A new IRMv1 penalty based on extrapolated per-sample gradient magnitudes is proposed, with reported gains over IRM baselines on synthetic and vision benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Robust Invariant Representation Learning by Distribution Extrapolation
A new IRMv1 penalty based on extrapolated per-sample gradient magnitudes is proposed, with reported gains over IRM baselines on synthetic and vision benchmarks.