The paper introduces Row-wise Matrix AdaGrad and Column-wise Matrix AdaGrad, two adaptive subgradient methods that extend AdaGrad to matrix-valued parameters by using row-wise and column-wise proximal functions. It presents a general Online Mirror Descent framework that derives these optimizers through online regret minimization, providing regret guarantees that can be tighter than entry-wise AdaGrad for structured gradients. Experiments on matrix factorization and deep neural-network training show that aligning adaptive scaling with matrix structure improves optimization stability, allows larger learning rates, and supports greater network depth.
arXiv:2405. 04376v4 Announce Type: replace Abstract: Hyperparameter tuning, particularly the selection of an appropriate learning rate in adaptive gradient training methods, remains a challenge.
arXiv:2402.11215v4 Announce Type: replace Abstract: The choice of batch size in minibatch stochastic gradient optimization is critical for both optimization and generalization performance in large-sc...
arXiv:2412. 19444v2 Announce Type: replace Abstract: Optimization algorithms such as AdaGrad and Adam have significantly advanced the training of deep models by dynamically adjusting the learning rate during the optimization process.
arXiv:2607. 06151v1 Announce Type: new Abstract: Generalization remains a pivotal challenge in deep learning, where traditional optimizers like Stochastic Gradient Descent (SGD) often converge to sharp minima, leading to overfitting and reduced performance on unseen data.
arXiv:2606. 16454v1 Announce Type: cross Abstract: Low-Rank Adaptation (LoRA) enables efficient adaptation of large pre-trained models to downstream tasks by parameterizing weight updates with low-rank matrices.
arXiv:2608. 16760v1 Announce Type: new Abstract: Reliable optimization is central to neural network (NN) training, yet Adam, the default optimizer for modern LLMs, rests on a fragile foundation.
arXiv:2406. 14340v2 Announce Type: replace-cross Abstract: The standard stochastic gradient descent (SGD) optimization method, as well as adaptive methods such as the Adam optimizer fail to converge if the learning rates do not converge to zero (particularly, in the situation of constant learning rates).
arXiv:2209. 03282v5 Announce Type: replace-cross Abstract: Accelerating the convergence of second-order optimization, particularly Newton-type methods, remains a pivotal challenge in algorithmic research.
The article surveys recent neural‑network optimizers, noting that the field has moved beyond simple Adam variants to encompass matrix‑ and layer‑level designs, time‑policy horizons, and state representations that survive sharding and low‑precision computation. It categorizes optimizers along four axes—temporal estimation, update geometry, horizon management, and representation & systems—highlighting methods such as Muon, Shampoo, SOAP, and quantized states. The survey concludes that while matrix‑aware methods are a genuine advance, no single optimizer universally replaces AdamW, and performance depends on model scale, data‑to‑parameter ratio, batch size, schedule, partitioning, tuning budget, and target metric.
arXiv:2606. 03831v1 Announce Type: new Abstract: This paper investigates non-stationary online learning using the metric of interval regret, which requires an online algorithm to perform well over every time interval.
arXiv:2606. 25971v1 Announce Type: new Abstract: Modern neural network training relies on optimizers such as Adam and Muon which act on each weight matrix as a single object.
arXiv:2509. 14969v2 Announce Type: replace Abstract: We introduce a new adaptive step-size strategy for convex optimization with stochastic gradient that exploits the local geometry of the objective function only by means of a first-order stochastic oracle and without any hyper-parameter tuning.