paper-with-me

홈 › Papers

MetaDIP: Accelerating Deep Image Prior with Meta Learning

2022-09-18 · Kevin Zhang, Mingyang Xie, Maharshi Gor, Yi-Ting Chen, Yvonne Zhou, Christopher A. Metzler

Deep image prior (DIP) is a recently proposed technique for solving imaging inverse problems by fitting the reconstructed images to the output of an untrained convolutional neural network. Unlike pretrained feedforward neural networks, the same DIP can generalize to arbitrary inverse problems, from denoising to phase retrieval, while offering competitive performance at each task. The central disadvantage of DIP is that, while feedforward neural networks can reconstruct an image in a single pass, DIP must gradually update its weights over hundreds to thousands of iterations, at a significant computational cost. In this work we use meta-learning to massively accelerate DIP-based reconstructions. By learning a proper initialization for the DIP weights, we demonstrate a 10x improvement in runtimes across a range of inverse imaging tasks. Moreover, we demonstrate that a network trained to quickly reconstruct faces also generalizes to reconstructing natural image patches.

📄 PDF Abstract BibTeX arXiv:2209.08452

Code (0)

등록된 구현이 없습니다.

Tasks

DenoisingMeta-LearningRetrieval

Similar Papers 제목 키워드 기반

Deep Random Projector: Accelerated Deep Image Prior

2023-01-01 · CVPR 2023 1 · Taihui Li, Hengkang Wang, Zhong Zhuang, Ju Sun

Deep image prior (DIP) has shown great promise in tackling a variety of image restoration (IR) and general visual inverse problems, needing no training data. However, the resulting optimization process is often very …

DenoisingImage DenoisingImage InpaintingImage Restoration+3

Accelerating Online Reinforcement Learning via Model-Based Meta-Learning

2021-03-15 · ICLR Workshop Learning_to_Learn 2021 5 · John D Co-Reyes, Sarah Feng, Glen Berseth, Jie Qui 외

Current reinforcement learning algorithms struggle to quickly adapt to new situations without large amounts of experience and usually without large amounts of optimization over that experience. In this work we seek to l…

Meta-Learningreinforcement-learningReinforcement LearningReinforcement Learning (RL)

FewShotNeRF: Meta-Learning-based Novel View Synthesis for Rapid Scene-Specific Adaptation

2024-08-09 · Piraveen Sivakumar, Paul Janson, Jathushan Rajasegaran, Thanuja Ambegoda

In this paper, we address the challenge of generating novel views of real-world objects with limited multi-view images through our proposed approach, FewShotNeRF. Our method utilizes meta-learning to acquire optimal init…

Meta-LearningNeRFNovel View Synthesis

CoMPS: Continual Meta Policy Search

2021-12-08 · ICLR 2022 4 · Glen Berseth, Zhiwei Zhang, Grace Zhang, Chelsea Finn 외

We develop a new continual meta-learning method to address challenges in sequential multi-task learning. In this setting, the agent's goal is to achieve high reward over any sequence of tasks quickly. Prior meta-reinforc…

Continual Learningcontinuous-controlContinuous ControlMeta-Learning+5

Big AI is accelerating the metacrisis: What can we do?

2025-12-31 · Steven Bird arxiv

The world is in the grip of ecological, meaning, and language crises that are converging into a metacrisis. Big AI is accelerating them all. LLM engineering sits at the core. Despite the public good motives of language e…