Theoretical Models of Learning to Learn

Abstract

A Machine can only learn if it is biased in some way. Typically the bias is supplied by hand, for example through the choice of an appropriate set of features. However, if the learning machine is embedded within an {\em environment} of related tasks, then it can {\em learn} its own bias by learning sufficiently many tasks from the environment. In this paper two models of bias learning (or equivalently, learning to learn) are introduced and the main theoretical results presented. The first model is a PAC-type model based on empirical process theory, while the second is a hierarchical Bayes model.

arXiv admin note: text overlap with arXiv:1106.0245

Cited by in corpus (256)

• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Neural Architecture Search with Reinforcement Learning
• Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
• Searching for Activation Functions
• Learning to learn by gradient descent by gradient descent
• Massively Multilingual Neural Machine Translation in the Wild: Findings and Challenges
• One-Shot Visual Imitation Learning via Meta-Learning
• Learning Task Grouping and Overlap in Multi-task Learning
• A Survey of Human Activity Recognition in Smart Homes Based on IoT Sensors Algorithms: Taxonomies, Challenges, and Opportunities with Deep Learning
• Low-Rank Pairwise Alignment Bilinear Network For Few-Shot Fine-Grained Image Classification
• Recasting Gradient-Based Meta-Learning as Hierarchical Bayes
• AutoDebias: Learning to Debias for Recommendation
• Learning to Optimize
• Curriculum Learning by Transfer Learning: Theory and Experiments with Deep Networks
• Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments
• Set Transformer: A Framework for Attention-based Permutation-Invariant Neural Networks
• Meta-learning with differentiable closed-form solvers
• Entanglement Devised Barren Plateau Mitigation
• AdaShare: Learning What To Share For Efficient Deep Multi-Task Learning
• An improved model-independent assessment of the late-time cosmic expansion
• Born to Learn: the Inspiration, Progress, and Future of Evolved Plastic Artificial Neural Networks
• Learning to Generalize: Meta-Learning for Domain Generalization
• One-Shot Imitation from Observing Humans via Domain-Adaptive Meta-Learning
• Learned Optimizers that Scale and Generalize
• Bilevel Programming for Hyperparameter Optimization and Meta-Learning
• Sample Efficient Adaptive Text-to-Speech
• The Benefit of Multitask Representation Learning
• Bottom-up and top-down approaches for the design of neuromorphic processing systems: Tradeoffs and synergies between natural and artificial intelligence
• Multitask learning over graphs: An Approach for Distributed, Streaming Machine Learning
• RS-MetaNet: Deep meta metric learning for few-shot remote sensing scene classification
• Learning to design from humans: Imitating human designers through deep learning
• Empirical Bayes Transductive Meta-Learning with Synthetic Gradients
• Analyzing Human-Human Interactions: A Survey
• Meta-Learning without Memorization
• Deep Representation Learning in Speech Processing: Challenges, Recent Advances, and Future Trends
• Sharing Knowledge in Multi-Task Deep Reinforcement Learning
• CrossTransformers: spatially-aware few-shot transfer
• Sparse coding for multitask and transfer learning
• Small Sample Learning in Big Data Era
• Automated Pruning for Deep Neural Network Compression
• Unsupervised Meta-Learning for Reinforcement Learning
• Alpha MAML: Adaptive Model-Agnostic Meta-Learning
• Learning Gradient Descent: Better Generalization and Longer Horizons
• Fast Rates by Transferring from Auxiliary Hypotheses
• Deep Online Learning via Meta-Learning: Continual Adaptation for Model-Based RL
• A Survey of Optimization Methods from a Machine Learning Perspective
• Few-shot Autoregressive Density Estimation: Towards Learning to Learn Distributions
• Meta-learning within Projective Simulation
• Learning from Few Samples: A Survey
• Transductive Episodic-Wise Adaptive Metric for Few-Shot Learning
• Few-Shot Goal Inference for Visuomotor Learning and Planning
• La-MAML: Look-ahead Meta Learning for Continual Learning
• Provable Guarantees for Gradient-Based Meta-Learning
• If MaxEnt RL is the Answer, What is the Question?
