Document Design with Interactive Evolution Juan C. Quiroz ,  Amit Banerjee, Sushil J. Louis, and  Sergiu Dascalu Departmen...
Outline <ul><li>Motivation </li></ul><ul><li>Interactive Genetic Algorithms </li></ul><ul><li>Related Work </li></ul><ul><...
Motivation <ul><li>Design process </li></ul><ul><ul><li>Conceptual design </li></ul></ul><ul><ul><li>Detailed design </li>...
Conceptual Design <ul><li>Subjective evaluation of alternative design concepts </li></ul><ul><ul><li>Aesthetics and other ...
Our Goal <ul><li>Support a simple design task </li></ul><ul><ul><li>Create a brochure document </li></ul></ul><ul><li>Obje...
<ul><li>Background and Related Work </li></ul>
Genetic Algorithms <ul><li>Population based search technique </li></ul><ul><ul><li>Natural selection </li></ul></ul><ul><u...
Interactive Genetic Algorithms <ul><li>Human guided evolution </li></ul><ul><li>Evaluation of subjective criteria </li></u...
Related Work <ul><li>Evolution of album page layouts (Geigel and Loui, 2003) </li></ul><ul><li>Album pages are encoded usi...
<ul><li>Interactive Document Design </li></ul>
Evolutionary Document Design <ul><li>Shapes: rectangles, ellipses, rounded rectangles </li></ul><ul><li>Shape scaling, alo...
Fitness Evaluation <ul><li>Objective heuristics </li></ul><ul><ul><li>White space evaluation </li></ul></ul><ul><ul><li>De...
Subjective Evaluation
Supported Customization <ul><li>Moving of shapes </li></ul><ul><li>Scaling of shapes </li></ul><ul><li>Insertion </li></ul...
Evolution of Brochures Generation  0 Generation  10
Pretest Experiment <ul><li>Three participants </li></ul><ul><li>Set of requirements: </li></ul><ul><ul><li>Design a brochu...
Preliminary Sample Brochures
User Generated Brochures
Discussion <ul><li>All three users were able to create brochures that met the requirements </li></ul><ul><li>Users liked t...
Discussion <ul><li>Interaction with IGA found to be limiting </li></ul><ul><ul><li>Document close to desired, but users no...
Discussion <ul><li>The ability to view and assess many documents in a few minutes </li></ul><ul><li>Useful when requiremen...
Affordance of IGAs <ul><li>The designer must make “appropriate actions perceptible and inappropriate ones invisible.” – Th...
Affordance of IGAs <ul><li>Most IGA applications s have conceptual models targeted to expert users </li></ul><ul><li>Conce...
Affordance Issues in Brochure Evolver <ul><li>Picking the best document </li></ul><ul><ul><li>Limiting </li></ul></ul><ul>...
Future Work <ul><li>User studies </li></ul><ul><ul><li>Creating brochure from scratch versus using IGA </li></ul></ul><ul>...
Conclusions <ul><li>Approach to document design based on human guided evolution </li></ul><ul><ul><li>In pretest all users...
Questions? <ul><li>Juan Quiroz </li></ul><ul><li>[email_address] </li></ul><ul><li>www.cse.unr.edu/~quiroz </li></ul>
Upcoming SlideShare
Loading in …5
×

KES IIMSS 2009: Document Design with Interactive Evolution

683 views

Published on

KES IIMSS 2009 presentation on the use of interactive genetic algorithms for document design. A pretest with three participants is presented with a discussion of affordance issues of interactive evolution.

