提出MISA方法，解决提示调优在持续学习中遗忘严重的问题。
Advancing Prompt-Based Methods for Replay-Independent General Continual Learning
- 用预训练数据初始化提示参数，提升模型泛化能力
- 采用非参数输出掩码，减少灾难性遗忘，无重放时性能提升超18%
- 无需重放缓冲、易部署，适合作为持续学习新基线
通用持续学习（GCL）描述现实世界中无明确任务边界的数据流场景，导致模型初始性能差、泛化能力弱且灾难性遗忘严重。尽管冻结预训练主干配合提示调优可缓解部分问题，但对可调参数的在线适应仍不理想。为此，本文提出MISA（Mask and Initial Session Adaption）方法：通过预训练数据初始化提示参数以增强泛化性，并引入输出层的非参数逻辑掩码机制抑制遗忘。实验证明，该方法在无重放缓冲条件下显著优于现有方法，在CIFAR-100、Tiny-ImageNet和ImageNet-R上分别提升18.39%、22.06%和11.96%。MISA具备即插即用性、无需重放、实现简便且不依赖持续学习相关超参数，可作为强基线用于后续研究。代码已开源。
原文摘要 · Abstract (English)
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
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。