SlideShare a Scribd company logo
1 of 25
Download to read offline
Model-Agnostic Meta-Learning for Fast
Adaptation of Deep Networks
Yoonho Lee
Department of Computer Science and Engineering
Pohang University of Science and Technology
September 14, 2017
Meta-learning
Learning to learn(faster, safer etc)
Meta-learning
RL methods take a long time to train: needs meta-learning
The meta-train set for humans would be: objects in real life,
experience playing different games etc
Previous Deep Meta-Learning Methods
Previous Deep Meta-Learning Methods
RNNs as learners[10][12][2][7]
Metric Learning[5][11]
Optimizer Learning[9][1]
Previous Deep Meta-Learning Methods
RNNs as learners
Previous Deep Meta-Learning Methods
RNNs as learners12
(assuming a sufficiently expressive RNN) Search space
includes all conceivable ML algorithms
Moves the burden of innovation to RNNs
Ignores advances achieved in ML by humans
Subpar results
1
Adam Santoro et al. “One-shot Learning with Memory-Augmented Neural
Networks”. In: ICML (2016).
2
Yan Duan et al. “RLˆ2: Fast Reinforcement Learning via Slow
Reinforcement Learning”. In: (2016).
Previous Deep Meta-Learning Methods
Metric Learning
Previous Deep Meta-Learning Methods
Metric Learning34
Learn a metric in input space
Specialized to one/few-shot classification(Omniglot,
MiniImageNet etc)
Cannot use in other problems (e.g. RL)
3
Gregory Koch, Richard Zemel, and Ruslan Salakhutdinov. “Siamese Neural
Networks for One-shot Image Recognition”. In: ICML (2015).
4
Oriol Vinyals et al. “Matching Networks for One Shot Learning”. In:
NIPS (2016).
Previous Deep Meta-Learning Methods
Optimizer Learning
Previous Deep Meta-Learning Methods
Optimizer Learning56
Learn parameter update given gradients (search space includes
SGD, RMSProp, Adam etc)
Applicable to any architecture/task
Best performance on Omniglot, MiniImageNet
5
Sachin Ravi and Hugo Larochelle. “Optimization as a Model for Few-shot
Learning”. In: ICLR (2017).
6
Marcin Andrychowicz et al. “Learning to learn by gradient descent by
gradient descent”. In: NIPS (2016).
Model-Agnostic Meta-Learning for Fast Adaptation
of Deep Networks
Chelsea Finn Pieter Abbeel, Sergey Levine
ImageNet pre-training
Pretrain network on ImageNet classification, then fine-tune
network on new task.
Enables NNs to learn new vision tasks using relatively small
datasets.
This works because we have a huge labelled image dataset
and the manifold of images have a somewhat consistent
structure even between different datasets and tasks.
How do we bring ’Initialization as meta-learning’ to non-vision
domains such as speech/NLP/RL?
Model-Agnostic Meta-Learning
Model-Agnostic Meta-Learning
Model-Agnostic Meta-Learning
Supervised Learning
Model-Agnostic Meta-Learning
Few-shot Sine wave regression experiments
Model-Agnostic Meta-Learning
One/few-shot Classification experiments
Model-Agnostic Meta-Learning
Reinforcement Learning
Score function estimator in lines 6/10, since we cannot
backprop through environment dynamics
Model-Agnostic Meta-Learning
RL experiments
https://sites.google.com/view/maml
Model-Agnostic Meta-Learning
Extension7
7
Zhenguo Li et al. “Meta-SGD: Learning to Learn Quickly for Few Shot
Learning”. In: (2017).
Discussion
Parameter space noise (as opposed to policy space noise) has
been shown to result in more consistent exploration89. This
supports MAML’s idea.
Why does MAML not overfit when taking multiple gradient
steps?
Do we need to overwrite all weights during adaptation?
8
Matthias Plappert et al. “Parameter Space Noise for Exploration”. In:
(2017).
9
Meire Fortunato et al. “Noisy Networks for Exploration”. In: (2017).
References I
[1] Marcin Andrychowicz et al. “Learning to learn by gradient
descent by gradient descent”. In: NIPS (2016).
[2] Yan Duan et al. “RLˆ2: Fast Reinforcement Learning via
Slow Reinforcement Learning”. In: (2016).
[3] Chelsea Finn, Pieter Abbeel, and Sergey Levine.
“Model-Agnostic Meta-Learning for Fast Adaptation of Deep
Networks”. In: (2017).
[4] Meire Fortunato et al. “Noisy Networks for Exploration”. In:
(2017).
[5] Gregory Koch, Richard Zemel, and Ruslan Salakhutdinov.
“Siamese Neural Networks for One-shot Image Recognition”.
In: ICML (2015).
[6] Zhenguo Li et al. “Meta-SGD: Learning to Learn Quickly for
Few Shot Learning”. In: (2017).
References II
[7] Nikhil Mishra, Mostafa Rohaninejad, and Xi UC Chen Pieter
Abbeel Berkeley. “Meta-Learning with Temporal
Convolutions”. In: (2017).
[8] Matthias Plappert et al. “Parameter Space Noise for
Exploration”. In: (2017).
[9] Sachin Ravi and Hugo Larochelle. “Optimization as a Model
for Few-shot Learning”. In: ICLR (2017).
[10] Adam Santoro et al. “One-shot Learning with
Memory-Augmented Neural Networks”. In: ICML (2016).
[11] Oriol Vinyals et al. “Matching Networks for One Shot
Learning”. In: NIPS (2016).
[12] Jx Wang et al. “Learning to Reinforcement Learn”. In:
(2016).
Thank You

