Zero-Shot Learning

Zero-shot learning (ZSL) is a model's ability to detect classes never seen during training. The condition is that the classes are not known during supervised learning.

Earlier work in zero-shot learning use attributes in a two-step approach to infer unknown classes. In the computer vision context, more recent advances learn mappings from image feature space to semantic space. Other approaches learn non-linear multimodal embeddings. In the modern NLP context, language models can be evaluated on downstream tasks without fine tuning.

Benchmark datasets for zero-shot learning include aPY, AwA, and CUB, among others.

( Image credit: Prototypical Networks for Few shot Learning in PyTorch )

Papers

Recently Added Most Hyped Most Active Needs Verification Most Verified

Showing 476–500 of 1864 papers

Title	Date	Tasks	Status	Hype
Entity6K: A Large Open-Domain Evaluation Dataset for Real-World Entity Recognition	Mar 19, 2024	Dense CaptioningImage Captioning	—Unverified	0
MEDBind: Unifying Language and Multimodal Medical Data Embeddings	Mar 19, 2024	Language ModelingLanguage Modelling	—Unverified	0
CLIP-VIS: Adapting CLIP for Open-Vocabulary Video Instance Segmentation	Mar 19, 2024	DecoderInstance Segmentation	CodeCode Available	1
UniBind: LLM-Augmented Unified and Balanced Representation Space to Bind Them All	Mar 19, 2024	AllContrastive Learning	—Unverified	0
Eye-gaze Guided Multi-modal Alignment for Medical Representation Learning	Mar 19, 2024	Diagnosticimage-classification	CodeCode Available	1
Leveraging Large Language Models to Extract Information on Substance Use Disorder Severity from Clinical Notes: A Zero-shot Learning Approach	Mar 18, 2024	DiagnosticZero-Shot Learning	—Unverified	0
Meta-Prompting for Automating Zero-shot Visual Recognition with LLMs	Mar 18, 2024	Language ModellingLarge Language Model	CodeCode Available	1
Boosting Continual Learning of Vision-Language Models via Mixture-of-Experts Adapters	Mar 18, 2024	Continual LearningIncremental Learning	CodeCode Available	3
CoLeCLIP: Open-Domain Continual Learning via Joint Task Prompt and Vocabulary Learning	Mar 15, 2024	class-incremental learningClass Incremental Learning	CodeCode Available	1
OpenGraph: Open-Vocabulary Hierarchical 3D Graph Representation in Large-Scale Outdoor Environments	Mar 14, 2024	Zero-Shot Learning	CodeCode Available	2
Select and Distill: Selective Dual-Teacher Knowledge Transfer for Continual Learning on Vision-Language Models	Mar 14, 2024	Continual LearningKnowledge Distillation	—Unverified	0
Sentinel-Guided Zero-Shot Learning: A Collaborative Paradigm without Real Data Exposure	Mar 14, 2024	Zero-Shot Learning	CodeCode Available	0
Efficient Prompt Tuning of Large Vision-Language Model for Fine-Grained Ship Classification	Mar 13, 2024	Language ModelingLanguage Modelling	CodeCode Available	0
Segmentation of Knee Bones for Osteoarthritis Assessment: A Comparative Analysis of Supervised, Few-Shot, and Zero-Shot Learning Approaches	Mar 13, 2024	Few-Shot LearningMorphological Analysis	—Unverified	0
MolBind: Multimodal Alignment of Language, Molecules, and Proteins	Mar 13, 2024	Contrastive LearningDrug Discovery	—Unverified	0
MoralBERT: A Fine-Tuned Language Model for Capturing Moral Values in Social Discussions	Mar 12, 2024	Domain AdaptationLanguage Modeling	CodeCode Available	1
MENTOR: Multilingual tExt detectioN TOward leaRning by analogy	Mar 12, 2024	Few-Shot LearningScene Text Detection	—Unverified	0
Towards Zero-shot Human-Object Interaction Detection via Vision-Language Integration	Mar 12, 2024	DecoderHuman-Object Interaction Detection	—Unverified	0
Improving deep learning with prior knowledge and cognitive models: A survey on enhancing explainability, adversarial robustness and zero-shot learning	Mar 11, 2024	Adversarial RobustnessDeep Learning	—Unverified	0
Zero-Shot ECG Classification with Multimodal Learning and Test-time Clinical Knowledge Enhancement	Mar 11, 2024	Clinical KnowledgeDescriptive	CodeCode Available	2
Split to Merge: Unifying Separated Modalities for Unsupervised Domain Adaptation	Mar 11, 2024	Domain AdaptationUnsupervised Domain Adaptation	CodeCode Available	1
Personalized LoRA for Human-Centered Text Understanding	Mar 10, 2024	Language ModelingLanguage Modelling	CodeCode Available	1
CSCNET: Class-Specified Cascaded Network for Compositional Zero-Shot Learning	Mar 9, 2024	AttributeCompositional Zero-Shot Learning	—Unverified	0
MedFLIP: Medical Vision-and-Language Self-supervised Fast Pre-Training with Masked Autoencoder	Mar 7, 2024	Domain AdaptationMedical Image Analysis	—Unverified	0
ACC-ViT : Atrous Convolution's Comeback in Vision Transformers	Mar 7, 2024	Contrastive LearningMedical Image Analysis	—Unverified	0

