Curriculum learning¶
Bengio, Y., Louradour, J., Collobert, R., & Weston, J. (2009). Curriculum learning. Proceedings of the 26th Annual International Conference on Machine Learning (ICML '09), 41-48.
Cited by¶
2 citations across 2 artifacts.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Pedagogy
- The pattern survives the removal of the schoolroom, the human teacher, and the conscious learner, which is what licenses calling it a prime rather than an education-domain specialty, as Bengio et al. (2009) demonstrated when their curriculum-learning result imported the easy-to-hard sequencing move directly into neural-network training without any human in the loop.
This sourceDemonstrates that ordering training examples easy-to-hard improves neural-network generalization, importing the pedagogical sequencing move into ML without a human in the loop — the load-bearing substrate-independence demonstration this prime cites.
Supported in partVerified against the publisher's abstract
The abstract confirms curriculum learning formalizes ordered, gradually-more-complex example presentation in ML with generalization gains, but does not itself state that no human is in the loop.
“Here, we formalize such training strategies in the context of machine learning, and call them "curriculum learning".”
- The pattern survives the removal of the schoolroom, the human teacher, and the conscious learner, which is what licenses calling it a prime rather than an education-domain specialty, as Bengio et al. (2009) demonstrated when their curriculum-learning result imported the easy-to-hard sequencing move directly into neural-network training without any human in the loop.
- Zone of Proximal Development (ZPD)
- The same ordered-difficulty intuition is formalized in machine-learning curriculum learning by Bengio, Louradour, Collobert, and Weston (2009), where models trained on progressively harder examples converge faster and to better optima — an algorithmic homologue of ZPD-targeted instructional sequencing.
This sourceACM. Introduces curriculum learning in machine learning: training models on progressively harder examples improves convergence and final performance — algorithmic homologue of ZPD-targeted instructional sequencing.
Supported in partVerified against the publisher's abstract
The paper's own abstract confirms it formalizes curriculum learning (training on gradually more complex examples) and reports generalization gains, but frames faster convergence and better local minima only as a hypothesis.
“We hypothesize that curriculum learning has both an effect on the speed of convergence of the training process to a minimum and, in the case of non-convex criteria, on the quality of the local minima obtained: curriculum learning can be seen as a particular form of continuation method (a general strategy for global optimization of non-convex functions).”
- The same ordered-difficulty intuition is formalized in machine-learning curriculum learning by Bengio, Louradour, Collobert, and Weston (2009), where models trained on progressively harder examples converge faster and to better optima — an algorithmic homologue of ZPD-targeted instructional sequencing.
Verification¶
Does it exist? Confirmed. This work's DOI resolves to a registered record, which fixes its identity. That is all it fixes.
Does it back the claim? Read against the text for 2 of 2 citations: 2 supported in part. Each verdict is shown under its citation below, with what in the work backs the sentence.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:cb06cd6fdba9 · see in the full table