AlphaGrad normalizes gradients layer-wise and applies tanh(alpha times the normalized gradient), reporting advantages over Adam in on-policy PPO but instability in off-policy DQN, with convergence to stationarity proven.
A., Veness, J., Bellemare, M
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
-
AlphaGrad: Non-Linear Gradient Normalization Optimizer
AlphaGrad normalizes gradients layer-wise and applies tanh(alpha times the normalized gradient), reporting advantages over Adam in on-policy PPO but instability in off-policy DQN, with convergence to stationarity proven.