Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Generalized Zero- and Few-Shot Learning via Aligned Variational Autoencoders
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use cross-alignment rather than distribution alignment alone (ablation) | CADA-VAE · CUB GZSL | H: 45.8 -> 50.2 | Helped |
| Add distribution alignment to the cross-aligned VAE (ablation) | CADA-VAE · CUB GZSL | H: 50.2 -> 52.4 | Helped |
| Use sentence embeddings instead of attributes as class side information (ablation) | CADA-VAE · CUB GZSL | H: 52.4% -> 53.4% | Helped |
| Use Word2Vec class embeddings instead of more visually discriminative side information (ablation) | CADA-VAE · CUB GZSL | H | Hurt |
| Mix attributes and sentence embeddings across seen and unseen classes (ablation) | CADA-VAE · CUB GZSL | accuracy: 40% at (X_S=0,X_U=50); 45% at (X_S=50,X_U=0); 47% at (X_S=50,X_U=100) | Mixed |
| Increase latent dimensionality up to 64, then increase beyond 100 (ablation) | CADA-VAE · CUB GZSL | H: 52.4% at d=64 | Mixed |
| Generate more latent features for unseen than seen classes (ablation) | CADA-VAE · CUB GZSL | accuracy: 37% -> 52%; 38% -> 52% | Helped |
| Use a dynamic training set instead of a fixed latent-feature dataset (ablation) | CADA-VAE · CUB GZSL | H | No clear effect |
| Increase the number of unseen-class training examples from zero to ten shots (ablation) | CADA-VAE · GFSL | H: 12% on CUB; 20% on AWA1&2 | Helped |
| Align latent spaces with attributes on coarse-grained AWA datasets (ablation) | CADA-VAE variants · AWA1 and AWA2 GFSL | H | No clear effect |
Closest papers by meaning
- A Variational Autoencoder with Deep Embedding Model for Generalized Zero-Shot Learning (AAAI 2020 · similarity 0.92)
- Learning Aligned Cross-Modal Representation for Generalized Zero-Shot Classification (AAAI 2022 · similarity 0.92)
- Latent Embeddings for Zero-Shot Classification (CVPR 2016 · similarity 0.91)
- Zero-Shot Learning via Class-Conditioned Deep Generative Models (AAAI 2018 · similarity 0.91)
- TPR: Topology-Preserving Reservoirs for Generalized Zero-Shot Learning (NeurIPS 2024 · similarity 0.89)
- GenZSL: Generative Zero-Shot Learning Via Inductive Variational Autoencoder (ICML 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.