PowerPoint Presentation
Diagnosis
Design & Creativity
Configuration
Diagnosis
Creativity
Design
Lesson Preview
Defining diagnosis
Data and hypothesis spaces
Mapping data to hypotheses
Two views of diagnosis
Patient:
A: Normal
B: High
C: Low
D: Normal
E: Normal
F: Normal
G: Normal
H: Low
Illnesses:
ο Alphaitis: Elevated A, Reduced C, Elevated F
ο Betatosis: Elevated B, Reduced C, Elevated E, Reduced H
ο Gammanoma: Elevated D, Elevated E, Elevated F
ο Deltacol: Elevated B, Reduced C
ο Epsicusus: Reduced H
ο Zetad: Elevated B, Reduced C, Reduced E, Reduced F
ο Etaemia: Elevated A, Reduced D, Reduced H
ο Thetadesis: Elevated B, Reduced C, Reduced H
ο Iotalgia: Elevated A, Reduced E, Elevated F, Elevated G
ο Kappacide: Reduced A, Reduced F, Reduced G
ο Lambdacrite: Reduced A, Reduced E, Reduced F, Reduced G
ο Mutension: Elevated A, Elevated G
What illness (or set of illnesses) would you use to diagnose this patient?
Diagnosis: To determine what is wrong with a malfunctioning device.
Diagnosis: To determine what is wrong with a malfunctioning device.
Image credit:
Tom Morris, https://commons.wikimedia.org/wiki/User:Tom_Morris
6
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
H10
H11
H12
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
…
DN
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
H10
H11
H12
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
…
DN
Abstract
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
H10
H11
H12
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
…
DN
Abstract
Map
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
H10
H11
H12
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
…
DN
Abstract
Map
Refine
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
H10
H11
H12
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
…
DN
Abstract
Map
Refine
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Problem #1: One data point, multiple hypotheses.
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Problem #2: One hypothesis, multiple sets of data.
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Problem #3: Multiple hypotheses, multiple sets of data.
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Problem #4: Mutually exclusive hypotheses.
But H3 and H6 are mutually exclusive.
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Problem #5: Interacting data points.
But D5 and D9 cancel each other out.
Rule
Cause
Effect
Rule
Cause
Effect
Deduction: Given the rule and the cause, deduce the effect.
Rule
Cause
Effect
Induction: Given a cause and an effect, induce a rule.
Rule
Cause
Effect
Abduction: Given a rule and an effect, abduce a cause.
Patient:
A: Normal
B: High
C: Low
D: Normal
E: Normal
F: Normal
G: Normal
H: Low
Illnesses:
ο Alphaitis: Elevated A, Reduced C, Elevated F
ο Betatosis: Elevated B, Reduced C, Elevated E, Reduced H
ο Gammanoma: Elevated D, Elevated E, Elevated F
ο Deltacol: Elevated B, Reduced C
ο Epsicusus: Reduced H
ο Zetad: Elevated B, Reduced C, Reduced E, Reduced F
ο Etaemia: Elevated A, Reduced D, Reduced H
ο Thetadesis: Elevated B, Reduced C, Reduced H
ο Iotalgia: Elevated A, Reduced E, Elevated F, Elevated G
ο Kappacide: Reduced A, Reduced F, Reduced G
ο Lambdacrite: Reduced A, Reduced E, Reduced F, Reduced G
What illness (or set of illnesses) would you use to diagnose this patient?
Criteria for Choosing Hypotheses
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
1. Hypotheses must cover as much of the data as possible.
Criteria for Choosing Hypotheses
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
1. Hypotheses must cover as much of the data as possible.
2. The smallest number of hypotheses ought to be used.
Criteria for Choosing Hypotheses
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
1. Hypotheses must cover as much of the data as possible.
2. The smallest number of hypotheses ought to be used.
3. Some hypotheses may be more likely than others.
Patient:
A: Normal
B: High
C: Low
D: Normal
E: Normal
F: Low
G: Normal
H: Low
Illnesses:
ο Alphaitis: Elevated A, Reduced C, Elevated F
ο Betatosis: Elevated B, Reduced C, Elevated E, Reduced H
ο Gammanoma: Elevated D, Elevated E, Elevated F
ο Deltacol: Elevated B, Reduced C
ο Epsicusus: Reduced H
ο Zetad: Elevated B, Reduced C, Reduced E, Reduced F
ο Etaemia: Elevated A, Reduced D, Reduced H
ο Thetadesis: Elevated B, Reduced C, Reduced H
ο Iotalgia: Elevated A, Reduced E, Elevated F, Elevated G
ο Kappacide: Reduced A, Reduced F, Reduced G
ο Lambdacrite: Reduced A, Reduced E, Reduced F, Reduced G
ο Mutension: Elevated A, Elevated G
What illness (or set of illnesses) would you use to diagnose this patient?
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Abstract
Map
Refine
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Abstract
Map
Refine
Treatment Space
T1
T2
T3
T4
T5
T6
T7
T8
T9
…
TN
Hypothesis Space
H1
H2
H3
H4
H5
H6
H7
H8
H9
…
HN
Data Space
D1
D2
D3
D4
D5
D6
D7
D8
D9
…
DN
Abstract
Map
Refine
Treatment Space
T1
T2
T3
T4
T5
T6
T7
T8
T9
…
TN
Chair Legs
count : 4
size : 10g
material : metal
cost : $4.00
Chair Seat
size : 100g
material : metal
cost : $10.00
Chair Arms
size : 0g
material : N/A
cost : $0.00
Chair Back
size : 20g
material : metal
cost : $2.00
Chair
mass : 160g
cost : $16
legs :
seat :
arms :
back :
What constraints dictated the design of this chair?
Image credit:
https://commons.wikimedia.org/wiki/File:Chair_4a.jpg
29
Assignment
How would you use diagnosis to design an agent that could answer Raven’s progressive matrices?
To recap…
Defining diagnosis
Process of diagnosis
Diagnosis as classification
Diagnosis as abduction
/docProps/thumbnail.jpeg