meta learning 17年:MAML SNAIL

2019 年 1 月 2 日 CreateAMind
元学习(meta learning) 最新进展综述论文
专知会员服务
226+阅读 · 2020年5月8日
因果图,Causal Graphs,52页ppt
专知会员服务
177+阅读 · 2020年4月19日
meta learning 17年:MAML SNAIL

maml


1

The process of training a model’s parameters such that a few gradient steps, or even a single gradient step, can pro- duce good results on a new task can be viewed from a fea- ture learning standpoint as building an internal representa- tion that is broadly suitable for many tasks. If the internal representation is suitable to many tasks, simply fine-tuning the parameters slightly (e.g. by primarily modifying the top layer weights in a feedforward model) can produce good results. In effect, our procedure optimizes for models that are easy and fast to fine-tune, allowing the adaptation to happen in the right space for fast learning. From a dynami- cal systems standpoint, our learning process can be viewed as maximizing the sensitivity of the loss functions of new tasks with respect to the parameters: when the sensitivity is high, small local changes to the parameters can lead to



2.2

we propose a method that can learn the parameters of any standard model via meta-learning in such a way as to prepare that model for fast adaptation. The intuition behind this approach is that some internal representations are more transferrable than others. For example, a neural network might learn internal features that are broadly applicable to all tasks in p(T ), rather than a single individual task. How can we en- courage the emergence of such general-purpose representa- tions? We take an explicit approach to this problem: since the model will be fine-tuned using a gradient-based learn- ing rule on a new task, we will aim to learn a model in such a way that this gradient-based learning rule can make rapid progress on new tasks drawn from p(T ), without overfit- ting. In effect, we will aim to find model parameters that are sensitive to changes in the task, such that small changes in the parameters will produce large improvements on the loss function of any task drawn from p(T ), when altered in the direction of the gradient of that loss (see Figure 1). We




https://github.com/tristandeleu/pytorch-maml-rl

https://github.com/cbfinn/maml_rl










-------------------------------

A SIMPLE NEURAL ATTENTIVE META-LEARNER


We propose a class of simple and generic meta-learner architectures that use a novel combination of temporal convolutions and soft attention; the former to aggregate information from past experience and the latter to pinpoint specific pieces of information.

 


1

Meta-learning can be formalized as a sequence-to-sequence problem; in existing approaches that adopt this view, the bottleneck is in the meta-learner’s ability to internalize and refer to past experience. Thus, we propose a class of model architectures that addresses this shortcoming: we combine temporal convolutions, which enable the meta-learner to aggregate contextual information from past experience, with causal attention, which allow it to pinpoint specific pieces of information within that context.


3

construct SNAIL by combining the two: we use temporal convolutions to produce the context over which we use a causal attention operation. By interleaving TC layers with causal attention layers, SNAIL can have high-bandwidth access over its past experience without constraints on the amount of experience it can effectively use. By using attention at multiple stages within a model that is trained end-to-end, SNAIL can learn what pieces of information to pick out from the experience it gathers, as well as a feature representation that is amenable to doing so easily. As an additional benefit, SNAIL architectures are easier to train than traditional RNNs such as LSTM or GRUs 

 

3.1 MODULAR BUILDING BLOCKS

 


 

实验领域及效果


5 EXPERIMENTS

Our experiments were designed to investigate the following questions:

• How does SNAIL’s generality affect its performance on a range of meta-learning tasks?

• How does its performance compare to existing approaches that are specialized to a particular

task domain, or have elements of a high-level strategy already built-in?

• HowdoesSNAILscalewithhigh-dimensionalinputsandlong-termtemporaldependencies?

5.1 FEW-SHOT IMAGE CLASSIFICATION



https://github.com/sagelywizard/snail

https://github.com/eambutu/snail-pytorch




登录查看更多
9

相关内容

MAML(Model-Agnostic Meta-Learning)是元学习(Meta learning)最经典的几个算法之一,出自论文《Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks》。 原文地址:https://arxiv.org/abs/1703.03400
元学习(meta learning) 最新进展综述论文
专知会员服务
226+阅读 · 2020年5月8日
因果图,Causal Graphs,52页ppt
专知会员服务
177+阅读 · 2020年4月19日
【Uber AI新论文】持续元学习,Learning to Continually Learn
专知会员服务
29+阅读 · 2020年2月27日
深度强化学习策略梯度教程,53页ppt
专知会员服务
130+阅读 · 2020年2月1日
【斯坦福大学Chelsea Finn-NeurIPS 2019】贝叶斯元学习
专知会员服务
35+阅读 · 2019年12月17日
强化学习最新教程,17页pdf
专知会员服务
90+阅读 · 2019年10月11日
Hierarchically Structured Meta-learning
CreateAMind
14+阅读 · 2019年5月22日
Transferring Knowledge across Learning Processes
CreateAMind
8+阅读 · 2019年5月18日
逆强化学习-学习人先验的动机
CreateAMind
6+阅读 · 2019年1月18日
强化学习的Unsupervised Meta-Learning
CreateAMind
7+阅读 · 2019年1月7日
无监督元学习表示学习
CreateAMind
22+阅读 · 2019年1月4日
Unsupervised Learning via Meta-Learning
CreateAMind
32+阅读 · 2019年1月3日
Hierarchical Imitation - Reinforcement Learning
CreateAMind
16+阅读 · 2018年5月25日
Auto-Encoding GAN
CreateAMind
5+阅读 · 2017年8月4日
强化学习 cartpole_a3c
CreateAMind
9+阅读 · 2017年7月21日
Meta-Transfer Learning for Zero-Shot Super-Resolution
Arxiv
40+阅读 · 2020年2月27日
Meta-Learning with Implicit Gradients
Arxiv
7+阅读 · 2019年9月10日
Arxiv
12+阅读 · 2019年1月26日
Arxiv
6+阅读 · 2018年12月26日
Arxiv
118+阅读 · 2018年10月8日
Multi-task Deep Reinforcement Learning with PopArt
Arxiv
3+阅读 · 2018年9月12日
Meta-Learning with Latent Embedding Optimization
Arxiv
6+阅读 · 2018年7月16日
小贴士
相关VIP内容
元学习(meta learning) 最新进展综述论文
专知会员服务
226+阅读 · 2020年5月8日
因果图,Causal Graphs,52页ppt
专知会员服务
177+阅读 · 2020年4月19日
【Uber AI新论文】持续元学习,Learning to Continually Learn
专知会员服务
29+阅读 · 2020年2月27日
深度强化学习策略梯度教程,53页ppt
专知会员服务
130+阅读 · 2020年2月1日
【斯坦福大学Chelsea Finn-NeurIPS 2019】贝叶斯元学习
专知会员服务
35+阅读 · 2019年12月17日
强化学习最新教程,17页pdf
专知会员服务
90+阅读 · 2019年10月11日
相关资讯
Hierarchically Structured Meta-learning
CreateAMind
14+阅读 · 2019年5月22日
Transferring Knowledge across Learning Processes
CreateAMind
8+阅读 · 2019年5月18日
逆强化学习-学习人先验的动机
CreateAMind
6+阅读 · 2019年1月18日
强化学习的Unsupervised Meta-Learning
CreateAMind
7+阅读 · 2019年1月7日
无监督元学习表示学习
CreateAMind
22+阅读 · 2019年1月4日
Unsupervised Learning via Meta-Learning
CreateAMind
32+阅读 · 2019年1月3日
Hierarchical Imitation - Reinforcement Learning
CreateAMind
16+阅读 · 2018年5月25日
Auto-Encoding GAN
CreateAMind
5+阅读 · 2017年8月4日
强化学习 cartpole_a3c
CreateAMind
9+阅读 · 2017年7月21日
相关论文
Meta-Transfer Learning for Zero-Shot Super-Resolution
Arxiv
40+阅读 · 2020年2月27日
Meta-Learning with Implicit Gradients
Arxiv
7+阅读 · 2019年9月10日
Arxiv
12+阅读 · 2019年1月26日
Arxiv
6+阅读 · 2018年12月26日
Arxiv
118+阅读 · 2018年10月8日
Multi-task Deep Reinforcement Learning with PopArt
Arxiv
3+阅读 · 2018年9月12日
Meta-Learning with Latent Embedding Optimization
Arxiv
6+阅读 · 2018年7月16日
Top
微信扫码咨询专知VIP会员