• Leveraging the Feature Distribution in Transfer-based Few-Shot Learning
• TaskNorm: Rethinking Batch Normalization for Meta-Learning
• Theoretical Convergence of Multi-Step Model-Agnostic Meta-Learning
• Modular meta-learning
• Meta-Graph: Few Shot Link Prediction via Meta Learning
• Client-server multi-task learning from distributed datasets
• Rewriting History with Inverse RL: Hindsight Inference for Policy Improvement
• Meta-Learned Confidence for Few-shot Learning
• Transfer Learning for Speech and Language Processing
• Sharing to learn and learning to share; Fitting together Meta-Learning, Multi-Task Learning, and Transfer Learning: A meta review
• Adversarial Meta-Learning
• Learning Output Kernels for Multi-Task Problems
• Neural Relational Inference with Fast Modular Meta-learning
• Convergence of Meta-Learning with Task-Specific Adaptation over Partial Parameters
• The Threat of Adversarial Attacks on Machine Learning in Network Security -- A Survey
• Task-Robust Model-Agnostic Meta-Learning
• Consistent Multitask Learning with Nonlinear Output Relations
• Learning to Warm-Start Bayesian Hyperparameter Optimization
• Meta-Learning Symmetries by Reparameterization
• Adaptive Gradient-Based Meta-Learning Methods
• Meta Transition Adaptation for Robust Deep Learning with Noisy Labels
• End-to-End Multi-Task Learning with Attention
• Meta-Learning Deep Energy-Based Memory Models
• Information-Theoretic Generalization Bounds for Meta-Learning and Applications
• Probing transfer learning with a model of synthetic correlated datasets
• Meta-Learning Mean Functions for Gaussian Processes
• Meta-Learning GNN Initializations for Low-Resource Molecular Property Prediction
• Learning Radio Resource Management in 5G Networks: Framework, Opportunities and Challenges
• Transfer Metric Learning: Algorithms, Applications and Outlooks
• Learning to Balance: Bayesian Meta-Learning for Imbalanced and Out-of-distribution Tasks
• Excess risk bounds for multitask learning with trace norm regularization
• Augmented Hard Example Mining for Generalizable Person Re-Identification
• On Fast Adversarial Robustness Adaptation in Model-Agnostic Meta-Learning
• CovidExpert: A Triplet Siamese Neural Network framework for the detection of COVID-19
• Learning to Learn Variational Semantic Memory
• Differentiable plasticity: training plastic neural networks with backpropagation
• Unsupervised Representation Learning with Future Observation Prediction for Speech Emotion Recognition
• Effective Cross-lingual Transfer of Neural Machine Translation Models without Shared Vocabularies
• Meta-Learning Stationary Stochastic Process Prediction with Convolutional Neural Processes
• Beyond backpropagation: bilevel optimization through implicit differentiation and equilibrium propagation
• Revisiting Meta-Learning as Supervised Learning
• Transfer Learning for Estimating Causal Effects using Neural Networks
• How to Train Your MAML to Excel in Few-Shot Classification
• Watch, Try, Learn: Meta-Learning from Demonstrations and Reward
• Learning Adaptive Loss for Robust Learning with Noisy Labels
• Evaluating Logical Generalization in Graph Neural Networks
• Universal linguistic inductive biases via meta-learning
• Model-based Adversarial Meta-Reinforcement Learning
• Meta-Learning with Sparse Experience Replay for Lifelong Language Learning
• Few-Shot Action Recognition with Compromised Metric via Optimal Transport
• Learning Tensors in Reproducing Kernel Hilbert Spaces with Multilinear Spectral Penalties
• Quantitative toxicity prediction using topology based multi-task deep neural networks
• Gradient Matching for Domain Generalization
• Biased Stochastic First-Order Methods for Conditional Stochastic Optimization and Applications in Meta Learning
• Few-shot Classification via Adaptive Attention
• Gradient-free Policy Architecture Search and Adaptation
• Improving Generalization in Meta-learning via Task Augmentation
• Learning to Transfer
• Sentence Embedding Alignment for Lifelong Relation Extraction
• Meta-learning autoencoders for few-shot prediction
• Continual Human-in-the-Loop Optimization
• Graph Meta Learning via Local Subgraphs
• Learning to Learn with Variational Information Bottleneck for Domain Generalization
• Non-Iterative Knowledge Fusion in Deep Convolutional Neural Networks
• Meta-Learning Acquisition Functions for Transfer Learning in Bayesian Optimization
• Generalizing from a few environments in safety-critical reinforcement learning
• Continuous Meta-Learning without Tasks
• MetaCURE: Meta Reinforcement Learning with Empowerment-Driven Exploration
• Fully-adaptive Feature Sharing in Multi-Task Networks with Applications in Person Attribute Classification
• Transfer learning approach for financial applications
• Towards Effective Context for Meta-Reinforcement Learning: an Approach based on Contrastive Learning
• Learning a Prior over Intent via Meta-Inverse Reinforcement Learning
• Structure Learning in Motor Control:A Deep Reinforcement Learning Model
• A Theoretically Sound Upper Bound on the Triplet Loss for Improving the Efficiency of Deep Distance Metric Learning
• Graph-based Interpolation of Feature Vectors for Accurate Few-Shot Classification
• Learning to Learn Single Domain Generalization
• Is Support Set Diversity Necessary for Meta-Learning?
