Deconstructed Generation-Based Zero-Shot Model
Dubing Chen, Yuming Shen, Haofeng Zhang, Philip H.S. Torr
Abstract
Recent research on Generalized Zero-Shot Learning (GZSL) has focused primarily on generation-based methods. However, current literature has overlooked the fundamental principles of these methods and has made limited progress in a complex manner. In this paper, we aim to deconstruct the generator-classifier framework and provide guidance for its improvement and extension. We begin by breaking down the generator-learned unseen class distribution into class-level and instance-level distributions. Through our analysis of the role of these two types of distributions in solving the GZSL problem, we generalize the focus of the generation-based approach, emphasizing the importance of (i) attribute generalization in generator learning and (ii) independent classifier learning with partially biased data. We present a simple method based on this analysis that outperforms SotAs on four public GZSL datasets, demonstrating the validity of our deconstruction. Furthermore, our proposed method remains effective even without a generative model, representing a step towards simplifying the generator-classifier structure. Our code is available at https://github.com/cdb342/DGZ.
BibTeX
@article{Chen_Shen_Zhang_Torr_2023, title={Deconstructed Generation-Based Zero-Shot Model}, volume={37}, url={https://ojs.aaai.org/index.php/AAAI/article/view/25102}, DOI={10.1609/aaai.v37i1.25102}, abstractNote={Recent research on Generalized Zero-Shot Learning (GZSL) has focused primarily on generation-based methods. However, current literature has overlooked the fundamental principles of these methods and has made limited progress in a complex manner. In this paper, we aim to deconstruct the generator-classifier framework and provide guidance for its improvement and extension. We begin by breaking down the generator-learned unseen class distribution into class-level and instance-level distributions. Through our analysis of the role of these two types of distributions in solving the GZSL problem, we generalize the focus of the generation-based approach, emphasizing the importance of (i) attribute generalization in generator learning and (ii) independent classifier learning with partially biased data. We present a simple method based on this analysis that outperforms SotAs on four public GZSL datasets, demonstrating the validity of our deconstruction. Furthermore, our proposed method remains effective even without a generative model, representing a step towards simplifying the generator-classifier structure. Our code is available at https://github.com/cdb342/DGZ.}, number={1}, journal={Proceedings of the AAAI Conference on Artificial Intelligence}, author={Chen, Dubing and Shen, Yuming and Zhang, Haofeng and Torr, Philip H.S.}, year={2023}, month={Jun.}, pages={295-303} }