AdaCubic dynamically adjusts the cubic regularization weight via an auxiliary cubic-constrained optimization problem, inherits local convergence guarantees from cubic Newton methods, and matches or exceeds standard optimizers in deep learning tasks with fixed hyperparameters.
However, f (yα )≤ f (x) contradicts (188) for M≥ 2 3LH , which in turn leads to (184), and the proof is complete
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
AdaCubic: An Adaptive Cubic Regularization Optimizer for Deep Learning
AdaCubic dynamically adjusts the cubic regularization weight via an auxiliary cubic-constrained optimization problem, inherits local convergence guarantees from cubic Newton methods, and matches or exceeds standard optimizers in deep learning tasks with fixed hyperparameters.