4dc3ed26a29c9c3df3ec373524377a5b-AuthorFeedback.pdf
–Neural Information Processing Systems
Asdiscussed44 in Section 6, a stricter condition is the "win-stay lose-shift" model, which is easier to validate without running the45 teaching algorithm. It remains an open problem to understand the complexity of the general sequential teaching setting as a52 sequentialoptimizationproblem.
Neural Information Processing Systems
Feb-12-2026, 03:10:32 GMT