Recasting Gradient-Based Meta-Learning as Hierarchical Bayes

Abstract

Meta-learning allows an intelligent agent to leverage prior learning episodes as a basis for quickly improving performance on a novel task. Bayesian hierarchical modeling provides a theoretical framework for formalizing meta-learning as inference for a set of parameters that are shared across tasks. Here, we reformulate the model-agnostic meta-learning algorithm (MAML) of Finn et al. (2017) as a method for probabilistic inference in a hierarchical Bayesian model. In contrast to prior methods for meta-learning via hierarchical Bayes, MAML is naturally applicable to complex function approximators through its use of a scalable gradient descent procedure for posterior inference. Furthermore, the identification of MAML as hierarchical Bayes provides a way to understand the algorithm's operation as a meta-learning procedure, as well as an opportunity to make use of computational strategies for efficient inference. We use this opportunity to propose an improvement to the MAML algorithm that makes use of techniques from approximate inference and curvature estimation.

References in corpus (3)

• Theoretical Models of Learning to Learn
• Bayesian Multitask Learning with Latent Hierarchies
• Few-Shot Learning Through an Information Retrieval Lens

Cited by in corpus (25)

• Empirical Bayes Transductive Meta-Learning with Synthetic Gradients
• Improving Generalization in Meta Reinforcement Learning using Learned Objectives
• Deep Learning Theory Review: An Optimal Control and Dynamical Systems Perspective
• Reward Shaping via Meta-Learning
• Meta-learning of Sequential Strategies
• Meta-Graph: Few Shot Link Prediction via Meta Learning
• Efficiently Identifying Task Groupings for Multi-Task Learning
• Warm Up Cold-start Advertisements: Improving CTR Predictions via Learning to Learn ID Embeddings
• Defining Benchmarks for Continual Few-Shot Learning
• Toward Multimodal Model-Agnostic Meta-Learning
• Meta-Learning surrogate models for sequential decision making
• Meta Dialogue Policy Learning
• Robust Meta-learning for Mixed Linear Regression with Small Batches
• Compositional Few-Shot Recognition with Primitive Discovery and Enhancing
• PAC-Bayes Bounds for Meta-learning with Data-Dependent Prior
• Non-Gaussian Gaussian Processes for Few-Shot Regression
• Information-Theoretic Analysis of Epistemic Uncertainty in Bayesian Meta-learning
• ST-MAML: A Stochastic-Task based Method for Task-Heterogeneous Meta-Learning
• Meta-learning for mixed linear regression
• An Optimization-Based Meta-Learning Model for MRI Reconstruction with Diverse Dataset
• Few Shot Learning With No Labels
• Quantifying Visual Image Quality: A Bayesian View
• Meta Learning in the Continuous Time Limit
• Knowledge as Invariance -- History and Perspectives of Knowledge-augmented Machine Learning
• Meta-Learning to Improve Pre-Training