More Related Content

What's hot

What's hot (20)

Model based rl
Model based rlModel based rl
Model based rl
 
Interpretability beyond feature attribution quantitative testing with concept...
Interpretability beyond feature attribution quantitative testing with concept...Interpretability beyond feature attribution quantitative testing with concept...
Interpretability beyond feature attribution quantitative testing with concept...
 
Understanding RNN and LSTM
Understanding RNN and LSTMUnderstanding RNN and LSTM
Understanding RNN and LSTM
 
【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive...
【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive...【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive...
【DL輪読会】Self-Supervised Learning from Images with a Joint-Embedding Predictive...
 
【DL輪読会】GPT-4Technical Report
【DL輪読会】GPT-4Technical Report【DL輪読会】GPT-4Technical Report
【DL輪読会】GPT-4Technical Report
 
Optimization as a model for few shot learning
Optimization as a model for few shot learningOptimization as a model for few shot learning
Optimization as a model for few shot learning
 
【DL輪読会】Hierarchical Text-Conditional Image Generation with CLIP Latents
【DL輪読会】Hierarchical Text-Conditional Image Generation with CLIP Latents【DL輪読会】Hierarchical Text-Conditional Image Generation with CLIP Latents
【DL輪読会】Hierarchical Text-Conditional Image Generation with CLIP Latents
 
Deep Q-Network 論文輪読会
Deep Q-Network 論文輪読会Deep Q-Network 論文輪読会
Deep Q-Network 論文輪読会
 
【DL輪読会】ViT + Self Supervised Learningまとめ
【DL輪読会】ViT + Self Supervised Learningまとめ【DL輪読会】ViT + Self Supervised Learningまとめ
【DL輪読会】ViT + Self Supervised Learningまとめ
 
【DL輪読会】論文解説:Offline Reinforcement Learning as One Big Sequence Modeling Problem
【DL輪読会】論文解説:Offline Reinforcement Learning as One Big Sequence Modeling Problem【DL輪読会】論文解説:Offline Reinforcement Learning as One Big Sequence Modeling Problem
【DL輪読会】論文解説:Offline Reinforcement Learning as One Big Sequence Modeling Problem
 
Introduction to Recurrent Neural Network
Introduction to Recurrent Neural NetworkIntroduction to Recurrent Neural Network
Introduction to Recurrent Neural Network
 
(DL輪読)Matching Networks for One Shot Learning
(DL輪読)Matching Networks for One Shot Learning(DL輪読)Matching Networks for One Shot Learning
(DL輪読)Matching Networks for One Shot Learning
 
[DL輪読会]“SimPLe”,“Improved Dynamics Model”,“PlaNet” 近年のVAEベース系列モデルの進展とそのモデルベース...
[DL輪読会]“SimPLe”,“Improved Dynamics Model”,“PlaNet” 近年のVAEベース系列モデルの進展とそのモデルベース...[DL輪読会]“SimPLe”,“Improved Dynamics Model”,“PlaNet” 近年のVAEベース系列モデルの進展とそのモデルベース...
[DL輪読会]“SimPLe”,“Improved Dynamics Model”,“PlaNet” 近年のVAEベース系列モデルの進展とそのモデルベース...
 
