Adaptive IMLE for Few-shot Pretraining-free Generative Modelling

Mehran Aghabozorgi, Shichong Peng, Ke Li
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:248-264, 2023.

Abstract

Despite their success on large datasets, GANs have been difficult to apply in the few-shot setting, where only a limited number of training examples are provided. Due to mode collapse, GANs tend to ignore some training examples, causing overfitting to a subset of the training dataset, which is small in the first place. A recent method called Implicit Maximum Likelihood Estimation (IMLE) is an alternative to GAN that tries to address this issue. It uses the same kind of generators as GANs but trains it with a different objective that encourages mode coverage. However, the theoretical guarantees of IMLE hold under a restrictive condition that the optimal likelihood at all data points is the same. In this paper, we present a more generalized formulation of IMLE which includes the original formulation as a special case, and we prove that the theoretical guarantees hold under weaker conditions. Using this generalized formulation, we further derive a new algorithm, which we dub Adaptive IMLE, which can adapt to the varying difficulty of different training examples. We demonstrate on multiple few-shot image synthesis datasets that our method significantly outperforms existing methods. Our code is available at https://github.com/mehranagh20/AdaIMLE.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-aghabozorgi23a, title = {Adaptive {IMLE} for Few-shot Pretraining-free Generative Modelling}, author = {Aghabozorgi, Mehran and Peng, Shichong and Li, Ke}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {248--264}, 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/aghabozorgi23a/aghabozorgi23a.pdf}, url = {https://proceedings.mlr.press/v202/aghabozorgi23a.html}, abstract = {Despite their success on large datasets, GANs have been difficult to apply in the few-shot setting, where only a limited number of training examples are provided. Due to mode collapse, GANs tend to ignore some training examples, causing overfitting to a subset of the training dataset, which is small in the first place. A recent method called Implicit Maximum Likelihood Estimation (IMLE) is an alternative to GAN that tries to address this issue. It uses the same kind of generators as GANs but trains it with a different objective that encourages mode coverage. However, the theoretical guarantees of IMLE hold under a restrictive condition that the optimal likelihood at all data points is the same. In this paper, we present a more generalized formulation of IMLE which includes the original formulation as a special case, and we prove that the theoretical guarantees hold under weaker conditions. Using this generalized formulation, we further derive a new algorithm, which we dub Adaptive IMLE, which can adapt to the varying difficulty of different training examples. We demonstrate on multiple few-shot image synthesis datasets that our method significantly outperforms existing methods. Our code is available at https://github.com/mehranagh20/AdaIMLE.} }
Endnote
%0 Conference Paper %T Adaptive IMLE for Few-shot Pretraining-free Generative Modelling %A Mehran Aghabozorgi %A Shichong Peng %A Ke Li %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-aghabozorgi23a %I PMLR %P 248--264 %U https://proceedings.mlr.press/v202/aghabozorgi23a.html %V 202 %X Despite their success on large datasets, GANs have been difficult to apply in the few-shot setting, where only a limited number of training examples are provided. Due to mode collapse, GANs tend to ignore some training examples, causing overfitting to a subset of the training dataset, which is small in the first place. A recent method called Implicit Maximum Likelihood Estimation (IMLE) is an alternative to GAN that tries to address this issue. It uses the same kind of generators as GANs but trains it with a different objective that encourages mode coverage. However, the theoretical guarantees of IMLE hold under a restrictive condition that the optimal likelihood at all data points is the same. In this paper, we present a more generalized formulation of IMLE which includes the original formulation as a special case, and we prove that the theoretical guarantees hold under weaker conditions. Using this generalized formulation, we further derive a new algorithm, which we dub Adaptive IMLE, which can adapt to the varying difficulty of different training examples. We demonstrate on multiple few-shot image synthesis datasets that our method significantly outperforms existing methods. Our code is available at https://github.com/mehranagh20/AdaIMLE.
APA
Aghabozorgi, M., Peng, S. & Li, K.. (2023). Adaptive IMLE for Few-shot Pretraining-free Generative Modelling. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:248-264 Available from https://proceedings.mlr.press/v202/aghabozorgi23a.html.

Related Material