SSLPM, a self-supervised method that reduces redundancy by making features hard to predict from one another, matches state-of-the-art performance, but higher-order redundancy reduction does not clearly improve downstream accuracy.
Hence, we conclude that the exact choice of the objective for the ridge regression is likely not of crucial importance for our method
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Beyond Pairwise Correlations: Higher-Order Redundancies in Self-Supervised Representation Learning
SSLPM, a self-supervised method that reduces redundancy by making features hard to predict from one another, matches state-of-the-art performance, but higher-order redundancy reduction does not clearly improve downstream accuracy.