Published in: Design, Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
683
On SlideShare
0
From Embeds
0
Number of Embeds
4
Actions
Shares
0
Downloads
5
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • What I’ll be covering today. First I’ll explain the motivation behind our work, why it is interesting and why this audience should care. We propose a computational model of creativity, we’ll see the potential for the model to be creative by looking at design space exploration. Then to test the model, we implemented a framework called IGAP, interactive genetic algorithm peer to peer. We conducted an experiment and evaluation using IGAP for floorplan design, with some interesting results. So the first question we must ask in a talk like this is, what is creativity?
  • The design process typically consists of these 4 phases: conceptual design, detailed design, evaluation, and iterative redesign. Computers have been used extensively for all these stages of design, except the creative conceptual design phase. So why is this the case?
  • Alternative design concepts during this design phase may need to be subjectively evaluated, especially when requirements include aesthetics and other subjective criteria. So how do designers evaluate subjective criteria? What’s the formula, or equation that we can code into an algorithm? It is very difficult if not impossible to do so. Finally, we are also interested in collaborative design, computer supported collaborative design. It has been argued that much of our intelligence and creativity results from interaction and collaboration with peers. We can question whether that if it had been the blue man instead of the blue man group, if the man would have been able to come up with the same creative stuff they do. So how do we propose to tackle these challenges?
  • KES IIMSS 2009: Document Design with Interactive Evolution