• Decoupling Exploration and Exploitation for Meta-Reinforcement Learning without Sacrifices
• Continuous Coordination As a Realistic Scenario for Lifelong Learning
• A survey of automatic de-identification of longitudinal clinical narratives
• PARN: Position-Aware Relation Networks for Few-Shot Learning
• Meta-learning algorithms for Few-Shot Computer Vision
• Nonstationary Nonparametric Online Learning: Balancing Dynamic Regret and Model Parsimony
• Task Selection Policies for Multitask Learning
• Robust Meta-learning for Mixed Linear Regression with Small Batches
• Multi-Task Learning with Language Modeling for Question Generation
• A Sample Complexity Separation between Non-Convex and Convex Meta-Learning
• Disentangled Cumulants Help Successor Representations Transfer to New Tasks
• Migrating Knowledge between Physical Scenarios based on Artificial Neural Networks
• MetaAlign: Coordinating Domain Alignment and Classification for Unsupervised Domain Adaptation
• Online Meta-Learning for Multi-Source and Semi-Supervised Domain Adaptation
• MxML: Mixture of Meta-Learners for Few-Shot Classification
• A Model-based Approach for Sample-efficient Multi-task Reinforcement Learning
• How can AI Automate End-to-End Data Science?
• How Fine-Tuning Allows for Effective Meta-Learning
• $k$Folden: $k$-Fold Ensemble for Out-Of-Distribution Detection
• Tailoring: encoding inductive biases by optimizing unsupervised objectives at prediction time
• Task-Adaptive Neural Network Search with Meta-Contrastive Learning
• Incremental Few-Shot Learning for Pedestrian Attribute Recognition
• Meta-strategy for Learning Tuning Parameters with Guarantees
• Text Generation with Exemplar-based Adaptive Decoding
• Efficient AutoML Pipeline Search with Matrix and Tensor Factorization
• A Channel Coding Benchmark for Meta-Learning
• Modular Universal Reparameterization: Deep Multi-task Learning Across Diverse Domains
• Few-Shot Learning by Integrating Spatial and Frequency Representation
• Neural Architecture Search Over a Graph Search Space
• MetaSelector: Meta-Learning for Recommendation with User-Level Adaptive Model Selection
• Exploring Versatile Generative Language Model Via Parameter-Efficient Transfer Learning
• Q-DeckRec: A Fast Deck Recommendation System for Collectible Card Games
• Meta Dropout: Learning to Perturb Features for Generalization
• Interactive Learning from Activity Description
• ORBIT: A Real-World Few-Shot Dataset for Teachable Object Recognition
• Learning State Abstractions for Transfer in Continuous Control
• Online Hyperparameter Meta-Learning with Hypergradient Distillation
• Improving Few-Shot Learning with Auxiliary Self-Supervised Pretext Tasks
• Elements of Active Continuous Learning and Uncertainty Self-Awareness: a Narrow Implementation for Face and Facial Expression Recognition
• Uncertainty-guided Model Generalization to Unseen Domains
• Universum Prescription: Regularization using Unlabeled Data
• Bilevel Optimization for Machine Learning: Algorithm Design and Convergence Analysis
• Discriminative Few-Shot Learning Based on Directional Statistics
• Improving Surrogate Model Accuracy for the LCLS-II Injector Frontend Using Convolutional Neural Networks and Transfer Learning
• Disentangled Relational Representations for Explaining and Learning from Demonstration
• Unsupervised Meta-Learning through Latent-Space Interpolation in Generative Models
• Meta-Unsupervised-Learning: A supervised approach to unsupervised learning
• Hierarchical Expert Networks for Meta-Learning
• Continual Learning from the Perspective of Compression
• Review, Analysis and Design of a Comprehensive Deep Reinforcement Learning Framework
• Intelligence, physics and information -- the tradeoff between accuracy and simplicity in machine learning
• Meta Learning with Differentiable Closed-form Solver for Fast Video Object Segmentation
• Hierarchical Meta Learning
• Meta-Optimization of Deep CNN for Image Denoising Using LSTM
• Reinforced Few-Shot Acquisition Function Learning for Bayesian Optimization
• Gaussian Process Meta Few-shot Classifier Learning via Linear Discriminant Laplace Approximation
• An Optimization Framework for Semi-Supervised and Transfer Learning using Multiple Classifiers and Clusterers
• Lifelong Learning in Multi-Armed Bandits
• Continual Learning in Task-Oriented Dialogue Systems
• How Tight Can PAC-Bayes be in the Small Data Regime?