[DL輪読会] Adversarial Skill Chaining for Long-Horizon Robot Manipulation via T...
[DL輪読会] Adversarial Skill Chaining for Long-Horizon Robot Manipulation via  T...[DL輪読会] Adversarial Skill Chaining for Long-Horizon Robot Manipulation via  T...
[DL輪読会] Adversarial Skill Chaining for Long-Horizon Robot Manipulation via T...
 
【DL輪読会】Transformers are Sample Efficient World Models
【DL輪読会】Transformers are Sample Efficient World Models【DL輪読会】Transformers are Sample Efficient World Models
【DL輪読会】Transformers are Sample Efficient World Models
 
Model-Based Reinforcement Learning @NIPS2017
Model-Based Reinforcement Learning @NIPS2017Model-Based Reinforcement Learning @NIPS2017
Model-Based Reinforcement Learning @NIPS2017
 
Transformers AI PPT.pptx
Transformers AI PPT.pptxTransformers AI PPT.pptx
Transformers AI PPT.pptx
 
Long Short Term Memory
Long Short Term MemoryLong Short Term Memory
Long Short Term Memory
 
Meta-Learning with Memory-Augmented Neural Networks (MANN)
Meta-Learning with Memory-Augmented Neural Networks (MANN)Meta-Learning with Memory-Augmented Neural Networks (MANN)
Meta-Learning with Memory-Augmented Neural Networks (MANN)
 
Segment Anything
Segment AnythingSegment Anything
Segment Anything
 

Similar to Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

REU_2015_poster_DustynTubbs_PatrickStreifel
REU_2015_poster_DustynTubbs_PatrickStreifelREU_2015_poster_DustynTubbs_PatrickStreifel
REU_2015_poster_DustynTubbs_PatrickStreifel
Patrick Streifel
 
deeplearningpresentation-180625071236.pptx
deeplearningpresentation-180625071236.pptxdeeplearningpresentation-180625071236.pptx
deeplearningpresentation-180625071236.pptx
JeetDesai14
 
Toward enhancement of deep learning techniques using fuzzy logic: a survey
Toward enhancement of deep learning techniques using fuzzy logic: a survey Toward enhancement of deep learning techniques using fuzzy logic: a survey
Toward enhancement of deep learning techniques using fuzzy logic: a survey
IJECEIAES
 

Similar to Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks (20)

Eat it, Review it: A New Approach for Review Prediction
Eat it, Review it: A New Approach for Review PredictionEat it, Review it: A New Approach for Review Prediction
Eat it, Review it: A New Approach for Review Prediction
 
Gradient-Based Meta-Learning with Learned Layerwise Metric and Subspace
Gradient-Based Meta-Learning with Learned Layerwise Metric and SubspaceGradient-Based Meta-Learning with Learned Layerwise Metric and Subspace
Gradient-Based Meta-Learning with Learned Layerwise Metric and Subspace
 
Gradient-based Meta-learning with learned layerwise subspace and metric
Gradient-based Meta-learning with learned layerwise subspace and metricGradient-based Meta-learning with learned layerwise subspace and metric
Gradient-based Meta-learning with learned layerwise subspace and metric
 
REU_2015_poster_DustynTubbs_PatrickStreifel
REU_2015_poster_DustynTubbs_PatrickStreifelREU_2015_poster_DustynTubbs_PatrickStreifel
REU_2015_poster_DustynTubbs_PatrickStreifel
 
Visual concept learning
Visual concept learningVisual concept learning
Visual concept learning
 
Research Directions - Full Stack Deep Learning
Research Directions - Full Stack Deep LearningResearch Directions - Full Stack Deep Learning
Research Directions - Full Stack Deep Learning
 
Deep learning presentation
Deep learning presentationDeep learning presentation
Deep learning presentation
 
deeplearningpresentation-180625071236.pptx
deeplearningpresentation-180625071236.pptxdeeplearningpresentation-180625071236.pptx
deeplearningpresentation-180625071236.pptx
 
Architecture Design for Deep Neural Networks III
Architecture Design for Deep Neural Networks IIIArchitecture Design for Deep Neural Networks III
Architecture Design for Deep Neural Networks III
 
