Chippy) –game playing (Bolo) MCL in these systems –had specific knowledge of the host system (domain) sufficient to properly respond to anomalies

Current Work Proof-of-concept work involved domain- specific instantiations of MCL The benefits of adding a metacognitive loop must outweigh the cost of incorporating it Current work is toward domain-neutrality –a single MCL that can be integrated with a variety of systems at a low cost

Domain Neutrality The roads to recovery in different domains share concepts at some level of abstraction –indications – contextual signal of an anomaly “a sensor failed to change as expected” –failures – underlying cause of indications “the sensor is malfunctioning” –responses – actions required to recover from and prevent anomaly “revise models to use alternate sensors”

Domain Neutral MCL Indications, Failures, and Responses ontologies nodes represent concepts at many levels of abstraction expressing various relationships between concepts implemented as graphical models Note, Assess, and Guide steps use the ontologies ontologies are now Bayes networks concepts have associated beliefs indicating the belief that they are true in the context of the current anomaly

Domain Neutral MCL Move from concrete indications to abstract Reason about underlying failures at an abstract, domain-neutral level Move from abstract repairs to concrete ones that can be implemented by the host expectations indicationsfailuresresponses actionable

MCL Overview expectations host indicationsfailuresresponses MCL actionable initialize: the host declares its sensing, acting, and cognitive capabilities to MCL specifications

Declaring Expectations expectations host indications step 1: when the host decides to act, it declares its expectations about what will happen failuresresponses MCL concrete action: move-to expectation: at-completion, location = N W expectation: distance-to-goal decreases expectation: action completes in < 2 minutes

Monitoring expectations host indications step 2: as the action unfolds, MCL monitors the state of the expectations failuresresponses MCL concrete monitor

Violation expectations host indications step 3: the agent encounters some ice, which slows its progress, violating an expectation failuresresponses MCL concrete action: move-to expectation: at-completion, location = N W expectation: distance-to-goal decreases expectation: action completes in < 2 minutes

Violation expectations host indications step 3: the agent encounters some ice, which slows its progress, violating an expectation failuresresponses MCL concrete action: move-to expectation: at-completion, location = N W expectation: distance-to-goal decreases expectation: action completes in < 2 minutes (this is ice)

Indication expectations host indications step 4: the properties of the expectation and how it is violated are used to create an initial configuration of the indications ontology failuresresponses MCL concrete

Indication Ontology Violation: Duration < 2mins Violation Type: miss/unchanged Violation Type: long of target Violation Type: short of target Source Type: sensor Source Type: temporal Data Type: continuous Source Type: reward Violation Type: CWA Violation Violation Type: missed target Violation Type: divergence Indication: deadline missed (actual ontology currently has 50+ nodes)

Inference: Failures expectations host indications step 5: the connectivity between the indications ontology and the failure ontology allows MCL to hypothesize the underlying failure failuresresponses MCL concrete

Failure Ontology Indication: deadline missed failure: resource error failure: knowledge error failure: eff. malfunction failure: effector error failure: effector noice failure: resource surfeit failure: resource defecit failure: model error failure: sensor error failure: predictive m.e. failure: procedural m.e. (from indication ontology) (actual ontology currently has 25+ nodes)

Inference: Responses expectations host indications step 6: the connectivity between the failure ontology and the response ontology allows MCL to generate beliefs that a particular response will fix the anomaly failuresresponses MCL concrete

Response Ontology failure: eff. malfunction failure: predictive m.e. failure: procedural m.e. (from failure ontology) response: rebuild model response: modify model response: amend model concrete response: revise expectations concrete response: rerun m.g.a. concrete response: reset policy concrete response: set  concrete response: eff. diagnostic (actual ontology will have many nodes)

Response Ontology failure: eff. malfunction failure: predictive m.e. failure: procedural m.e. (from failure ontology) response: rebuild model response: modify model response: amend model concrete response: revise expectations concrete response: rerun m.g.a. concrete response: reset policy concrete response: set  concrete response: eff. diagnostic (only those nodes actionable by the host will be active)

Response Generation expectations host indications step 6: MCL computes the utility associated with each concrete response available to the host and selects the highest utility response failuresresponses MCL concrete response: perform effector diagnostic addresses: effector malfunction

Feedback expectations host indications step 7: the host implements the response. if the response fails, MCL treats the feedback as evidence against it in the underlying Bayes nets. failuresresponses MCL concrete feedback from response

Interactive Repair expectations host indications MCL and the host iterate over responses until one is found that prevents the anomaly from occurring again. failuresresponses MCL concrete feedback from response highest utility response

Current State Trimmed-down ontologies implemented with simple Bayes inference Deployed in testbed applications Transferring to openPNL-based Bayes Net Redeploying PNL-based dnMCL Reinforcement learning Bolo player Dialog agent

Conclusion Lots of evidence that a meta-level monitor and control can make AI systems more robust, more efficient Anderson, Perlis et al. Goel, Stroulia, Murdock, et al. Our intuition Concepts used in reasoning about anomalies generalize across domains Encode these concepts into ontologies, use Bayesian techniques to endow AI systems to reason about and recover from their own failures

Future Work Deploy dnMCL on new domains Expand ontologies Learn expectations Recursive MCL –Expectations about repairs & failures Evaluation methods

Mahalo nui loa

Reinforcement Learning Chippy is a reinforcement learner who learns an action policy in a reward-yielding grid world domain. He maintains expectations for rewards average reward average time between rewards If his experience deviates from his expectations (due to changing the reward schedule) he assesses the anomaly and chooses from a range of responses increase learning rate increase exploration rate start learning from scratch

Comparison of the per-turn performance of non-MCL and simple-MCL with a perturbation moving the locations and degrees of rewards in turn 10,001.