• Improving Context-Based Meta-Reinforcement Learning with Self-Supervised Trajectory Contrastive Learning
• Embracing the Disharmony in Medical Imaging: A Simple and Effective Framework for Domain Adaptation
• Population-Based Evolution Optimizes a Meta-Learning Objective
• The Traveling Observer Model: Multi-task Learning Through Spatial Variable Embeddings
• Learning the Step-size Policy for the Limited-Memory Broyden-Fletcher-Goldfarb-Shanno Algorithm
• Adversarial Imitation Learning with Trajectorial Augmentation and Correction
• Incremental Meta-Learning via Indirect Discriminant Alignment
• Learning to Learn and Predict: A Meta-Learning Approach for Multi-Label Classification
• Meta-Learning with Network Pruning
• Meta-learning for mixed linear regression
• Immuno-mimetic Deep Neural Networks (Immuno-Net)
• Large-Scale Meta-Learning with Continual Trajectory Shifting
• Constructing Complexity-efficient Features in XCS with Tree-based Rule Conditions
• Learning-to-learn non-convex piecewise-Lipschitz functions
• Modular meta-learning in abstract graph networks for combinatorial generalization
• Neural Mask Generator: Learning to Generate Adaptive Word Maskings for Language Model Adaptation
• AlphaNet: Improving Long-Tail Classification By Combining Classifiers
• Visual Transfer Learning: Informal Introduction and Literature Overview
• Combining Domain-Specific Meta-Learners in the Parameter Space for Cross-Domain Few-Shot Classification
• Explainability-aided Domain Generalization for Image Classification
• Meta-learning Transferable Representations with a Single Target Domain
• A partition-based similarity for classification distributions
• Local Nonparametric Meta-Learning
• Meta Variational Monte Carlo
• No-regret Non-convex Online Meta-Learning
• Meta-Learner with Linear Nulling
• Towards Recognizing New Semantic Concepts in New Visual Domains
• Universal Policies for Software-Defined MDPs
• The Joy of Neural Painting
• Meta-Learning with Variational Semantic Memory for Word Sense Disambiguation
• Unsupervised Domain Adaptation with Variational Approximation for Cardiac Segmentation
• Selecting for Selection: Learning To Balance Adaptive and Diversifying Pressures in Evolutionary Search
• Deep3DPose: Realtime Reconstruction of Arbitrarily Posed Human Bodies from Single RGB Images
• Margin-Based Transfer Bounds for Meta Learning with Deep Feature Embedding
• Personalizing Dialogue Agents via Meta-Learning
• Learning Efficient and Effective Exploration Policies with Counterfactual Meta Policy
• On Bi-gram Graph Attributes
• Learning to Transfer: A Foliated Theory
• Learning to Initialize Gradient Descent Using Gradient Descent
• Center Loss Regularization for Continual Learning
• Diverse Distributions of Self-Supervised Tasks for Meta-Learning in NLP
• Learning to Generalize Compositionally by Transferring Across Semantic Parsing Tasks
• Transfer Regression via Pairwise Similarity Regularization
• Reasoning about Unforeseen Possibilities During Policy Learning
• Domain Agnostic Learning for Unbiased Authentication
• Adaptive Submodular Meta-Learning
• Online Parameter-Free Learning of Multiple Low Variance Tasks
• Meta Learning in the Continuous Time Limit
• Don't Wait, Just Weight: Improving Unsupervised Representations by Learning Goal-Driven Instance Weights
• Generalized Reinforcement Meta Learning for Few-Shot Optimization
• RLCache: Automated Cache Management Using Reinforcement Learning
• Training few-shot classification via the perspective of minibatch and pretraining
• Beyond Categorical Label Representations for Image Classification
• Submodular Meta-Learning
• Dataset Bias in Few-shot Image Recognition
• Does the Adam Optimizer Exacerbate Catastrophic Forgetting?
• Generalization Bounds for Meta-Learning via PAC-Bayes and Uniform Stability
• On regret bounds for continual single-index learning
• Towards Enabling Meta-Learning from Target Models
• Meta Auxiliary Learning for Facial Action Unit Detection
• On Data Efficiency of Meta-learning
• Meta-Regularization by Enforcing Mutual-Exclusiveness
• Uniform Sampling over Episode Difficulty
• Meta Learning Low Rank Covariance Factors for Energy-Based Deterministic Uncertainty
• Meta-Learning with Adjoint Methods