Introduction to Soft Computing
Introduction to Soft Computing Introduction to Soft Computing
Introduction to Soft Computing
 
Image captioning using DL and NLP.pptx
Image captioning using DL and NLP.pptxImage captioning using DL and NLP.pptx
Image captioning using DL and NLP.pptx
 
3234150
32341503234150
3234150
 
Toward enhancement of deep learning techniques using fuzzy logic: a survey
Toward enhancement of deep learning techniques using fuzzy logic: a survey Toward enhancement of deep learning techniques using fuzzy logic: a survey
Toward enhancement of deep learning techniques using fuzzy logic: a survey
 
DLD meetup 2017, Efficient Deep Learning
DLD meetup 2017, Efficient Deep LearningDLD meetup 2017, Efficient Deep Learning
DLD meetup 2017, Efficient Deep Learning
 
Introduction to Deep learning
Introduction to Deep learningIntroduction to Deep learning
Introduction to Deep learning
 
Survey on Text Prediction Techniques
Survey on Text Prediction TechniquesSurvey on Text Prediction Techniques
Survey on Text Prediction Techniques
 
Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...
Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...
Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...
 
Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...
Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...
Two Cognitive Architectures for General Intelligence - Cortical Feedback & Ep...
 
Ravi's SOP Princeton
Ravi's SOP Princeton Ravi's SOP Princeton
Ravi's SOP Princeton
 
Emotional Learning in a Simulated Model of the Mental Apparatus
Emotional Learning in a Simulated Model of the Mental Apparatus Emotional Learning in a Simulated Model of the Mental Apparatus
Emotional Learning in a Simulated Model of the Mental Apparatus
 

More from Yoonho Lee

More from Yoonho Lee (11)

Meta-learning and the ELBO
Meta-learning and the ELBOMeta-learning and the ELBO
Meta-learning and the ELBO
 
On First-Order Meta-Learning Algorithms
On First-Order Meta-Learning AlgorithmsOn First-Order Meta-Learning Algorithms
On First-Order Meta-Learning Algorithms
 
New Insights and Perspectives on the Natural Gradient Method
New Insights and Perspectives on the Natural Gradient MethodNew Insights and Perspectives on the Natural Gradient Method
New Insights and Perspectives on the Natural Gradient Method
 
Parameter Space Noise for Exploration
Parameter Space Noise for ExplorationParameter Space Noise for Exploration
Parameter Space Noise for Exploration
 
Meta Learning Shared Hierarchies
Meta Learning Shared HierarchiesMeta Learning Shared Hierarchies
Meta Learning Shared Hierarchies
 
Continuous Adaptation via Meta Learning in Nonstationary and Competitive Envi...
Continuous Adaptation via Meta Learning in Nonstationary and Competitive Envi...Continuous Adaptation via Meta Learning in Nonstationary and Competitive Envi...
Continuous Adaptation via Meta Learning in Nonstationary and Competitive Envi...
 
The Predictron: End-to-end Learning and Planning
The Predictron: End-to-end Learning and PlanningThe Predictron: End-to-end Learning and Planning
The Predictron: End-to-end Learning and Planning
 
Dueling Network Architectures for Deep Reinforcement Learning
Dueling Network Architectures for Deep Reinforcement LearningDueling Network Architectures for Deep Reinforcement Learning
Dueling Network Architectures for Deep Reinforcement Learning
 
Modular Multitask Reinforcement Learning with Policy Sketches
Modular Multitask Reinforcement Learning with Policy SketchesModular Multitask Reinforcement Learning with Policy Sketches
Modular Multitask Reinforcement Learning with Policy Sketches
 
Evolution Strategies as a Scalable Alternative to Reinforcement Learning
Evolution Strategies as a Scalable Alternative to Reinforcement LearningEvolution Strategies as a Scalable Alternative to Reinforcement Learning
Evolution Strategies as a Scalable Alternative to Reinforcement Learning
 
Gradient Estimation Using Stochastic Computation Graphs
Gradient Estimation Using Stochastic Computation GraphsGradient Estimation Using Stochastic Computation Graphs
Gradient Estimation Using Stochastic Computation Graphs
 

Recently uploaded

If this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New NigeriaIf this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New Nigeria
Kayode Fayemi
 
Uncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac FolorunsoUncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac Folorunso
Kayode Fayemi
 
