New theory results for machine teaching appearing at #NeurIPS2019!
We characterize spaced repetition teaching under the half-life regression (HLR) forgetting model.
https://arxiv.org/abs/1805.08322
HLR is very popular, also used by orgs like Duolingo:
https://making.duolingo.com/how-we-learn-how-you-learn …
(1/5)
-
Prikaži ovu nit
-
Spaced repetition is well-understood for teaching single concepts, but much less so for multiple concepts. We develop a generalization of string submodularity to characterize total concept recall over time over multiple concepts, each with a different HLR rate. (2/5)
1 reply 0 proslijeđenih tweetova 3 korisnika označavaju da im se sviđaPrikaži ovu nit -
Use these new theoretical tools, we can analyze the performance of the standard adaptive greedy algorithm for spaced repetition (over multiple concepts). We provide sufficient conditions where the adaptive greedy algorithm is guaranteed to perform well. (3/5)
1 reply 0 proslijeđenih tweetova 4 korisnika označavaju da im se sviđaPrikaži ovu nit -
These ideas have been implemented & tested in online flashcard tutoring systems for biodiversity & German: https://www.teaching-biodiversity.cc/ https://www.teaching-german.cc/ Try them out! (4/5)
1 reply 0 proslijeđenih tweetova 5 korisnika označava da im se sviđaPrikaži ovu nit
This work was led by Adish Singla's group at MPI-SWS, with Anette Hunziker as first author:
https://machineteaching.mpi-sws.org/adishsingla.html …
Also joint work with @yuxinch, @arkrause, @oisinmacaodha, Pietro Perona, and Manuel Gomez-Rodriguez.
(5/5)
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.