GenZSL: Generative Zero-Shot Learning Via Inductive Variational Autoencoder
This work addresses zero-shot learning for AI systems by improving generative methods, though it is incremental as it builds on existing generative ZSL approaches.
The paper tackles the problem of suboptimal generative performance and limited scene generalization in zero-shot learning by proposing GenZSL, an inductive variational autoencoder that generates new class samples from similar seen classes using weak semantic vectors, resulting in 24.7% performance gains and over 60x faster training speed on the AWA2 dataset.
Remarkable progress in zero-shot learning (ZSL) has been achieved using generative models. However, existing generative ZSL methods merely generate (imagine) the visual features from scratch guided by the strong class semantic vectors annotated by experts, resulting in suboptimal generative performance and limited scene generalization. To address these and advance ZSL, we propose an inductive variational autoencoder for generative zero-shot learning, dubbed GenZSL. Mimicking human-level concept learning, GenZSL operates by inducting new class samples from similar seen classes using weak class semantic vectors derived from target class names (i.e., CLIP text embedding). To ensure the generation of informative samples for training an effective ZSL classifier, our GenZSL incorporates two key strategies. Firstly, it employs class diversity promotion to enhance the diversity of class semantic vectors. Secondly, it utilizes target class-guided information boosting criteria to optimize the model. Extensive experiments conducted on three popular benchmark datasets showcase the superiority and potential of our GenZSL with significant efficacy and efficiency over f-VAEGAN, e.g., 24.7% performance gains and more than $60\times$ faster training speed on AWA2. Codes are available at https://github.com/shiming-chen/GenZSL.