Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.23641 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918468392058880 |
|---|---|
| author | Yan, Jiawei |
| author_facet | Yan, Jiawei |
| contents | This paper introduces VDLF-Net, which attaches a compact VAE to a multi-scale CNN backbone. Latent vectors and softmax-gate support the backbone feature maps, while $\ell_2$-normalized embeddings from the gated maps contribute toward supervised classification or episodic few-shot prediction. Under standard CIFAR-100 and Mini-ImageNet protocols, VDLF-Net demonstrates an improved performance over ResNet-50 Enhanced, VGG-16, Prototypical Networks, and Matching Networks. Extensive ablations show that removing the fine-resolution scale has the greatest impact on VDLF-Net's performance. At the same time, KL and reconstruction at the chosen $α$ pose a minor performance reduction, demonstrating that performance gains over classical episodic baselines mainly originate from the full VDLF-Net architecture and training strategy. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_23641 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | VDLF-Net: Variational Feature Fusion for Adaptive and Few-Shot Visual Learning Yan, Jiawei Computer Vision and Pattern Recognition This paper introduces VDLF-Net, which attaches a compact VAE to a multi-scale CNN backbone. Latent vectors and softmax-gate support the backbone feature maps, while $\ell_2$-normalized embeddings from the gated maps contribute toward supervised classification or episodic few-shot prediction. Under standard CIFAR-100 and Mini-ImageNet protocols, VDLF-Net demonstrates an improved performance over ResNet-50 Enhanced, VGG-16, Prototypical Networks, and Matching Networks. Extensive ablations show that removing the fine-resolution scale has the greatest impact on VDLF-Net's performance. At the same time, KL and reconstruction at the chosen $α$ pose a minor performance reduction, demonstrating that performance gains over classical episodic baselines mainly originate from the full VDLF-Net architecture and training strategy. |
| title | VDLF-Net: Variational Feature Fusion for Adaptive and Few-Shot Visual Learning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.23641 |