PTLOC adds per-source local constraints to self-supervised acoustic pre-training via first-order MAML-style updates, reporting improved downstream ASR word error rates over a baseline that is not compute-matched.
Local constraints are imposed to ensure that the models optimize each data source to its local optimum after K- step gradient descent initialized from the model
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Heterogeneous Self-Supervised Acoustic Pre-Training with Local Constraints
PTLOC adds per-source local constraints to self-supervised acoustic pre-training via first-order MAML-style updates, reporting improved downstream ASR word error rates over a baseline that is not compute-matched.