</> Code with Papers by Syntology

Papers › Generalized Zero- and Few-Shot Learning via Aligned Variational Autoencoders

Generalized Zero- and Few-Shot Learning via Aligned Variational Autoencoders

1 Jun 2019CVPR 2019 6archive 2025-07-28

Edgar Schonfeld, Sayna Ebrahimi, Samarth Sinha, Trevor Darrell, Zeynep Akata

Many approaches in generalized zero-shot learning rely on cross-modal mapping between the image feature space and the class embedding space. As labeled images are expensive, one direction is to augment the dataset by generating either images or image features. However, the former misses fine-grained details and the latter requires learning a mapping associated with class embeddings. In this work, we take feature generation one step further and propose a model where a shared latent space of image features and class embeddings is learned by modality-specific aligned variational autoencoders. This leaves us with the required discriminative information about the image and classes in the latent features, on which we train a softmax classifier. The key to our approach is that we align the distributions learned from images and from side-information to construct latent features that contain the essential multi-modal information associated with unseen classes. We evaluate our learned latent features on several benchmark datasets, i.e. CUB, SUN, AWA1 and AWA2, and establish a new state of the art on generalized zero-shot as well as on few-shot learning. Moreover, our results on ImageNet with various zero-shot splits show that our latent features generalize well in large-scale settings.

PaperPDFCode

Code

edgarschnfld/CADA-VAE-PyTorch officialmentioned in paperpytorchMIT report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Few-Shot LearningGeneralized Few-Shot LearningGeneralized Zero-Shot LearningGeneralized Zero-Shot Learning - UnseenLong-tail learning with class descriptorsZero-Shot Learning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Generalized Few-Shot Learning AwA2 DA-VAE Per-Class Accuracy (1-shot) 68.0 #3 of 6 Archive leaderboard report
Generalized Few-Shot Learning AwA2 DA-VAE Per-Class Accuracy (10-shots) 76.8 #3 of 6 Archive leaderboard report
Generalized Few-Shot Learning AwA2 DA-VAE Per-Class Accuracy (2-shots) 73.0 #3 of 6 Archive leaderboard report
Generalized Few-Shot Learning AwA2 DA-VAE Per-Class Accuracy (5-shots) 75.6 #3 of 6 Archive leaderboard report
Generalized Few-Shot Learning AwA2 CA-VAE Per-Class Accuracy (1-shot) 64.0 #5 of 6 Archive leaderboard report
Generalized Few-Shot Learning AwA2 CA-VAE Per-Class Accuracy (10-shots) 79.0 #5 of 6 Archive leaderboard report
Generalized Few-Shot Learning AwA2 CA-VAE Per-Class Accuracy (2-shots) 71.3 #5 of 6 Archive leaderboard report
Generalized Few-Shot Learning AwA2 CA-VAE Per-Class Accuracy (5-shots) 76.6 #5 of 6 Archive leaderboard report
Generalized Few-Shot Learning CUB CADA-VAE Per-Class Accuracy (2-shots) 59.2 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CADA-VAE Per-Class Accuracy (1-shot) 55.2 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CADA-VAE Per-Class Accuracy (10-shots) 64.9 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CADA-VAE Per-Class Accuracy (20-shots) 66.0 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CADA-VAE Per-Class Accuracy (5-shots) 63.0 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CA-VAE Per-Class Accuracy (2-shots) 54.4 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CA-VAE Per-Class Accuracy (1-shot) 50.6 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CA-VAE Per-Class Accuracy (10-shots) 62.2 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB CA-VAE Per-Class Accuracy (5-shots) 59.6 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB DA-VAE Per-Class Accuracy (2-shots) 54.6 #4 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB DA-VAE Per-Class Accuracy (1-shot) 49.2 #4 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB DA-VAE Per-Class Accuracy (10-shots) 60.8 #4 of 5 Archive leaderboard report
Generalized Few-Shot Learning CUB DA-VAE Per-Class Accuracy (5-shots) 58.8 #4 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN DA-VAE Per-Class Accuracy (1-shot) 40.6 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN DA-VAE Per-Class Accuracy (10-shots) 47.6 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN DA-VAE Per-Class Accuracy (2-shots) 43.0 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN DA-VAE Per-Class Accuracy (5-shots) 46.0 #2 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CADA-VAE Per-Class Accuracy (1-shot) 37.8 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CADA-VAE Per-Class Accuracy (10-shots) 45.8 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CADA-VAE Per-Class Accuracy (2-shots) 41.4 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CADA-VAE Per-Class Accuracy (5-shots) 44.2 #3 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CA-VAE Per-Class Accuracy (1-shot) 37.8 #4 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CA-VAE Per-Class Accuracy (10-shots) 45.1 #4 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CA-VAE Per-Class Accuracy (2-shots) 40.8 #4 of 5 Archive leaderboard report
Generalized Few-Shot Learning SUN CA-VAE Per-Class Accuracy (5-shots) 43.6 #4 of 5 Archive leaderboard report
Long-tail learning with class descriptors AWA-LT CADA-VAE Long-Tailed Accuracy 89.5 #3 of 5 Archive leaderboard report
Long-tail learning with class descriptors AWA-LT CADA-VAE Per-Class Accuracy 73.5 #3 of 5 Archive leaderboard report
Long-tail learning with class descriptors CUB-LT CADA-VAE Long-Tailed Accuracy 57.4 #5 of 5 Archive leaderboard report
Long-tail learning with class descriptors CUB-LT CADA-VAE Per-Class Accuracy 48.3 #5 of 5 Archive leaderboard report
Long-tail learning with class descriptors ImageNet-LT-d CADA-VAE Per-Class Accuracy 49.3 #4 of 5 Archive leaderboard report
Long-tail learning with class descriptors SUN-LT CADA-VAE Long-Tailed Accuracy 35.1 #4 of 5 Archive leaderboard report
Long-tail learning with class descriptors SUN-LT CADA-VAE Per-Class Accuracy 32.8 #4 of 5 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Softmax

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections