Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
By Chelsea Finn, Pieter Abbeel, Sergey Levine. 2017
14,813 citations recorded by the source corpus.
We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning problems, including classification, regression, and reinforcement learning. The goal of meta-learning is to train a model on a variety of learning tasks, such that it can solve new learning tasks using only a small number of training samples. In our approach, the parameters of the model are explicitly trained such that a small number of gradient steps with a small amount of training data from a new task will produce good generalization performance on that task. In effect, our method trains the model to be easy to fine-tune. We demonstrate that this approach leads to state-of-the-art performance on two few-shot image classification benchmarks, produces good results on few-shot regression, and accelerates fine-tuning for policy gradient reinforcement learning with neural network policies.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Sergey Levine
- Yoshua Bengio
- Mohit Bansal
- Quoc V. Le
- Jianfeng Gao
- Percy Liang
- Ruslan Salakhutdinov
- Trevor Darrell
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Trevor Darrell
- Ming-Hsuan Yang
- Xiaogang Wang
- Jitendra Malik
- Bernt Schiele
- Andrew Zisserman
In Representation & Self-Supervised Learning
- Yoshua Bengio
- Trevor Darrell
- Sergey Levine
- Han-Jia Ye
- Dacheng Tao
- De-chuan Zhan
- Andrea Vedaldi
- Fei Sha
Citations in the corpus
- Learning and Evaluating General Linguistic Intelligence (2019, background): “Standard meta learning methods (Finn et al., 2017; Nichol et al., 2018) require the distribution of tasks that we want the model to adapt to to be known in advance (i.e., we need to know all the tasks that we want to evaluate on).”
- Meta-Learning for Low-Resource Neural Machine Translation (2018, methodology): “Following Finn et al. (2017), we achieve this goal by defining the meta-objective function as where k ∼ U ( { 1 , . . . , K } ) refers to one meta-learning episode, and D T , D follow the uniform distribution over T ’s data.”
- The Natural Language Decathlon: Multitask Learning as Question Answering (2018, background): “, 2016; Schmidhuber, 1992), and maximize the degree to which models can learn new tasks (Finn et al., 2017).”
- Few-shot Autoregressive Density Estimation: Towards Learning to Learn Distributions (2017, background): “However, as shown in Finn et al. (2017b) and similar to the setup in Neu & Szepesv´ari (2012), we actually have considerable flexibility here to make the inner and outer objectives different.”
- Massively Multilingual Neural Machine Translation in the Wild: Findings and Challenges (2019, background): “…(Nichol et al., 2018; Baydin et al., 2017), model parameters (Ha et al., 2016a; Platanios et al., 2018) and models that learn new tasks with high sample ef-ficiency (Finn et al., 2017) without forgetting existing tasks or languages (Rusu et al., 2016; Kirk-patrick et al., 2017; Lakew et al., 2018).”
- Do Better ImageNet Models Transfer Better? (2018, methodology): “This problem is closely related to few-shot learning [39, 71, 55, 62, 21, 62, 49], but these methods are typically evaluated with training and test classes from the same distribution.”
- Scaling Laws for Transfer (2021, background): “[FAL17] showed that trying to design a model to be fine-tunable can improve few-shot performance.”
- Finetuned Language Models Are Zero-Shot Learners (2021, background): “In addition, meta-learning (Finn et al., 2017; Vanschoren, 2018, inter alia) also broadly tries to train models that adapt quickly to unseen tasks, typically based on a few examples.”
- Shortcut learning in deep neural networks (2020, background): “An intermediate goal is to learn representations that can adapt quickly to new conditions [136, 137, 138].”
- A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark (2019, neutral): “Ideally, we would sample a new task for each evaluation, as is possible in procedural environments, e.g. (Finn et al., 2017).”
- Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need? (2020, background): “Recently, significant advances [55, 54, 52, 12, 46, 48, 56, 34, 44, 60, 26, 28] have been made to tackle this problem using the ideas of meta-learning or “learning to learn”.”
- Learning with Latent Language (2017, neutral): “, 2016) to generate candidate descriptions more efficiently, or MAML (Finn et al., 2017) rather than zero-shot reward as the training criterion for the interpretation model.”
- Generating images from caption and vice versa via CLIP-Guided Generative Latent Space Search (2021, methodology): “The resulting dataset contains text from over 45 million was trained using a Model-Agnostic Meta-Learning (MAML) method (Finn et al., 2017), and tested with zero-shot approach on a variety of tasks: text generation, translation, summarization, question answering, etc.”
- One-Shot Imitation Learning (2017, background): “Recently, [19] propose a generic framework for meta learning across several aforementioned domains.”
- Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing (2021, neutral): “There are a wide variety of methods to achieve few-shot learning including model agnostic meta-learning [29] (learning features rapidly adaptable to new tasks), embedding learning [8] (embedding each sample in a lower-dimensional space where similar samples are close together), memorybased learning ”
- Repurposing Pretrained Models for Robust Out-of-domain Few-Shot Learning (2021, methodology): “Third, optimization-based approaches search for good initial parameters during training and adapt the pretrained model for new tasks (Finn et al., 2017; Lee & Choi, 2018; Grant et al., 2018).”
- True Few-Shot Learning with Language Models (2021, methodology): “CV has also been used in prior work on multi-distribution few-shot learning [26, 27].”
- Flamingo: a Visual Language Model for Few-Shot Learning (2022, methodology): “In this paper, we take inspiration from the in-context [11] few-shot learning technique instead of more involved few-shot learning approaches based on metric learning [24, 103, 112, 117] or meta-learning [6, 7, 27, 31, 91, 155].”
- Feature Learning in Infinite-Width Neural Networks (2020, methodology): “NTK/GP Feature Learning Infinite-Width Networks on Real Tasks We explicitly calculate this limit for the tasks of Word2Vec [28, 29] and few-shot learning on Omniglot via MAML [13],2 two standard tasks relying crucially on feature learning.”