Absolute Zero-Shot Learning
This addresses data privacy and copyright issues for machine learning practitioners by enabling model training without access to real data, though it appears incremental as it builds on existing zero-shot learning paradigms.
The paper tackles the problem of training classifiers without any real data due to copyright and privacy concerns by introducing Absolute Zero-Shot Learning (AZSL), which uses a teacher model to guide training without data leakage. The framework achieves state-of-the-art performance in zero-shot and generalized zero-shot learning under white-box scenarios while showing promising results in black-box settings.
Considering the increasing concerns about data copyright and privacy issues, we present a novel Absolute Zero-Shot Learning (AZSL) paradigm, i.e., training a classifier with zero real data. The key innovation is to involve a teacher model as the data safeguard to guide the AZSL model training without data leaking. The AZSL model consists of a generator and student network, which can achieve date-free knowledge transfer while maintaining the performance of the teacher network. We investigate `black-box' and `white-box' scenarios in AZSL task as different levels of model security. Besides, we also provide discussion of teacher model in both inductive and transductive settings. Despite embarrassingly simple implementations and data-missing disadvantages, our AZSL framework can retain state-of-the-art ZSL and GZSL performance under the `white-box' scenario. Extensive qualitative and quantitative analysis also demonstrates promising results when deploying the model under `black-box' scenario.