Problem instance variability strongly affects FIC/DK-P solvability¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2024 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? For the same dataset configuration (C = 400, K = 400, S = 10), different random-seed problem instances yield different fitness outcomes, indicating difficulty depends on both course catalog and student objective. → Lebis 2024

Evidence¶

Lebis 2024¶

Lebis, A., Humeau, J., Fleury, A., Lucas, F., & Vermeulen, M. (2024). Fully Individualized Curriculum with Decaying Knowledge, a New Hard Problem: Investigation and Recommendations. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-023-00376-9

q2 · i? · design · r2

Dependency study across 10 problem instances from the same configuration (C = 400, K = 400, S = 10, P = 4), each with 30 GA runs, presented in Fig. 5. Results confirm "solving FIC/DK-P seems to be strongly dependent on the problem instance."

"solving FIC/DK-P seems to be strongly dependent on the problem instance. For the same data-set configuration, C = 400, K = 400, S = 10, we produce 10 different problem instances by changing the random seed and then performing for each one 30 different runs."

Discussion¶

Related Claims¶

• FIC/DK-P exhibits combinatorial explosion making exact solving intractable beyond small problem sizes — related
• Increasing the number of courses per semester does not necessarily make FIC/DK-P easier to solve well — related
• A genetic algorithm can find near-faultless curricula for full-scale FIC/DK-P instances despite intractability of exact methods — related
• Meeting the requisites of the student objective is the hardest constraint to satisfy in FIC/DK-P — related