AdamL: A fast adaptive gradient method incorporating loss function
This work addresses generalization issues in deep learning optimizers, offering an incremental improvement for training neural networks.
The authors tackled the problem of poor generalization in adaptive first-order optimizers by proposing AdamL, a variant of Adam that incorporates loss function information, achieving faster convergence or lower objective values on benchmark functions and deep learning tasks compared to Adam, EAdam, and AdaBelief.
Adaptive first-order optimizers are fundamental tools in deep learning, although they may suffer from poor generalization due to the nonuniform gradient scaling. In this work, we propose AdamL, a novel variant of the Adam optimizer, that takes into account the loss function information to attain better generalization results. We provide sufficient conditions that together with the Polyak-Lojasiewicz inequality, ensure the linear convergence of AdamL. As a byproduct of our analysis, we prove similar convergence properties for the EAdam, and AdaBelief optimizers. Experimental results on benchmark functions show that AdamL typically achieves either the fastest convergence or the lowest objective function values when compared to Adam, EAdam, and AdaBelief. These superior performances are confirmed when considering deep learning tasks such as training convolutional neural networks, training generative adversarial networks using vanilla convolutional neural networks, and long short-term memory networks. Finally, in the case of vanilla convolutional neural networks, AdamL stands out from the other Adam's variants and does not require the manual adjustment of the learning rate during the later stage of the training.