QUALITATIVE MODELLING AND REASONING Ivan Bratko Faculty of Computer and Information Science Ljubljana University Slovenia.

Slides:



Advertisements
Similar presentations
Model-based Technology George M. Coghill. Introduction Description of the Field Motivations for development Applications of MBT Overview Models Background.
Advertisements

Formal Computational Skills
Qualitative Reasoning About Population and Community Ecology Reha K. Gerçeker Boğaziçi University, 2005.
Boyce/DiPrima 9th ed, Ch 2.4: Differences Between Linear and Nonlinear Equations Elementary Differential Equations and Boundary Value Problems, 9th edition,
Signals and Circuits 2 Time responses The nature of the time response at the electrical system is the same as at the mechanical system. As example in the.
Sahar Mosleh PageCalifornia State University San Marcos 1 Introductory Concepts This section of the course introduces the concept of digital circuits and.
Chapter 4: Basic Properties of Feedback
INTRODUCTION TO MODELING
2. the intensity of light bulb A increases.
Markov processes in a problem of the Caspian sea level forecasting Mikhail V. Bolgov Water Problem Institute of Russian Academy of Sciences.
Start of Presentation Mathematical Modeling of Physical Systems © Prof. Dr. François E. Cellier Electrical CircuitsI This lecture discusses the mathematical.
Blogics! It’s a logic circuit simulator aimed at beginners. It introduces simple concepts in the design of interactive physical computing systems such.
Description of Systems M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl 1.
EECE Hybrid and Embedded Systems: Computation T. John Koo, Ph.D. Institute for Software Integrated Systems Department of Electrical Engineering and.
Universidad de La Habana Lectures 5 & 6 : Difference Equations Kurt Helmes 22 nd September - 2nd October, 2008.
Automata theory and formal languages Andrej Bogdanov The Chinese University of Hong Kong Fall 2008.
UNIVERSITY OF SOUTH CAROLINA Department of Computer Science and Engineering CSCE 580 Artificial Intelligence Ch.6 [P]: Reasoning Under Uncertainty Sections.
Unit A2.1 Causality Kenneth D. Forbus Qualitative Reasoning Group Northwestern University.
Feedback Control Systems (FCS)
RC (Resistor-Capacitor) Circuits
Electrical Electricity Fuses Relays Solenoids Electrical Testing
INEL 3105 Electrical Systems Analysis Electrical and Computer Engineering Department University of Puerto Rico at Mayagüez.
Chapter 19 Linear Programming McGraw-Hill/Irwin
FIRST ORDER TRANSIENT CIRCUITS
09/16/2010© 2010 NTUST Today Course overview and information.
Time Response of Reactive Circuits
Qualitative Induction Dorian Šuc and Ivan Bratko Artificial Intelligence Laboratory Faculty of Computer and Information Science University of Ljubljana,
MATHEMATICAL FOUNDATIONS OF QUALITATIVE REASONING Louise-Travé-Massuyès, Liliana Ironi, Philippe Dague Presented by Nuri Taşdemir.
1 An Introduction to Formal Languages and Automata Provided by : Babak Salimi webAdd:
