Selection Without Search

wrennie
Note by , created over 6 years ago

Note on Selection Without Search, created by wrennie on 04/22/2013.

870
0
0
Tags No tags specified
wrennie
Created by wrennie over 6 years ago
Health and Social Care Flashcards
Kelsey Phillips
GCSE History – The early years and the Weimar Republic 1918-1923
Ben C
4 Lesson Planning Tips for Teachers
Micheal Heffernan
Teaching Using GoConqr's Tools
Micheal Heffernan
Biology - B2 - AQA - GCSE - Exam Style Questions
Josh Anderson
Realidad De Nuestra Identidad Cultural
53831
An Inspector calls - Gerald Croft
Rattan Bhorjee
B3, C3, P3
George Moores
Physical Geography
clongworth25
P1 quiz
I M Wilson

Page 1

Naive model with the aim of replicating the behaviour of the person studied by Anzai and Simon (1979) on her first attempt at the five-disk problem

At each stage in the solution process: enumerate the possible moves;evaluate those moves with respect to local information;select the move with the highest evaluation;apply the selected move to the current state;if the goal state has not been achieved, repeat the process.This approach can be applied to any well-specified problem (Newell and Simon 1972).

Modelling this in Cogent requires: a buffer to hold the current state a buffer to hold the representation of operators a process to manipulate buffer contents

Anzai and Simon (1979) found that on the subject's first attempt she avoids backtracking and moving the same disk twice she never moves the small disk back to the peg it was on two moves previously These need to be incorporated into the evaluation function

Add a further propositional buffer, Previous Move. When a move is executed, store the move in this buffer. Use the buffer contents to give low evaluations (e.g., 0) to moves that move the same disk twice in succession

If one adopts the 2nd constraint, all moves are determined by the initial move.

Model needs to give move in which the small disk is returned to the peg it was on two moves previously receive low evaluations. Rename Previous Move to Previous Moves and set its capacity to 2. OnExccess is Oldest.Represent the previous move in PreviousMoves by a proposition of the form history(Move, -1) and the move prior to that by a proposition of the form history(Move, -2)Need to update indices each time a move is made.

Properties of selection without search: selection of the first move is random if the model selects the wrong first move, it can go off in an unproductive region of the problem space the model will find a solution eventually, but it can be very inefficient

Nevertheless, Anzai and Simon (1979) found subjects seemingly use this strategy first

Anzai and Simon's (1979) participant did make the wrong first move on her first attempt at the problem, but she quickly realised her mistake. After six moves she began expressing doubts, and after nine moves she abandoned her attempt and began afresh. On her second attempt, she explicitly avoided making the same first move, and subsequently solved the problem. 

Strategy

Implementation in Cogent

Operator Selection and Operation Application

New Evaluation Function

Properties