Grid-based Genetic Operators for Graphical Layout Generation
The slides for the PACMHCI / EICS 2021 paper titled "Grid-based Genetic Operators for Graphical Layout Generation" authored by Morteza Shiripour, Niraj Ramesh Dayama, Antti Oulasvirta
For more Details: https://doi.org/10.1145/3461730
Many-objective optimization problem
𝑀is the number of objectives
𝑥 is the feasible set of decision variables
Ω is the set of all boundaries and constraints
3
4.
How to solvethis problem?
● Classical approaches:
○ Weighted-sum and 𝜖-constraint
i. Multiple runs are required
ii. Obtaining the full Pareto front is difficult
iii. Sensitive to the Pareto front’s shape
iv. Linear approximation techniques for some objectives
● Evolutionary approaches:
○ Non-dominated sorting genetic algorithm (NSGA) III
i. Easy to implement
ii. A broad applicability
iii. Robust to dynamic changes
4
Conclusion
● All thesolutions generated satisfy the no-overlap requirement
● Representing well-formed layouts
● The efficiency of the generated layouts was verified by users
Future work
● Exploring new objectives and their effects on performance
● Emerging work that combines machine learning approaches
15