iTAML: An Incremental Task-Agnostic Meta-learning Approach

@article{Rajasegaran2020iTAMLAI,
  title={iTAML: An Incremental Task-Agnostic Meta-learning Approach},
  author={Jathushan Rajasegaran and Salman Khan and Munawar Hayat and F. Khan and M. Shah},
  journal={2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2020},
  pages={13585-13594}
}
Humans can continuously learn new knowledge as their experience grows. In contrast, previous learning in deep neural networks can quickly fade out when they are trained on a new task. In this paper, we hypothesize this problem can be avoided by learning a set of generalized parameters, that are neither specific to old nor new tasks. In this pursuit, we introduce a novel meta-learning approach that seeks to maintain an equilibrium between all the encountered tasks. This is ensured by a new meta… Expand
18 Citations
BI-MAML: Balanced Incremental Approach for Meta Learning
  • PDF
Generalising via Meta-Examples for Continual Learning in the Wild
  • PDF
Class-incremental learning: survey and performance evaluation
  • 4
  • PDF
GDumb: A Simple Approach that Questions Our Progress in Continual Learning
  • 23
  • PDF
Meta-Learned Attribute Self-Gating for Continual Generalized Zero-Shot Learning
  • 1
  • Highly Influenced
  • PDF
Few-Shot Unsupervised Continual Learning through Meta-Examples
  • 2
  • PDF
An Appraisal of Incremental Learning Methods
  • 1
  • PDF
...
1
2
...

References

SHOWING 1-10 OF 36 REFERENCES
Task Agnostic Meta-Learning for Few-Shot Learning
  • 84
  • PDF
Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
  • 3,036
  • PDF
Meta-Learning Representations for Continual Learning
  • 70
  • Highly Influential
  • PDF
On First-Order Meta-Learning Algorithms
  • 539
  • Highly Influential
  • PDF
Class-incremental Learning via Deep Model Consolidation
  • 52
  • PDF
Overcoming Catastrophic Forgetting by Incremental Moment Matching
  • 265
  • PDF
Learning without Forgetting
  • Zhizhong Li, Derek Hoiem
  • Computer Science, Mathematics
  • IEEE Transactions on Pattern Analysis and Machine Intelligence
  • 2018
  • 938
  • PDF
End-to-End Incremental Learning
  • 211
  • PDF
Overcoming catastrophic forgetting in neural networks
  • 1,597
  • Highly Influential
  • PDF
...
1
2
3
4
...