No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...
No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...
No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...
Sheetaleventcompany
 
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptxChiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptx
raffaeleoman
 

Recently uploaded (20)

If this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New NigeriaIf this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New Nigeria
 
Uncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac FolorunsoUncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac Folorunso
 
SaaStr Workshop Wednesday w/ Lucas Price, Yardstick
SaaStr Workshop Wednesday w/ Lucas Price, YardstickSaaStr Workshop Wednesday w/ Lucas Price, Yardstick
SaaStr Workshop Wednesday w/ Lucas Price, Yardstick
 
Presentation on Engagement in Book Clubs
Presentation on Engagement in Book ClubsPresentation on Engagement in Book Clubs
Presentation on Engagement in Book Clubs
 
Dreaming Marissa Sánchez Music Video Treatment
Dreaming Marissa Sánchez Music Video TreatmentDreaming Marissa Sánchez Music Video Treatment
Dreaming Marissa Sánchez Music Video Treatment
 
No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...
No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...
No Advance 8868886958 Chandigarh Call Girls , Indian Call Girls For Full Nigh...
 
lONG QUESTION ANSWER PAKISTAN STUDIES10.
lONG QUESTION ANSWER PAKISTAN STUDIES10.lONG QUESTION ANSWER PAKISTAN STUDIES10.
lONG QUESTION ANSWER PAKISTAN STUDIES10.
 
The workplace ecosystem of the future 24.4.2024 Fabritius_share ii.pdf
The workplace ecosystem of the future 24.4.2024 Fabritius_share ii.pdfThe workplace ecosystem of the future 24.4.2024 Fabritius_share ii.pdf
The workplace ecosystem of the future 24.4.2024 Fabritius_share ii.pdf
 
Call Girl Number in Khar Mumbai📲 9892124323 💞 Full Night Enjoy
Call Girl Number in Khar Mumbai📲 9892124323 💞 Full Night EnjoyCall Girl Number in Khar Mumbai📲 9892124323 💞 Full Night Enjoy
Call Girl Number in Khar Mumbai📲 9892124323 💞 Full Night Enjoy
 
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort ServiceBDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
 
My Presentation "In Your Hands" by Halle Bailey
My Presentation "In Your Hands" by Halle BaileyMy Presentation "In Your Hands" by Halle Bailey
My Presentation "In Your Hands" by Halle Bailey
 
Air breathing and respiratory adaptations in diver animals
Air breathing and respiratory adaptations in diver animalsAir breathing and respiratory adaptations in diver animals
Air breathing and respiratory adaptations in diver animals
 
Dreaming Music Video Treatment _ Project & Portfolio III
Dreaming Music Video Treatment _ Project & Portfolio IIIDreaming Music Video Treatment _ Project & Portfolio III
Dreaming Music Video Treatment _ Project & Portfolio III
 
Causes of poverty in France presentation.pptx
Causes of poverty in France presentation.pptxCauses of poverty in France presentation.pptx
Causes of poverty in France presentation.pptx
 
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdfAWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
 
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
 
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptxChiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptx
 
ANCHORING SCRIPT FOR A CULTURAL EVENT.docx
ANCHORING SCRIPT FOR A CULTURAL EVENT.docxANCHORING SCRIPT FOR A CULTURAL EVENT.docx
ANCHORING SCRIPT FOR A CULTURAL EVENT.docx
 
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptx
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptxMohammad_Alnahdi_Oral_Presentation_Assignment.pptx
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptx
 
Thirunelveli call girls Tamil escorts 7877702510
Thirunelveli call girls Tamil escorts 7877702510Thirunelveli call girls Tamil escorts 7877702510
Thirunelveli call girls Tamil escorts 7877702510
 

