摘要:
论文“Bi-directional Distribution Alignment for Transductive Zero-Shot Learning”提出Bi-VAEGAN,它以f-VAEGAN-D2为Baseline,进一步发展了TF-VAEGAN通过利用所见数据和反馈模块增强生成的视觉特征思 阅读全文
摘要:
前面介绍了将VAE+GAN解决零样本学习的方法:f-VAEGAN-D2,这里继续讨论引入生成模型处理零样本学习(Zero-shot Learning, ZSL)问题。论文“Latent Embedding Feedback and Discriminative Features for Zero-S 阅读全文