How to join the virtual workshop: The 2020 Workshop on Meta-Learning will be a series of streamed pre-recorded talks + live question-and-answer (Q&A) periods, and poster sessions on Gather.Town. You can participate by:
* Accessing the livestream on our [ protected link dropped ] 2;
* MetaLearn 2020 Rocket.Chat!
* Entering panel discussion questions in this sli.do!
Focus of the workshop: Recent years have seen rapid progress in meta-learning methods, which transfer knowledge across tasks and domains to learn new tasks more efficiently, optimize the learning process itself, and even generate new learning methods from scratch. Meta-learning can be seen as the logical conclusion of the arc that machine learning has undergone in the last decade, from learning classifiers and policies over hand-crafted features, to learning representations over which classifiers and policies operate, and finally to learning algorithms that themselves acquire representations, classifiers, and policies. Meta-learning methods are also of substantial practical interest. For instance, they have been shown to yield new state-of-the-art automated machine learning algorithms and architectures, and have substantially improved few-shot learning systems. Moreover, the ability to improve one’s own learning capabilities through experience can also be viewed as a hallmark of intelligent beings, and there are strong connections with work on human learning in cognitive science and reward learning in neuroscience.
| Introduction and opening remarks (introduction) | |
| Introduction for invited speaker, Frank Hutter (remarks) | |
| Meta-learning neural architectures, initial weights, hyperparameters, and algorithm components (invited talk) | |
| Q/A for invited talk #1 (question period) | |
| On episodes, Prototypical Networks, and few-shot learning (contributed talk) | |
| Poster session #1 (poster session) | |
| Introduction for invited speaker, Luisa Zintgraf (remarks) | |
| Exploration in meta-reinforcement learning (invited talk) | |
| Q/A for invited talk #2 (question period) | |
| Introduction for invited speaker, Tim Hospedales (remarks) | |
| Meta-Learning: Representations and Objectives (invited talk) | |
| Q/A for invited talk #3 (question period) | |
| Break (break) | |
| Poster session #2 (poster session) | |
| Introduction for invited speaker, Louis Kirsch (remarks) | |
| General meta-learning (invited talk) | |
| Q/A for invited talk #4 (question period) | |
| Introduction for invited speaker, Fei-Fei Li (remarks) | |
| Creating diverse tasks to catalyze robot learning (invited talk) | |
| Q/A for invited talk #5 (question period) | |
| Poster session #3 (poster session) | |
| Introduction for invited speaker, Kate Rakelly (remarks) | |
| An inference perspective on meta-reinforcement learning (invited talk) | |
| Q/A for invited talk #6 (question period) | |
| Reverse engineering learned optimizers reveals known and novel mechanisms (contributed talk) | |
| Bayesian optimization by density ratio estimation (contributed talk) | |
| Panel discussion (discussion panel) | |
| Meta-Learning via Hypernetworks (Poster) | |
| Model-Agnostic Graph Regularization for Few-Shot Learning (Poster) | |
| MPLP: Learning a Message Passing Learning Protocol (Poster) | |
| Model-Based Meta-Reinforcement Learning for Flight with Suspended Payloads (Poster) | |
| Prototypical Region Proposal Networks for Few-shot Localization and Classification (Poster) | |
| Putting Theory to Work: From Learning Bounds to Meta-Learning Algorithms (Poster) | |
| Similarity of classification tasks (Poster) | |
| Tailoring: encoding inductive biases by optimizing unsupervised objectives at prediction time (Poster) | |
| Task Meta-Transfer from Limited Parallel Labels (Poster) | |
| Task Similarity Aware Meta Learning: Theory-inspired Improvement on MAML (Poster) | |
| Training more effective learned optimizers (Poster) | |
| Towards Meta-Algorithm Selection (Poster) | |
| Uniform Priors for Meta-Learning (Poster) | |
| Hyperparameter Transfer Across Developer Adjustments (Poster) | |
| Few-Shot Unsupervised Continual Learning through Meta-Examples (Poster) | |
| Synthetic Petri Dish: A Novel Surrogate Model for Rapid Architecture Search (Poster) | |
| Pareto-efficient Acquisition Functions for Cost-Aware Bayesian Optimization (Poster) | |
| Learning in Low Resource Modalities via Cross-Modal Generalization (Poster) | |
| How Important is the Train-Validation Split in Meta-Learning? (Poster) | |
| HyperVAE: Variational Hyper-Encoding Network (Poster) | |
| Learning not to learn: Nature versus nurture in silico (Poster) | |
| Meta-Learning Backpropagation And Improving It (Poster) | |
| Adaptive Risk Minimization: A Meta-Learning Approach for Tackling Group Shift (Poster) | |
| Continual Model-Based Reinforcement Learning with Hypernetworks (Poster) | |
| NAS-Bench-301 and the Case for Surrogate Benchmarks for Neural Architecture Search (Poster) | |
| Prior-guided Bayesian Optimization (Poster) | |
| Multi-Objective Multi-Fidelity Hyperparameter Optimization with Application to Fairness (Poster) | |
| Open-Set Incremental Learning via Bayesian Prototypical Embeddings (Poster) | |
| Meta-Learning Bayesian Neural Network Priors Based on PAC-Bayesian Theory (Poster) | |
| MobileDets: Searching for Object Detection Architecture for Mobile Accelerators (Poster) | |
| A Meta-Learning Approach for Graph Representation Learning in Multi-Task Settings (Poster) | |
| Contextual HyperNetworks for Novel Feature Adaptation (Poster) | |
| Continual learning with direction-constrained optimization (Poster) | |
| Data Augmentation for Meta-Learning (Poster) | |
| Decoupling Exploration and Exploitation in Meta-Reinforcement Learning without Sacrifices (Poster) | |
| Defining Benchmarks for Continual Few-Shot Learning (Poster) | |
| Exploring Representation Learning for Flexible Few-Shot Tasks (Poster) | |
| Few-shot Sequence Learning with Transformers (Poster) | |
| Flexible Dataset Distillation: Learn Labels Instead of Images (Poster) | |
| Is Support Set Diversity Necessary for Meta-Learning? (Poster) | |
| Learning Flexible Classifiers with Shot-CONditional Episodic (SCONE) Training (Poster) | |
| Learning to Generate Noise for Multi-Attack Robustness (Poster) | |
| MAster of PuPpets: Model-Agnostic Meta-Learning via Pre-trained Parameters for Natural Language Generation (Poster) | |
| Measuring few-shot extrapolation with program induction (Poster) | |
| Meta-Learning Initializations for Image Segmentation (Poster) | |
| Meta-Learning of Compositional Task Distributions in Humans and Machines (Poster) | |