approximation and visualization of interactive decision maps short course of lectures

53
Approximation and Visualization of Interactive Decision Maps Short course of lectures Alexander V. Lotov Dorodnicyn Computing Center of Russian Academy of Sciences and Lomonosov Moscow State University

Upload: ketan

Post on 18-Jan-2016

25 views

Category:

Documents


1 download

DESCRIPTION

Approximation and Visualization of Interactive Decision Maps Short course of lectures. Alexander V. Lotov Dorodnicyn Computing Center of Russian Academy of Sciences and Lomonosov Moscow State University. Lecture 7 . Non-linear Feasible Goals Method and its applications. Plan of the lecture - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Approximation and Visualization of Interactive Decision Maps

Short course of lectures

Alexander V. Lotov

Dorodnicyn Computing Center of Russian Academy of Sciences and

Lomonosov Moscow State University

Page 2: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Lecture 7. Non-linear Feasible Goals Method and its applications

Plan of the lecture1. Approximation for visualization of the feasible objective set2. Application of the FGNL for conceptual design of future

aircrafts3. Identification of economic systems by visualization of Y=f(X)4. Approximation for visualization of the EPH5. Hybrid methods for approximating the EPH in the non-

convex case6. Statistical tests7. A simple hybrid method: combination of two-phase, three-

phase and plastering methods8. Study of a cooling equipment in continuous casting of steel9. Parallel computing

Page 3: Approximation and Visualization of Interactive Decision Maps Short course of lectures

The main problems that arise in the non-linear case:

1. non-convexity of the set Y=f(X); and

2. time-consuming algorithms for global scalar optimization.

Page 4: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Approximation of the feasible objective set f(X) may be needed at least in two cases: 1) decision maker does not want to maximize or minimize the performance indicators; 2) identification problem is considered. We approximate the set f(X) by using simulation of random decisions, filtering their outputs and approximating f(X) by a system of cubes. Then, on-line visualization of the feasible objective set is possible. Thus, we apply simulation-based multi-criteria optimization. Such an approach can be applied in the case of models given by computational modules, too.

Approximation for visualization of the feasible objective set

Page 5: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Example model: the well-known Peak Function

),(),(),(),( 21321221121 xxuxxuxxuxxf 10

,e)x(),x(xu)(xx 212

212

1211 13

, )4

1(10),(

22

215

23

11212

xxexxxxxu

22

2)11(

213 3

1),(

xxexxu

]6 ;5.3[ ],2.3 ;9.4[ 21 xxwhere

Page 6: Approximation and Visualization of Interactive Decision Maps Short course of lectures

which are subject of maximization.Let us consider an example. Imagine that we

want to locate the monitoring station at the point where maximal pollution occurs. Let be the pollution level forecasted by the i-th expert. Different values of criteria characterize difference in knowledge of experts concerning pollution distribution.

),(),( 2121 iii xxfxxf

Let us consider five criterion functions

),( 21 xxf i

Page 7: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Let us consider several three-criterion graphs

Page 8: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 9: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 10: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 11: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 12: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Software demonstration

Page 13: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Application of the FGNL for conceptual design of future aircrafts

Four construction parameters were considered 1. Draught of the engine per weight (P0);Overall drag coefficient of the aircraft (Cx0);Inductive drag coefficient of the aircraft (A);4. Lift coefficient of the aircraft (CY).

The aircraft was described by flight characteristics:1. Rotation speed at a given height (W_max5000);2. Time of elevation to a given height (TimeH);3. Time of acceleration to a given speed (TimeV).

Page 14: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 15: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Exploration of the decision space:Squeezed variety of feasible values of draft of the engine (P0), frontal resistance (СX0) and elevating

force (CY).

Page 16: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Identification of economic systems by visualization of Y=f(X)

Page 17: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Approximation for visualization of the EPH

The EPH is approximated by the set T* that is the union of the non-negative cones

with apexes in a finite number of points y of the set Y=f(X). The set of such points y is called the approximation base and is denoted by T.

