{
  "id": "2112.08654",
  "version": "v2",
  "published": "2021-12-16T06:17:07.000Z",
  "updated": "2022-03-21T19:26:32.000Z",
  "title": "Learning to Prompt for Continual Learning",
  "authors": [
    "Zifeng Wang",
    "Zizhao Zhang",
    "Chen-Yu Lee",
    "Han Zhang",
    "Ruoxi Sun",
    "Xiaoqi Ren",
    "Guolong Su",
    "Vincent Perot",
    "Jennifer Dy",
    "Tomas Pfister"
  ],
  "comment": "Published at CVPR 2022 as a conference paper",
  "categories": [
    "cs.LG",
    "cs.CV"
  ],
  "abstract": "The mainstream paradigm behind continual learning has been to adapt the model parameters to non-stationary data distributions, where catastrophic forgetting is the central challenge. Typical methods rely on a rehearsal buffer or known task identity at test time to retrieve learned knowledge and address forgetting, while this work presents a new paradigm for continual learning that aims to train a more succinct memory system without accessing task identity at test time. Our method learns to dynamically prompt (L2P) a pre-trained model to learn tasks sequentially under different task transitions. In our proposed framework, prompts are small learnable parameters, which are maintained in a memory space. The objective is to optimize prompts to instruct the model prediction and explicitly manage task-invariant and task-specific knowledge while maintaining model plasticity. We conduct comprehensive experiments under popular image classification benchmarks with different challenging continual learning settings, where L2P consistently outperforms prior state-of-the-art methods. Surprisingly, L2P achieves competitive results against rehearsal-based methods even without a rehearsal buffer and is directly applicable to challenging task-agnostic continual learning. Source code is available at https://github.com/google-research/l2p.",
  "revisions": [
    {
      "version": "v2",
      "updated": "2022-03-21T19:26:32.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "continual learning",
      "popular image classification benchmarks",
      "rehearsal buffer",
      "task identity",
      "test time"
    ],
    "tags": [
      "conference paper"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}