1 Unit 1: Automata Theory and Formal Languages Readings 1, 2.2, 2.3.
Copyright © 2013, 2009, 2005 Pearson Education, Inc. 1 2 Graphs and Functions Copyright © 2013, 2009, 2005 Pearson Education, Inc.
Linear Programming McGraw-Hill/Irwin Copyright © 2012 by The McGraw-Hill Companies, Inc. All rights reserved.
Fundamentals of Electric Circuits Chapter 16 Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
PROLOG SYNTAX AND MEANING Ivan Bratko University of Ljubljana Faculty of Computer and Info. Sc. Ljubljana, Slovenia.
A METHODOLOGY FOR QUALITATIVE MODELLING OF ECONOMIC, FINANCIAL SYSTEMS WITH EMPHASIS TO CHAOS THEORY Tomas Vicha, Mirko Dohnal Faculty of Business and.
Chapter 1 Introduction n Introduction: Problem Solving and Decision Making n Quantitative Analysis and Decision Making n Quantitative Analysis n Model.
October 2004Computer Hardware Lecture 5 Slide1 / 29 Lecture 5 Physical Realisation of Logic Gates.
Barnett/Ziegler/Byleen Business Calculus 11e1 Chapter 13 Review Important Terms, Symbols, Concepts 13.1 Antiderivatives and Indefinite Integrals A function.
Copyright © 2010 Pearson Education, Inc. All rights reserved Sec
Week 3 Capacitance & Inductance Transitions Try your hand at graphing these functions Graph: y = e x Graph: y = e -x Graph: y = 1 - e -x.
Functions and Their Representations
EE102 – SYSTEMS & SIGNALS Fall Quarter, Instructor: Fernando Paganini.
Logic UNIT 1.
Properties of Functions. First derivative test. 1.Differentiate 2.Set derivative equal to zero 3.Use nature table to determine the behaviour of.
Copyright © 2006 Pearson Education, Inc. Publishing as Pearson Addison-Wesley.
Linear Programming McGraw-Hill/Irwin Copyright © 2012 by The McGraw-Hill Companies, Inc. All rights reserved.
Theory of Computation Automata Theory Dr. Ayman Srour.
 A transistor is the basic building block of electronic components.  The average computer may have millions of them within its circuits.  Essentially,
Introduction to Modeling Technology Enhanced Inquiry Based Science Education.
Circuits Electric Current Series vs. Parallel. Let’s Review 0 What is electricity?
Modelling & Simulation of Semiconductor Devices Lecture 1 & 2 Introduction to Modelling & Simulation.
Lesson 54 - FTC PART 2 Calculus - Santowski 7/6/2016Calculus - Santowski1.
CEN352 Dr. Nassim Ammour King Saud University
G. Pullaiah College of Engineering and Technology
Introduction to Physical Science
Description and Analysis of Systems
Linear Inequalities and Absolute Value
Autonomous Cyber-Physical Systems: Dynamical Systems
§1—2 State-Variable Description The concept of state
Natural Sciences and Technology Grade 6
CPSC 121: Models of Computation
Basic Concepts of Algorithm
Circuit Components.
CPSC 121: Models of Computation
P2 - Physics Circuits.
Formal Languages and Automata Theory BODDEDA HARITHA LAKSHMI,
This work is licensed under a Creative Commons Attribution 4
Chapter 5 Integration Section R Review.
Automata theory and formal languages COS 3112 – AUTOMATA THEORY PRELIM PERIOD WEEK 1 AND 2.
Presentation transcript:

