• Corpus ID: 168169908

Meta-Learning Representations for Continual Learning

@inproceedings{Javed2019MetaLearningRF,
  title={Meta-Learning Representations for Continual Learning},
  author={Khurram Javed and Martha White},
  booktitle={NeurIPS},
  year={2019}
}
A continual learning agent should be able to build on top of existing knowledge to learn on new data quickly while minimizing forgetting. [] Key Method Moreover, our method is complementary to existing continual learning strategies, like MER, which can learn more effectively from representations learned by our objective. Finally, we demonstrate that a basic online updating strategy with our learned representation is competitive with rehearsal based methods for continual learning. We release an implementation…

Figures and Tables from this paper

Self-Attention Meta-Learner for Continual Learning
TLDR
This paper proposes a new method, named Self-Attention MetaLearner (SAM) which learns a prior knowledge for continual learning that permits learning a sequence of tasks, while avoiding catastrophic forgetting, and empirically shows that it can achieve a better performance than several state-of-the-art methods by building on top of the selected representation learned by SAM.
Learning to Remember from a Multi-Task Teacher
TLDR
This paper argues that the outputs of neural networks are subject to rapid changes when learning a new data distribution, and networks that appear to "forget" everything still contain useful representation towards previous tasks, and proposes a novel meta-learning algorithm to overcome this issue.
Online Fast Adaptation and Knowledge Accumulation: a New Approach to Continual Learning
TLDR
It is empirically shown that Continual-MAML, an online extension of the popular MAML algorithm, is better suited to the new scenario of OSAKA, where an agent must quickly solve new (out-of-distribution) tasks, while also requiring fast remembering.
A Meta-Learned Neuron model for Continual Learning
TLDR
This work replaces the standard neuron by a metalearned neuron model whom inference and update rules are optimized to minimize catastrophic interference, and can memorize dataset-length sequences of training samples, and its learning capabilities generalize to any domain.
La-MAML: Look-ahead Meta Learning for Continual Learning
TLDR
This work proposes Look-ahead MAML (La-MAML), a fast optimisation-based meta-learning algorithm for online-continual learning, aided by a small episodic memory, and proposed modulation of per-parameter learning rates in this update provides a more flexible and efficient way to mitigate catastrophic forgetting compared to conventional prior-based methods.
Online Fast Adaptation and Knowledge Accumulation (OSAKA): a New Approach to Continual Learning
TLDR
It is shown in an empirical study that ContinualMAML, an online extension of the popular MAML algorithm, is better suited to the new scenario than the aforementioned methodologies including standard continual learning and meta-learning approaches.
Learning to Continually Learn
TLDR
A Neuromodulated Meta-Learning Algorithm (ANML) enables continual learning without catastrophic forgetting at scale: it produces state-of-the-art continual learning performance, sequentially learning as many as 600 classes (over 9,000 SGD updates).
Prototypes-Guided Memory Replay for Continual Learning
TLDR
A dynamic prototypesguided memory replay module is devised, incorporating it into an online meta-learning model, and the experimental results testify the superiority of the method in alleviating catastrophic forgetting and enabling efficient knowledge transfer.
Generative vs. Discriminative: Rethinking The Meta-Continual Learning
TLDR
This work develops a generative classifier that efficiently uses data-driven experience to learn new concepts even from few samples while being immune to forgetting, suggesting that generative models provide a natural way for alleviating catastrophic forgetting, which is a major drawback of discriminative models.
iTAML: An Incremental Task-Agnostic Meta-learning Approach
TLDR
A novel meta-learning approach that seeks to maintain an equilibrium between all the encountered tasks, ensured by a new meta-update rule which avoids catastrophic forgetting and is task-agnostic.
...
...

References

SHOWING 1-10 OF 34 REFERENCES
Deep Online Learning via Meta-Learning: Continual Adaptation for Model-Based RL
TLDR
This work develops a method for continual online learning from an incoming stream of data, using deep neural network models, and demonstrates that MOLe outperforms alternative prior methods, and enables effective continuous adaptation in non-stationary task distributions such as varying terrains, motor failures, and unexpected disturbances.
Continual Learning with Tiny Episodic Memories
TLDR
It is observed that a very simple baseline, which jointly trains on both examples from the current task as well as examples stored in the memory, outperforms state-of-the-art CL approaches with and without episodic memory.
On Tiny Episodic Memories in Continual Learning
TLDR
This work empirically analyze the effectiveness of a very small episodic memory in a CL setup where each training example is only seen once and finds that repetitive training on even tiny memories of past tasks does not harm generalization, on the contrary, it improves it.
Gradient based sample selection for online continual learning
TLDR
This work formulation of sample selection as a constraint reduction problem based on the constrained optimization view of continual learning shows that it is equivalent to maximizing the diversity of samples in the replay buffer with parameters gradient as the feature.
Gradient Episodic Memory for Continual Learning
TLDR
A model for continual learning, called Gradient Episodic Memory (GEM) is proposed that alleviates forgetting, while allowing beneficial transfer of knowledge to previous tasks.
Overcoming catastrophic forgetting in neural networks
TLDR
It is shown that it is possible to overcome the limitation of connectionist models and train networks that can maintain expertise on tasks that they have not experienced for a long time and selectively slowing down learning on the weights important for previous tasks.
Memory Aware Synapses: Learning what (not) to forget
TLDR
This paper argues that, given the limited model capacity and the unlimited new information to be learned, knowledge has to be preserved or erased selectively and proposes a novel approach for lifelong learning, coined Memory Aware Synapses (MAS), which computes the importance of the parameters of a neural network in an unsupervised and online manner.
Learning to Learn with Gradients
TLDR
This thesis discusses gradient-based algorithms for learning to learn, or meta-learning, which aim to endow machines with flexibility akin to that of humans, and shows how these methods can be extended for applications in motor control by combining elements of meta- learning with techniques for deep model-based reinforcement learning, imitation learning, and inverse reinforcement learning.
Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning
Meta-Learning Update Rules for Unsupervised Representation Learning
TLDR
This work target semi-supervised classification performance, and meta-learn an algorithm -- an unsupervised weight update rule -- that produces representations useful for this task that is constrain to be a biologically-motivated, neuron-local function which enables it to generalize to different neural network architectures, datasets, and data modalities.
...
...