Multiple slices of such an approximation can be computed and displayed fairly fast.

mRy

Page 18: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 19: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Visualization example for 8 criteria

Page 20: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 21: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Goal identification

Page 22: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 23: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Demonstration of Pareto Front Viewer

Page 24: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Hybrid methods for approximating the EPH in the non-convex case

Page 25: Approximation and Visualization of Interactive Decision Maps Short course of lectures

The models under studyComputing of the objective functions (the model)

can be given by a computational module (black box) provided by user and unknown for the researcher. Thus, a very broad scope of non-linear models can be studied.

Our methods provide inputs which depend on the method; the module computes outputs of these inputs (or by using simulation, or by solving boundary problem, or in some other way).

Due to it we can even use simulation-based local optimization of random decisions or genetic optimization.

Page 26: Approximation and Visualization of Interactive Decision Maps Short course of lectures

The scheme of the methods

Page 27: Approximation and Visualization of Interactive Decision Maps Short course of lectures

We apply hybrid methods that include:

1) global random search;2) adaptive local optimization;3) importance sampling;4) genetic algorithm.

Statistical tests of the approximation quality play the leading role in approximation process.

Page 28: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Statistical testsQuality of an approximation T* is studied

by using the concept of completeness hT = Pr {f(x)  T* : x X }.

We estimate Pr { hT > h* } for a given reliability by using a random sample

HN = {x1, … ,  xN}. Let hT

(N)= n/N, where n=|f(xi)  T*|. Then, hT

(N) is a non-biased estimate of hT .Moreover, hT

(N) – (– ln (1 – ) / (2N) )1/2

describes the confidence interval.

Page 29: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Completeness function

Let (T*)ε be the ε–neighborhood of T*.

Then, hT (ε)= Pr {f(x)  (T*)ε : x X } is the completeness function.

Important characteristics of the function hT(N)(ε)

is the value εmax=δ(f(HN), T*).

Page 30: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Two optimization-based completeness functions for different iterations (1 and 7)

Page 31: Approximation and Visualization of Interactive Decision Maps Short course of lectures

The optimization-based completenessIn the problems of high dimension of decision variable can happen that the sample completeness is equal to 1,

but the approximation is bad. In this case optimization-based completeness function is used

hT (ε) = Pr{f(Φ(x0)) (T*)ε : x X } where Φ:X → X is the “improvement” mapping, which is usually based on local optimization of a

scalar function of criteria. The mapping moves the point f(Φ(x0)) in the direction of the Pareto frontier. As usually, a random sample HN {x1, … ,  xN} is generated

and hT(N)(ε)=n(ε)/N, where n(ε)=|f(Φ(xi

0))  (T*)ε| is computed.

Page 32: Approximation and Visualization of Interactive Decision Maps Short course of lectures

One-phase methodAn iteration. A current approximation base T must be

given. 1. Testing the base T. Generate a random sample HN X , compute hT

(N)(ε). If hT(N)(ε) (or some

values as hT(N)(0) and εmax=δ(f(HN),T*) in

automatic testing) satisfy the requirements, stop.

2. Forming new base. Form a list that includes points of T and sample points that not belong to T*, exclude dominated points. By this a new approximation base is found. Start next iteration.

Page 33: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Two-phase method

An iteration. A current approximation base T must be given. 1. Testing the base T. Generate a random sample HN X , compute Φ(HN). Construct hT

(N)(ε). If the function hT

(N)(ε) (or some values as hT(N)(0) and

εmax=δ(f(Φ(HN)),T*) in the case of automatic testing) satisfy the requirements, stop.2. Forming new base. As usually. Start next iteration.

Page 34: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Three-phase methodAn iteration. Current base T and a neighborhood B

of decisions which images constitute T must be given 1. Testing the base T. Generate two random samples H1X and H2 B, compute Φ(H1) and Φ(H2). Construct hT

(N)(ε). If hT(N)(ε) satisfies

the requirements, stop.

2. Forming new base. As usually. 3. Forming new neighborhood B using statistics of extreme values. Start next iteration.

Page 35: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Three-phase methodAn iteration. Current base T and a neighborhood B

of decisions which images constitute T must be given 1. Testing the base T. Generate two random samples H1X and H2 B, compute Φ(H1) and Φ(H2). Construct hT

(N)(ε). If hT(N)(ε) satisfies

the requirements, stop.

