Adversarial attacks aim to disturb the functionality of a target system ...
The problem of long-tailed recognition (LTR) has received attention in r...
For best performance, today's semantic segmentation methods use large an...
Recent works have shown that unstructured text (documents) from online
s...
Despite the tremendous progress in zero-shot learning(ZSL), the majority...
Any-shot image classification allows to recognize novel classes with onl...
Human-annotated attributes serve as powerful semantic embeddings in zero...
Parts represent a basic unit of geometric and semantic similarity across...
Compositional Zero-Shot learning (CZSL) aims to recognize unseen composi...
Having access to multi-modal cues (e.g. vision and audio) empowers some
...
Being able to segment unseen classes not observed during training is an
...
In compositional zero-shot learning, the goal is to recognize unseen
com...
Compositional Zero-Shot learning (CZSL) requires to recognize state-obje...
Reducing the amount of supervision required by neural networks is especi...
From the beginning of zero-shot learning research, visual attributes hav...
Few-shot learning methods operate in low data regimes. The aim is to lea...
Intuitively, image classification should profit from using spatial
infor...
When labeled training data is scarce, a promising data augmentation appr...
Suffering from the extreme training data imbalance between seen and unse...
Due to the importance of zero-shot learning, i.e. classifying images whe...
Due to the importance of zero-shot learning, the number of proposed
appr...
We present a novel latent embedding model for learning a compatibility
f...