    1. 1. Document Design with Interactive Evolution Juan C. Quiroz , Amit Banerjee, Sushil J. Louis, and Sergiu Dascalu Department of Computer Science & Engineering University of Nevada, Reno
    2. 2. Outline <ul><li>Motivation </li></ul><ul><li>Interactive Genetic Algorithms </li></ul><ul><li>Related Work </li></ul><ul><li>Evolutionary Document Design </li></ul><ul><li>Results </li></ul><ul><li>Affordance of Interactive Genetic Algorithms </li></ul><ul><li>Future Work </li></ul>
    3. 3. Motivation <ul><li>Design process </li></ul><ul><ul><li>Conceptual design </li></ul></ul><ul><ul><li>Detailed design </li></ul></ul><ul><ul><li>Evaluation </li></ul></ul><ul><ul><li>Iterative redesign </li></ul></ul>
    4. 4. Conceptual Design <ul><li>Subjective evaluation of alternative design concepts </li></ul><ul><ul><li>Aesthetics and other subjective criteria </li></ul></ul><ul><li>What is the formula for how designers evaluate subjective criteria? </li></ul>
    5. 5. Our Goal <ul><li>Support a simple design task </li></ul><ul><ul><li>Create a brochure document </li></ul></ul><ul><li>Objective requirements </li></ul><ul><li>Subjective exploration of designs </li></ul><ul><li>Interactive Genetic Algorithms for exploring brochure document designs </li></ul>
    6. 6. <ul><li>Background and Related Work </li></ul>
    7. 7. Genetic Algorithms <ul><li>Population based search technique </li></ul><ul><ul><li>Natural selection </li></ul></ul><ul><ul><li>Survival of the fittest </li></ul></ul>
    8. 8. Interactive Genetic Algorithms <ul><li>Human guided evolution </li></ul><ul><li>Evaluation of subjective criteria </li></ul><ul><ul><li>Aesthetics </li></ul></ul><ul><ul><li>Emotion </li></ul></ul><ul><ul><li>Intuition </li></ul></ul>
    9. 9. Related Work <ul><li>Evolution of album page layouts (Geigel and Loui, 2003) </li></ul><ul><li>Album pages are encoded using a 4-tuple for each image in the page: </li></ul><ul><ul><li>X, Y, Scaling, Rotation </li></ul></ul><ul><li>User specifies set of preferences at start of GA run </li></ul><ul><ul><li>Preference values help guide the evolutionary process </li></ul></ul><ul><li>With IGA, user interacts every generation </li></ul>
    10. 10. <ul><li>Interactive Document Design </li></ul>
    11. 11. Evolutionary Document Design <ul><li>Shapes: rectangles, ellipses, rounded rectangles </li></ul><ul><li>Shape scaling, along x and y axis, by up to 10% </li></ul><ul><ul><li>Scaling up or down </li></ul></ul><ul><li>Shapes can serve as placeholders for content </li></ul><ul><li>Collections of shapes can serve as a background design </li></ul>
    12. 12. Fitness Evaluation <ul><li>Objective heuristics </li></ul><ul><ul><li>White space evaluation </li></ul></ul><ul><ul><li>Degree of shape overlap </li></ul></ul><ul><ul><li>Spatial balance </li></ul></ul><ul><li>Subjective heuristics </li></ul><ul><ul><li>Small subset displayed from large population </li></ul></ul><ul><ul><li>User evaluation by picking the solution the user likes the best from the subset </li></ul></ul><ul><ul><li>Fitness interpolation of the rest of population based on similarity to user selected best </li></ul></ul>
    13. 13. Subjective Evaluation
    14. 14. Supported Customization <ul><li>Moving of shapes </li></ul><ul><li>Scaling of shapes </li></ul><ul><li>Insertion </li></ul><ul><ul><li>Images, Text boxes </li></ul></ul><ul><li>Deletion </li></ul><ul><ul><li>Shapes, Text boxes </li></ul></ul><ul><li>Color </li></ul><ul><ul><li>Change color scheme </li></ul></ul><ul><ul><li>Change individual shape color </li></ul></ul><ul><li>Save to file </li></ul>
    15. 15. Evolution of Brochures Generation 0 Generation 10
    16. 16. Pretest Experiment <ul><li>Three participants </li></ul><ul><li>Set of requirements: </li></ul><ul><ul><li>Design a brochure that includes: </li></ul></ul><ul><ul><ul><li>A header </li></ul></ul></ul><ul><ul><ul><li>At least one paragraph </li></ul></ul></ul><ul><ul><ul><li>At least two images </li></ul></ul></ul><ul><ul><li>Brochure to advertise minor in digital interactive games </li></ul></ul>
    17. 17. Preliminary Sample Brochures
    18. 18. User Generated Brochures
    19. 19. Discussion <ul><li>All three users were able to create brochures that met the requirements </li></ul><ul><li>Users liked the ability to explore alternative designs </li></ul><ul><li>Interaction with IGA found to be limiting </li></ul>
    20. 20. Discussion <ul><li>Interaction with IGA found to be limiting </li></ul><ul><ul><li>Document close to desired, but users not being able to fine tune the solution by simple picking </li></ul></ul><ul><ul><li>No support for injecting edits made by user back to population </li></ul></ul>
    21. 21. Discussion <ul><li>The ability to view and assess many documents in a few minutes </li></ul><ul><li>Useful when requirements are open-ended </li></ul><ul><ul><li>When user has to create a conceptual model for the given requirements </li></ul></ul>
    22. 22. Affordance of IGAs <ul><li>The designer must make “appropriate actions perceptible and inappropriate ones invisible.” – The Design of Everyday Things </li></ul><ul><li>Typical IGA experience </li></ul><ul><ul><li>Maximum of 20 generations </li></ul></ul><ul><ul><li>User fatigue </li></ul></ul><ul><ul><li>Frustration </li></ul></ul><ul><ul><li>Boredom </li></ul></ul>
    23. 23. Affordance of IGAs <ul><li>Most IGA applications s have conceptual models targeted to expert users </li></ul><ul><li>Conceptual model understood only by author researchers </li></ul>
    24. 24. Affordance Issues in Brochure Evolver <ul><li>Picking the best document </li></ul><ul><ul><li>Limiting </li></ul></ul><ul><ul><li>Reduces user fatigue </li></ul></ul><ul><li>Building a conceptual model of the IGA </li></ul><ul><ul><li>End-user should not need an understanding of GAs to use system </li></ul></ul><ul><li>Robustness of evolutionary system </li></ul><ul><ul><li>Is the system working properly? </li></ul></ul><ul><ul><li>Is the system performing poorly because it is a hard problem? </li></ul></ul>
    25. 25. Future Work <ul><li>User studies </li></ul><ul><ul><li>Creating brochure from scratch versus using IGA </li></ul></ul><ul><ul><li>Difference in quality in designed brochures </li></ul></ul><ul><ul><li>Creative Product Semantic Scale as evaluation criteria </li></ul></ul><ul><li>Allow exploration starting with a prototype brochure </li></ul>
    26. 26. Conclusions <ul><li>Approach to document design based on human guided evolution </li></ul><ul><ul><li>In pretest all users were able to generate diverse brochures that met the given requirements </li></ul></ul><ul><li>Limitations in IGA and proposed tool due to affordance issues </li></ul>
    27. 27. Questions? <ul><li>Juan Quiroz </li></ul><ul><li>[email_address] </li></ul><ul><li>www.cse.unr.edu/~quiroz </li></ul>

    ×