2. Forming new base. As usually. 3. Forming new neighborhood B using statistics of extreme values. Start next iteration.

Page 36: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Forming new neighborhood BThe neighborhood B is the constituted of the balls in

decision space with centers in current Pareto-optimal decisions. They have the same radius k . The value of the radius is computed using the statistics of extreme values. Namely, we consider the distances of new Pareto-optimal decisions the old Pareto-optimal decisions. Then, we order the distances in accordance to their growth d(N), d(N-l),… where d(N) is the most distanced point. Then, k = d(N) + , where

θ = r(l, χ)( d(N) – d(N-l)), while , is the reliability, 0<<1. Here

r(l, χ) = {[1-(1-χ) 1/l](-1/a) – 1}(-1) and a = (ln l) / ln[(d(N) – d(N-l)) / (d(N) – d(N-1))],

l << N (we took l=10).

Page 37: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Plastering method

“Plastering” method that has some properties of genetic algorithms (as cross-over) is used at the very end of the approximation process.

An iteration. Current approximation base T and numbers q, 1, 2 must be given. 1. Testing the base T. Let H be the set of inputs that result in points of the approximation base T. Select N random pairs (hi,  hj) that satisfy 1 ≤ d(f(hi), f(hj)) ≤ 2 from the set H. Select q random points on the segment connecting the points hi and  hj, and denote them by Hl, l=1,…,N. Compute objective points for the points x Hl, l=1,…,N. Construct hT

(N)(ε). If hT(N)(ε) satisfies the

requirements, stop.2. Forming new base. 3. Filtering if needed

Start next iteration.

Page 38: Approximation and Visualization of Interactive Decision Maps Short course of lectures

A simple hybrid method: combination of two-phase, three-phase and plastering

methods

• Iterations of two-phase method are carried out until hT

(N)(0) and εmax=δ(f(Φ(HN)),T*) are close to zero. • Iterations of three-phase method are carried out

until hT(N)(0) and εmax=δ(f(Φ(HN)), T*) for it are

close to zero. • Iterations the genetic method carried out until hT

(N)

(0) and εmax=δ(f(Φ(HN)), T*) for it satisfy some requirements.

Page 39: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Study of a cooling equipment in continuous casting of steel.

The research was carried out jointly with

Dr. Kaisa Miettinen, Finland,at the University of Jyvaskyla,

Finland.

Page 40: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Cooling in the continuous casting process

Page 41: Approximation and Visualization of Interactive Decision Maps Short course of lectures

CriteriaJ1 is the original single optimization criterion: deviation from the desired surface temperature of the steel strand must be minimized.J2 to J5 are the penalty criteria introduced to describe violation of constraints imposed on : J2 – surface temperature; J3 – gradient of surface temperature along the strand;

J4 – on the temperature after point z3; and J5 – on the temperature at point z5.

J2 to J5 were considered in this study.

Page 42: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Description of the module

FEM/FDM module was developed in Finland,by researchers from University of Jyvaskyla. Properties of the model: 325 control variables that describe intensity of water application.

Properties of local simulation-based optimization:one local optimization required about 11-12 calculations of the gradient and about 1000-2000 additional calculations of the value of f(x).

Page 43: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Next pictures demonstrate the approximation

Page 44: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 45: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 46: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Parallel computing

Page 47: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Parallel computing (processor clusters and grid-computing)

The method has the form that can be used in parallel computing.

Thus, it can be easily implemented at parallel platforms –

it is sufficient to separate data generation and data analysis

(Research in the framework of contract with Russian Federal Agency for Science).

Page 48: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Important property of our hybrid methods

Since our methods are based on random sampling, partial loss of the results is not dangerous. It influences the reliability of the results but does not destroy the process. Due to it, application in GRID network is possible.

Page 49: Approximation and Visualization of Interactive Decision Maps Short course of lectures
Page 50: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Two platform implementation is needed

Page 51: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Example of scenario template for hybrid method

Page 52: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Scenario editor-1

Page 53: Approximation and Visualization of Interactive Decision Maps Short course of lectures

Scenario editor - 2