A unified momentum-based optimization algorithm with time-varying parameters is shown to converge almost surely under generalized Robbins-Monro and Kiefer-Wolfowitz-Blum conditions, even with biased, unbounded-variance stochastic gradients.
Advances in optimizing recurrent networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Convergence of Momentum-Based Optimization Algorithms with Time-Varying Parameters
A unified momentum-based optimization algorithm with time-varying parameters is shown to converge almost surely under generalized Robbins-Monro and Kiefer-Wolfowitz-Blum conditions, even with biased, unbounded-variance stochastic gradients.