arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2411.01409cs.LGcs.CLcs.CV
收藏

分类器引导的梯度调制用于增强多模态学习

Classifier-guided Gradient Modulation for Enhanced Multimodal Learning

Zhejiang University(浙江大学);Shanghai AI Lab(上海人工智能实验室)

Zirun Guo, Tao Jin, Jingyuan Chen, Zhou Zhao

AI总结:

提出CGGM方法,通过分类器引导同时调制梯度大小和方向,平衡多模态学习,在四个数据集上优于基线,验证有效性和通用性。

摘要:

近年来,多模态学习发展非常迅速。然而,在多模态训练过程中,模型倾向于仅依赖一种模态,因为基于该模态学习更快,从而导致对其他模态利用不足。现有的平衡训练过程的方法在损失函数、优化器和模态数量上总有一些限制,并且只考虑调制梯度的大小而忽略梯度的方向。为了解决这些问题,本文提出了一种新颖的方法,通过分类器引导的梯度调制(CGGM)来平衡多模态学习,同时考虑梯度的大小和方向。我们在四个多模态数据集上进行了大量实验:UPMC-Food 101、CMU-MOSI、IEMOCAP和BraTS 2021,涵盖了分类、回归和分割任务。结果表明,CGGM在所有基线和最新方法中表现一致地优于它们,证明了其有效性和通用性。我们的代码可在https://github.com/zrguo/CGGM获取。

英文摘要:

Multimodal learning has developed very fast in recent years. However, during the multimodal training process, the model tends to rely on only one modality based on which it could learn faster, thus leading to inadequate use of other modalities. Existing methods to balance the training process always have some limitations on the loss functions, optimizers and the number of modalities and only consider modulating the magnitude of the gradients while ignoring the directions of the gradients. To solve these problems, in this paper, we present a novel method to balance multimodal learning with Classifier-Guided Gradient Modulation (CGGM), considering both the magnitude and directions of the gradients. We conduct extensive experiments on four multimodal datasets: UPMC-Food 101, CMU-MOSI, IEMOCAP and BraTS 2021, covering classification, regression and segmentation tasks. The results show that CGGM outperforms all the baselines and other state-of-the-art methods consistently, demonstrating its effectiveness and versatility. Our code is available at https://github.com/zrguo/CGGM.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