重新审视 Softmax 掩码：通过停止梯度增强基于回放的持续学习的稳定性
Revisiting Softmax Masking: Stop Gradient for Enhancing Stability in Replay-based Continual Learning
AI总结：
本文针对持续学习中 softmax 交叉熵诱发遗忘的问题，提出可调节新旧类别梯度尺度的通用掩码 softmax，以提升回放方法的稳定性，并在极小缓冲区下验证其性能增益。
摘要：
在持续学习的基于回放方法中，回放情景记忆中的输入样本已被证明能有效缓解灾难性遗忘。然而，带有 softmax 的交叉熵损失在导致灾难性遗忘方面的潜在关键因素尚未得到充分研究。本文分析了 softmax 的影响，并重新审视使用负无穷的 softmax 掩码，以阐明其缓解灾难性遗忘的能力。基于这些分析，研究发现负无穷掩码 softmax 并不总是与暗知识兼容。为提高这种兼容性，我们提出了一种通用的掩码 softmax，通过调整流向旧类别和新类别的梯度尺度来控制稳定性。我们证明，将该方法用于其他基于回放的方法时可带来更好的性能，主要是通过在持续学习基准测试中增强模型稳定性，即使在缓冲区大小被设置为极小值时也是如此。
英文摘要：
In replay-based methods for continual learning, replaying input samples in episodic memory has shown its effectiveness in alleviating catastrophic forgetting. However, the potential key factor of cross-entropy loss with softmax in causing catastrophic forgetting has been underexplored. In this paper, we analyze the effect of softmax and revisit softmax masking with negative infinity to shed light on its ability to mitigate catastrophic forgetting. Based on the analyses, it is found that negative infinity masked softmax is not always compatible with dark knowledge. To improve the compatibility, we propose a general masked softmax that controls the stability by adjusting the gradient scale to old and new classes. We demonstrate that utilizing our method on other replay-based methods results in better performance, primarily by enhancing model stability in continual learning benchmarks, even when the buffer size is set to an extremely small value.
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。
