aiqbench takeMachine Learning

This paper introduces AdamNX, an optimization algorithm that modifies the exponential decay of the second-moment estimate in standard Adam to better approximate momentum-SGD behavior during training plateaus. The authors demonstrate performance gains across image classification and object detection tasks while explicitly noting these results do not yet prove broader statistical effects or generalization improvements beyond their specific benchmarks.

Source coveringarXiv

What this isA model's reading of the article and related coverage, written for aiqbench. Not the publisher's words.

AdamNX: An Adam improvement algorithm based on a novel exponential decay mechanism for the second-order moment estimate

arXivJuly 27, 20261 min read

arXiv:2511.13465v5 Announce Type: replace Abstract: This paper studies the exponential decay mechanism of the second-moment estimate in Adam. We propose AdamNX and a time-varying second-moment decay rate that gradually weakens the correction applied to the update scale. Under the assumptions used in our analysis, this mechanism makes the updates approach momentum-SGD-like behavior during the training plateau phase. We report results on the image-classification, object-detection, and semantic-segmentation tasks, configurations, and comparators included in this paper. These results do not establish multi-seed statistical effects, flatness, or generalization beyond the reported tasks. Our code is open-sourced at https://github.com/mengzhu0308/AdamNX.

Abstract:This paper studies the exponential decay mechanism of the second-moment estimate in Adam. We propose AdamNX and a time-varying second-moment decay rate that gradually weakens the correction applied to the update scale. Under the assumptions used in our analysis, this mechanism makes the updates approach momentum-SGD-like behavior during the training plateau phase. We report results on the image-classification, object-detection, and semantic-segmentation tasks, configurations, and comparators included in this paper. These results do not establish multi-seed statistical effects, flatness, or generalization beyond the reported tasks. Our code is open-sourced at this https URL.

Machine Learning (cs.LG); Machine Learning (stat.ML)

arXiv:2511.13465 [cs.LG]

(or arXiv:2511.13465v5 [cs.LG] for this version)

Full text under the publisher's open licence · 106 words

Read at arXiv ↗Ask about this article