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

  • 1. Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks Yoonho Lee Department of Computer Science and Engineering Pohang University of Science and Technology September 14, 2017
  • 3. Meta-learning RL methods take a long time to train: needs meta-learning The meta-train set for humans would be: objects in real life, experience playing different games etc
  • 5. Previous Deep Meta-Learning Methods RNNs as learners[10][12][2][7] Metric Learning[5][11] Optimizer Learning[9][1]
  • 6. Previous Deep Meta-Learning Methods RNNs as learners
  • 7. Previous Deep Meta-Learning Methods RNNs as learners12 (assuming a sufficiently expressive RNN) Search space includes all conceivable ML algorithms Moves the burden of innovation to RNNs Ignores advances achieved in ML by humans Subpar results 1 Adam Santoro et al. “One-shot Learning with Memory-Augmented Neural Networks”. In: ICML (2016). 2 Yan Duan et al. “RLˆ2: Fast Reinforcement Learning via Slow Reinforcement Learning”. In: (2016).
  • 8. Previous Deep Meta-Learning Methods Metric Learning
  • 9. Previous Deep Meta-Learning Methods Metric Learning34 Learn a metric in input space Specialized to one/few-shot classification(Omniglot, MiniImageNet etc) Cannot use in other problems (e.g. RL) 3 Gregory Koch, Richard Zemel, and Ruslan Salakhutdinov. “Siamese Neural Networks for One-shot Image Recognition”. In: ICML (2015). 4 Oriol Vinyals et al. “Matching Networks for One Shot Learning”. In: NIPS (2016).
  • 10. Previous Deep Meta-Learning Methods Optimizer Learning
  • 11. Previous Deep Meta-Learning Methods Optimizer Learning56 Learn parameter update given gradients (search space includes SGD, RMSProp, Adam etc) Applicable to any architecture/task Best performance on Omniglot, MiniImageNet 5 Sachin Ravi and Hugo Larochelle. “Optimization as a Model for Few-shot Learning”. In: ICLR (2017). 6 Marcin Andrychowicz et al. “Learning to learn by gradient descent by gradient descent”. In: NIPS (2016).
  • 12. Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks Chelsea Finn Pieter Abbeel, Sergey Levine
  • 13. ImageNet pre-training Pretrain network on ImageNet classification, then fine-tune network on new task. Enables NNs to learn new vision tasks using relatively small datasets. This works because we have a huge labelled image dataset and the manifold of images have a somewhat consistent structure even between different datasets and tasks. How do we bring ’Initialization as meta-learning’ to non-vision domains such as speech/NLP/RL?
  • 17. Model-Agnostic Meta-Learning Few-shot Sine wave regression experiments
  • 19. Model-Agnostic Meta-Learning Reinforcement Learning Score function estimator in lines 6/10, since we cannot backprop through environment dynamics
  • 21. Model-Agnostic Meta-Learning Extension7 7 Zhenguo Li et al. “Meta-SGD: Learning to Learn Quickly for Few Shot Learning”. In: (2017).
  • 22. Discussion Parameter space noise (as opposed to policy space noise) has been shown to result in more consistent exploration89. This supports MAML’s idea. Why does MAML not overfit when taking multiple gradient steps? Do we need to overwrite all weights during adaptation? 8 Matthias Plappert et al. “Parameter Space Noise for Exploration”. In: (2017). 9 Meire Fortunato et al. “Noisy Networks for Exploration”. In: (2017).
  • 23. References I [1] Marcin Andrychowicz et al. “Learning to learn by gradient descent by gradient descent”. In: NIPS (2016). [2] Yan Duan et al. “RLˆ2: Fast Reinforcement Learning via Slow Reinforcement Learning”. In: (2016). [3] Chelsea Finn, Pieter Abbeel, and Sergey Levine. “Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks”. In: (2017). [4] Meire Fortunato et al. “Noisy Networks for Exploration”. In: (2017). [5] Gregory Koch, Richard Zemel, and Ruslan Salakhutdinov. “Siamese Neural Networks for One-shot Image Recognition”. In: ICML (2015). [6] Zhenguo Li et al. “Meta-SGD: Learning to Learn Quickly for Few Shot Learning”. In: (2017).
  • 24. References II [7] Nikhil Mishra, Mostafa Rohaninejad, and Xi UC Chen Pieter Abbeel Berkeley. “Meta-Learning with Temporal Convolutions”. In: (2017). [8] Matthias Plappert et al. “Parameter Space Noise for Exploration”. In: (2017). [9] Sachin Ravi and Hugo Larochelle. “Optimization as a Model for Few-shot Learning”. In: ICLR (2017). [10] Adam Santoro et al. “One-shot Learning with Memory-Augmented Neural Networks”. In: ICML (2016). [11] Oriol Vinyals et al. “Matching Networks for One Shot Learning”. In: NIPS (2016). [12] Jx Wang et al. “Learning to Reinforcement Learn”. In: (2016).