QUALITATIVE MODELLING AND REASONING Ivan Bratko Faculty of Computer and Information Science Ljubljana University Slovenia

CONTENTS  Common sense, qualitative reasoning and naive physics  Why qualitative reasoning and modelling  Qualitative reasoning about static systems  Qualitative reasoning about dynamic systems  Qualitative differential equations (QDEs) and QSIM

QUALITATIVE REASONING, NAIVE PHYSICS  Describe physical processes qualitatively, without numbers or exact numerical relations  Close to common sense descriptions

EXAMPLE: BATH TUB  What will happen? Amount A of water will keep increasing, so will level L, until the level reaches the top.

EXAMPLE: U-TUBE  What will happen? La Lb

EXAMPLE: U-TUBE  What will happen? La Lb Level La will be decreasing, and Lb increasing, until La = Lb.

QUANTITATIVE MODELLING  Traditional modelling and simulation:  Differential equations  Numerical methods  Arithmetic computation  World represented by numbers  But: Representing world is only a special case of describing the world

QUALITATIVE REASONING  Alternative representations, reasoning  Common sense reasoning  Naive physics, as opposed to "proper physics"  Qualitative modelling, as opposed to quantitative modelling

QUALITATIVE ABSTRACTIONS OF QUANTITATIVE INFORMATION Quantitative statement Qualitative statement Level( 3.2 sec) = 2.6 cm Level(t1) = pos Level( 3.2 sec) = 2.6 cm Level(t1) = zero..top Level(3.2 sec) = 0.12 m/sec Level(t1) increasing Amount = Level * (Level + 5.7) M + ( Amount, Level)

QUALITATIVE ABSTRACTIONS  numbers --> symbolic values and intervals  time derivatives --> directions of change  functions --> monotonic relations  monotonic sequence of values in time --> one symbolic value + direction of change

WHY QUALITATIVE MODELLING?  Exact relations may not be known, e.g. in physiology  Numerical parameters may be hard to measure  Quantitative model may be computationally complex  Qualitative models facilitate:  Explanation – how system works?  Diagnostic reasoning for fault diagnosis  Interpreting alarms in process control  Functional reasoning  Structural synthesis, invention from first principles

SOME APPROACHES TO QUALITATIVE REASONING AND MODELLING  Qualitative Physics Theory, QPT; notion of "process" (Forbus 1984)  Qualitative simulation, QSIM; abstraction of differential equations into qualitative differential equations (QDE) (Kuipers 1986)  Envisionment; based on "confluences", a kind of QDE (de Kleer and Brown, 1984)  KARDIO, qualitative model of the heart; symbolic descriptions in logic (Bratko, Mozetic, Lavrac 1989)

QUALITATIVE REASONING ABOUT STATIC SYSTEMS  Circuit 1 BulbVolt SwVolt

CIRCUIT 2 S1 B1 S2 B2 B3 S3 A diagnostic question about circuit 1: If the switch on and the bulb dark, what is state of the bulb? A diagnostic question about circuit 2: Seeing that bulb 2 is light and bulb 3 is dark, can we conclude that bulb 3 is blown?

Abstracting real number X into qualitative value  if X > 0 then "pos"  if X = 0 then "zero"  if X < 0 then "neg"

QUALITATIVE SUMMATION % qsum( Q1, Q2, Q3): % Q3 = Q1 + Q2, qualitative sum over domain [pos,zero,neg] qsum( pos, pos, pos). qsum( pos, zero, pos). qsum( pos, neg, pos). qsum( pos, neg, zero). qsum( pos, neg, neg). qsum( zero, pos, pos)....

ELECTRIC COMPONENTS % Definition of switch % switch( SwitchPosition, Voltage, Current) switch( on, zero, AnyCurrent). % Switch on: zero voltage switch( off, AnyVoltage, zero). % Switch off: zero current % Definition of bulb % bulb( BulbState, Lightness, Voltage, Current) bulb( blown, dark, AnyVoltage, zero). bulb( ok, light, pos, pos). bulb( ok, light, neg, neg). bulb( ok, dark, zero, zero).

DEFINITION OF CIRCUIT 1 % A simple circuit consisting of a bulb, switch and battery circuit1( SwitchPos, BulbState, Lightness) :- switch( SwitchPos, SwVolt, Curr), bulb( BulbState, Lightness, BulbVolt, Curr), qsum( SwVolt, BulbVolt, pos). % Battery voltage = pos

CIRCUIT 2 circuit2( Sw1, Sw2, Sw3, B1, B2, B3, L1, L2, L3) :- switch( Sw1, VSw1, C1), bulb( B1, L1, VB1, C1), switch( Sw2, VSw2, C2), bulb( B2, L2, VB2, C2), qsum( VSw2, VB2, V3), switch( Sw3, V3, CSw3), bulb( B3, L3, V3, CB3), qsum( VSw1, VB1, V1), qsum( V1, V3, pos), qsum( CSw3, CB3, C3), qsum( C2, C3, C1).

