You are on page 1of 37

AI and Design

Krzysztof Gajos
CS 182, Fall 2011
Hall of Fame?
Hall of Shame?

Eisenberg, Bryan. How to Decrease Sales by 90 Percent. ClickZ. [Online] Feb 21, 2003. http://www.clickz.com/showPage.html?page=1588161
Kohavi et al. Practical guide to controlled experiments on the web: listen to your customers not to the hippo. Proceedings (2007) pp. 959-967
What Is Design?
Design by Genius

Specification
Design by Exploration

[Buxton, Sketching User Experiences]


Design as Optimization
Variables and Domains
Classroom:
!: { , , }

Light Bank: A/V:


!: { , , } !: { , }

Light: Projector:
Light ... Light ...
!: { , } !: { , }

Light Level: Power: Screen: Power: Input: Vent:


!: <int, [0,10]> !: bool !: bool ! : bool !: <string, {data1,data2, video}> !: <int, [0,3]>
Branch and Bound
• Search in any order you want
• Use an admissible heuristic to
compute the minimum cost of the solution
reachable from the current state; if this cost
is higher than the best solution found so
far, prune this branch!
• Orthogonal to any CSP techniques you
might want to use!
Adaptation to Screen Size
Adaptation to Devices
Demo
Algorithm Performance
20,000
Worst case
number of nodes

15,000
expanded

10,000

5,000

Average case
0
size constraint
Parallel Algorithm

• No change in average case performance


• Worst-case performance improved from 54
minutes to slightly over 1 minute
Average- Worst-
Possible
Application case time case time
Interfaces
(ms) (ms)
Map 2.16E+02 70 142
Test interface A 3.51E+02 133 783
Test Interface B 2.84E+04 45 837
Email 7.74E+05 28 343
Classroom 7.80E+07 35 516
Test Interface C 1.87E+08 144 2,666
Ribbon 5.44E+08 317 3,096
Synthetic 1.27E+11 51 441
Print Dialog 3.54E+13 591 30,307
Font Formatting 2.76E+15 355 23,171
Stereo 2.79E+17 30 66,150
Vs. 8.9 million years
Concerns in UI Design

• Perceptual effort
• Cognitive effort
• Motor effort
• Aesthetics
Adapting to Motor Abilities

cost( ) = time
Collect Motor
Performance Data
Pointing Dragging List Selection
Generate User Interfaces that
minimize the expected time
the person would take to
complete a set of common tasks
Results
One Participant’s
Story
• Severely impaired hand and
arm control:
• Trackball controlled with chin
• Keyboard controlled with a head-mounted wand
• Severely impaired speech so not possible to use
speech recognition
• Has a CS degree and works as an IT
consultant
One Participant’s
Story
• Severely impaired hand and
arm control:
With Supple he closed half the
• Trackball controlled with chin
performance gap between himself
• andcontrolled
Keyboard able-bodied participants
with a head-mounted wand
• Severely impaired speech so not possible to use
speech recognition
• Has a CS degree and works as an IT
consultant
Models of Motor Performance
From Unobtrusive OBservations
Mouse pointer movements Mouse pointer movements Movements recorded in situ
recorded in a formal experiment recorded in situ and then filtered
#!!!!" #!!!!" #!!!!"

+!!!" +!!!" +!!!"

*!!!" *!!!" *!!!"

)!!!" )!!!" )!!!"


Movement Time (ms)

(!!!" (!!!" (!!!"

'!!!" '!!!" '!!!"

&!!!" &!!!" &!!!"

%!!!" %!!!" %!!!"

$!!!" $!!!" $!!!"

#!!!" #!!!" #!!!"

!" !" !"


#" $" %" &" '" (" #" $" %" &" '" (" #" $" %" &" '" ("
Index of Difficulty Index of Difficulty Index of Difficulty
Three Things to
Remember

• Design == constrained optimization


• Branch and Bound
• Phase Transition in combinatorial problems

You might also like