推进基于提示的方法以实现独立于重放的通用持续学习
Advancing Prompt-Based Methods for Replay-Independent General Continual Learning
AI总结：
针对通用持续学习中模糊任务边界导致的灾难性遗忘问题，提出MISA方法，通过遗忘感知的初始会话适应和非参数化logit掩码提升基于提示方法的性能，无需重放即取得显著提升。
摘要：
通用持续学习（GCL）是一个描述现实世界持续学习（CL）问题的广泛概念，这些问题通常具有在线数据流且任务之间没有明显转换（即模糊的任务边界）的特征。这些要求导致初始性能差、泛化能力有限以及严重的灾难性遗忘，严重影响了从头训练的主流GCL模型的有效性。虽然使用冻结的预训练骨干网络结合适当的提示微调可以部分应对这些挑战，但这种基于提示的方法对于在运行中微调剩余可调参数的CL来说仍然不是最优的。对此，我们提出了一种名为MISA（Mask and Initial Session Adaption）的创新方法，以推进GCL中基于提示的方法。它包含一个遗忘感知的初始会话适应，利用预训练数据初始化提示参数并提高泛化能力，以及输出层的非参数化logit掩码以缓解灾难性遗忘。实验结果表明，与近期的竞争方法相比，我们的方法取得了显著的性能提升，尤其是在没有重放缓冲区的情况下（例如，在CIFAR-100、Tiny-ImageNet和ImageNet-R上分别取得了高达18.39%、22.06%和11.96%的性能领先）。此外，我们的方法具有基于提示方法的即插即用特性、独立于重放、易于实现以及避免CL相关超参数的特点，可作为GCL研究的强基线。我们的源代码已在 https://github.com/kangzhiq/MISA 公开。
英文摘要：
General continual learning (GCL) is a broad concept to describe real-world continual learning (CL) problems, which are often characterized by online data streams without distinct transitions between tasks, i.e., blurry task boundaries. Such requirements result in poor initial performance, limited generalizability, and severe catastrophic forgetting, heavily impacting the effectiveness of mainstream GCL models trained from scratch. While the use of a frozen pretrained backbone with appropriate prompt tuning can partially address these challenges, such prompt-based methods remain suboptimal for CL of remaining tunable parameters on the fly. In this regard, we propose an innovative approach named MISA (Mask and Initial Session Adaption) to advance prompt-based methods in GCL. It includes a forgetting-aware initial session adaption that employs pretraining data to initialize prompt parameters and improve generalizability, as well as a non-parametric logit mask of the output layers to mitigate catastrophic forgetting. Empirical results demonstrate substantial performance gains of our approach compared to recent competitors, especially without a replay buffer (e.g., up to 18.39%, 22.06%, and 11.96% performance lead on CIFAR-100, Tiny-ImageNet, and ImageNet-R, respectively). Moreover, our approach features the plug-in nature for prompt-based methods, independence of replay, ease of implementation, and avoidance of CL-relevant hyperparameters, serving as a strong baseline for GCL research. Our source code is publicly available at https://github.com/kangzhiq/MISA
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。
