Deep Reinforcement Learning That Matters.¶
Henderson, P., Islam, R., Bachman, P., Pineau, J., Precup, D., & Meger, D. (2018). Deep Reinforcement Learning That Matters. Proceedings of the AAAI Conference on Artificial Intelligence, 3207-3214.
Cited by¶
1 citation across 1 artifact.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Minimal Pairs
- Phonological pairs leak prosody; A/B tests leak time-of-day and traffic source; ablations leak training noise and seed variance; lesion studies leak compensatory reorganization.
This sourceDocuments how random-seed variance and non-determinism confound reported gains, and prescribes averaging over seeds / pre-declaring held-constant factors — the leakage failure of ablation-style comparisons and its defense.
Supported in partVerified against the work's full text
Henderson et al. show that random-seed variance alone can produce statistically different results in deep RL, supporting only the seed-variance clause of the sentence.
“even averaging several learning results together across totally dif- ferent random seeds can lead to the reporting of misleading results”
- Phonological pairs leak prosody; A/B tests leak time-of-day and traffic source; ablations leak training noise and seed variance; lesion studies leak compensatory reorganization.
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 1 of 1 citation: 1 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:57b9cbcd027c · see in the full table