MetaModulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Tasks

Wenfang Sun, Yingjun Du, Xiantong Zhen, Fan Wang, Ling Wang, Cees G. M. Snoek
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:32847-32858, 2023.

Abstract

Meta-learning algorithms are able to learn a new task using previously learned knowledge, but they often require a large number of meta-training tasks which may not be readily available. To address this issue, we propose a method for few-shot learning with fewer tasks, which we call MetaModulation. The key idea is to use a neural network to increase the density of the meta-training tasks by modulating batch normalization parameters during meta-training. Additionally, we modify parameters at various neural network levels, rather than just a single layer, to increase task diversity. To account for the uncertainty caused by the reduced number of training tasks, we propose a variational MetaModulation where the modulation parameters are treated as latent variables. We also introduce learning variational feature hierarchies by the variational MetaModulation, which modulates features at all layers and can take into account task uncertainty and generate more diverse tasks. The ablation studies illustrate the advantages of utilizing a learnable task modulation at different levels and demonstrate the benefit of incorporating probabilistic variants in few-task meta-learning. Our MetaModulation and its variational variants consistently outperform state-of-the-art alternatives on four few-task meta-learning benchmarks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-sun23b, title = {{M}eta{M}odulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Tasks}, author = {Sun, Wenfang and Du, Yingjun and Zhen, Xiantong and Wang, Fan and Wang, Ling and Snoek, Cees G. M.}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {32847--32858}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/sun23b/sun23b.pdf}, url = {https://proceedings.mlr.press/v202/sun23b.html}, abstract = {Meta-learning algorithms are able to learn a new task using previously learned knowledge, but they often require a large number of meta-training tasks which may not be readily available. To address this issue, we propose a method for few-shot learning with fewer tasks, which we call MetaModulation. The key idea is to use a neural network to increase the density of the meta-training tasks by modulating batch normalization parameters during meta-training. Additionally, we modify parameters at various neural network levels, rather than just a single layer, to increase task diversity. To account for the uncertainty caused by the reduced number of training tasks, we propose a variational MetaModulation where the modulation parameters are treated as latent variables. We also introduce learning variational feature hierarchies by the variational MetaModulation, which modulates features at all layers and can take into account task uncertainty and generate more diverse tasks. The ablation studies illustrate the advantages of utilizing a learnable task modulation at different levels and demonstrate the benefit of incorporating probabilistic variants in few-task meta-learning. Our MetaModulation and its variational variants consistently outperform state-of-the-art alternatives on four few-task meta-learning benchmarks.} }
Endnote
%0 Conference Paper %T MetaModulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Tasks %A Wenfang Sun %A Yingjun Du %A Xiantong Zhen %A Fan Wang %A Ling Wang %A Cees G. M. Snoek %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-sun23b %I PMLR %P 32847--32858 %U https://proceedings.mlr.press/v202/sun23b.html %V 202 %X Meta-learning algorithms are able to learn a new task using previously learned knowledge, but they often require a large number of meta-training tasks which may not be readily available. To address this issue, we propose a method for few-shot learning with fewer tasks, which we call MetaModulation. The key idea is to use a neural network to increase the density of the meta-training tasks by modulating batch normalization parameters during meta-training. Additionally, we modify parameters at various neural network levels, rather than just a single layer, to increase task diversity. To account for the uncertainty caused by the reduced number of training tasks, we propose a variational MetaModulation where the modulation parameters are treated as latent variables. We also introduce learning variational feature hierarchies by the variational MetaModulation, which modulates features at all layers and can take into account task uncertainty and generate more diverse tasks. The ablation studies illustrate the advantages of utilizing a learnable task modulation at different levels and demonstrate the benefit of incorporating probabilistic variants in few-task meta-learning. Our MetaModulation and its variational variants consistently outperform state-of-the-art alternatives on four few-task meta-learning benchmarks.
APA
Sun, W., Du, Y., Zhen, X., Wang, F., Wang, L. & Snoek, C.G.M.. (2023). MetaModulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Tasks. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:32847-32858 Available from https://proceedings.mlr.press/v202/sun23b.html.

Related Material