SOME QUESTIONS  A diagnostic question: ?- circuit2( _, _, off, B1, B2, B3, light, _, dark). B1 = ok, B2 = ok, B3 = blown  A control-type question: ?- circuit2( SwPos1, SwPos2, SwPos3, ok, ok, ok, _, _, light). SwPos1 = on, SwPos2 = on, SwPOs3 = off; Swpos1 = on, SwPos2 = off, SwPos3 = off

QUALITATIVE REASONING ABOUT DYNAMIC SYSTEMS  Bath tub with open drain and constant input flow.

WATER LEVEL IN TIME Level top steady inc. zero Time  Qualitatively Time Level t0 zero / inc (t0, t1) zero..top / inc t1 zero..top / std

RELATION BETWEEN AMOUNT AND LEVEL  The greater the amount, the greater the level A = M + (L)  A is a monotonically increasing function of L

MONOTONIC FUNCTIONS Y = M + (X) specifies a family of functions X Y

QUALITATIVE CONSTRAINTS  M + (X,Y) Y is a monotonically increasing function of X  M - (X,Y) Y is a monotonically decreasing function of X  sum(X,Y,Z) Z = X+Y  minus(X,Y) Y = -X  mult(X,Y,Z) Z = X*Y  deriv(X,Y) Y = dX/dt (Y is time derivative of X)

QUALITATIVE MODEL OF BATH TUB Quantities:  Level = level of water  Amount = amount of water  Inflow = input flow  Outflow = output flow  Netflow = net flow (Netflow = Inflow – Outflow)

QUALITATIVE MODEL OF BATH TUB Constraints:  M 0 + ( Amount, Level)  M 0 + ( Level, Outflow)  sum( Outflow, Netflow, Inflow)  deriv( Amount, Netflow)  Inflow = constant = inflow/std

GRAPHICAL REPRESENTATION OF BATH-TUB MODEL deriv Amount Netflow M Inflow = inflow/std Level Outflow M 0 +

QUALITATIVE SIMULATION - EXAMPLE deriv Inlow = IF/std M 0 + M 0 + M Time Amount Level Pressure Outflow Netflow to 0

QUALITATIVE STATE TRANSITIONS Main principle: Assume all variables are continuous and smooth in time.

CONTINUITY AND SMOOTHNESS  Therefore in single transition step: a variable can either stay unchanged or change between neighbour landmarks; e.g.: Landmark1 ---> Landmark1.. Landmark2 Landmark1.. Landmark2 ---> Landmark2  variable's direction of change can stay unchanged, or change to its neighbour value; e.g.: dec ---> std std ---> inc However, dec ---> inc is impossible.

TABLE OF POSSIBLE STATE TRANSITIONS Let L1 and L2 be two successive landmarks. Possible state transitions are:  L1 / std ---> L1 / std  L1 / std ---> L1.. L2 / inc  L1.. L2 / inc ---> L1.. L2 / std  L1.. L2 / inc ---> L2 / inc  L1.. L2 / inc ---> L2 / std  etc. Variable cannot jump over a landmark.

SKETCH OF SIMULATION ALGORITHM Given: A model M as a set of constraints, and initial state S0 Initialise current state S := S0 While transition possible to next state do begin Find a next state NextS (using state transition table) such that NextS satisfies the constraints in model M; S := NextS end Note: This algortihm non-deterministically generates possible behaviours.

SPURIOUS BEHAVIOURS QSIM algorithm has the following mathematical properties:  1. QSIM is complete: it is guaranteed to generate all possible qualitative behaviours of a simulated system.  2. QSIM is not sound: it may generate qualitative behaviours that the simulated system cannot exhibit. These are called spurious behaviours. Unfortunately this deficiency seems to be very difficult to fix.

SPURIOUS SOLUTIONS Q3, spurious! QDE Q1 Q2 DE1 S1 DE2 S2 DE3 S3