Show:10 25 50

← PrevPage 20 of 75Next →

All datasets CUB-200-2011 MedConceptsQA SUN Attribute AwA2 Caltech-101 CIFAR-10 CIFAR-100 COCO-MLT DTD FGVC-Aircraft Flowers-102 Food-101

Benchmark Results

#	Model	Metric	Claimed	Verified	Status
1	ZeroDiff	average top-1 classification accuracy	87.5	—	Unverified
2	DUET	average top-1 classification accuracy	72.3	—	Unverified
3	Composer	average top-1 classification accuracy	69.4	—	Unverified
4	HDC-ZSC-MLP	average top-1 classification accuracy	65.6	—	Unverified
5	ZSL_TF-VAEGAN	average top-1 classification accuracy	64.9	—	Unverified
6	ZLaP	Accuracy	64.3	—	Unverified
7	ZLaP*	Accuracy	64.2	—	Unverified
8	HDC-ZSC	average top-1 classification accuracy	63.8	—	Unverified
9	SPOT	average top-1 classification accuracy	62.9	—	Unverified
10	f-VAEGAN-D2	average top-1 classification accuracy	61	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	dmis-lab/biobert-v1.1	Accuracy	26.15	—	Unverified
2	meta-llama/Meta-Llama-3-8B-Instruct	Accuracy	25.84	—	Unverified
3	epfl-llm/meditron-7b	Accuracy	25.75	—	Unverified
4	dmis-lab/meerkat-7b-v1.0	Accuracy	25.68	—	Unverified
5	meta-llama/Meta-Llama-3-8B-Instruct	Accuracy	25.65	—	Unverified
6	HuggingFaceH4/zephyr-7b-beta	Accuracy	25.54	—	Unverified
7	dmis-lab/biobert-v1.1	Accuracy	25.46	—	Unverified
8	epfl-llm/meditron-70b	Accuracy	25.36	—	Unverified
9	epfl-llm/meditron-70b	Accuracy	25.26	—	Unverified
10	HuggingFaceH4/zephyr-7b-beta	Accuracy	25.06	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZeroDiff	average top-1 classification accuracy	77.3	—	Unverified
2	SPOT (VAEGAN)	average top-1 classification accuracy	66.04	—	Unverified
3	ZSL_TF-VAEGAN	average top-1 classification accuracy	66	—	Unverified
4	f-VAEGAN	average top-1 classification accuracy	64.7	—	Unverified
5	DUET (Ours)	average top-1 classification accuracy	64.4	—	Unverified
6	LisGAN	average top-1 classification accuracy	61.7	—	Unverified
7	TCN	average top-1 classification accuracy	61.5	—	Unverified
8	f-CLSWGAN	average top-1 classification accuracy	60.8	—	Unverified
9	Cycle-WGAN	average top-1 classification accuracy	59.9	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZeroDiff	average top-1 classification accuracy	86.4	—	Unverified
2	ZSL-KG	average top-1 classification accuracy	78.08	—	Unverified
3	ZSL_TF-VAEGAN	average top-1 classification accuracy	72.2	—	Unverified
4	DUET (Ours)	average top-1 classification accuracy	69.9	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP	Accuracy	84	—	Unverified
2	ZLaP*	Accuracy	83.1	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP*	Accuracy	93.6	—	Unverified
2	ZLaP	Accuracy	93.4	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP*	Accuracy	74.2	—	Unverified
2	ZLaP	Accuracy	74	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ViT-B/16	Average mAP	60.17	—	Unverified
2	ResNet-50	Average mAP	56.19	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP	Accuracy	51.2	—	Unverified
2	ZLaP*	Accuracy	51	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP	Accuracy	29.1	—	Unverified
2	ZLaP*	Accuracy	29	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP	Accuracy	75.9	—	Unverified
2	ZLaP*	Accuracy	75.5	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP*	Accuracy	87.9	—	Unverified
2	ZLaP	Accuracy	87.8	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP	Top 1 Accuracy	72.1	—	Unverified
2	ZLaP*	Top 1 Accuracy	72.1	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	HiTeA	Accuracy	21.7	—	Unverified
2	HiTeA	Accuracy	0.46	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	HiTeA	Accuracy	37.4	—	Unverified
2	HiTeA	Accuracy	0.56	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	SPOT	average top-1 classification accuracy	71.9	—	Unverified
2	ZSL_TF-VAEGAN	average top-1 classification accuracy	70.8	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP	Accuracy	90	—	Unverified
2	ZLaP*	Accuracy	89	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP*	Accuracy	71.8	—	Unverified
2	ZLaP	Accuracy	71.2	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP*	Accuracy	71.4	—	Unverified
2	ZLaP	Accuracy	71	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP*	Accuracy	76.3	—	Unverified
2	ZLaP	Accuracy	76.3	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	CLIP(ViT-B/16)	Average mAP	85.77	—	Unverified
2	CLIP(ResNet-50)	Average mAP	84.3	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZSL-KG	Top-1	60.54	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	zsl_ADA	Average Per-Class Accuracy	70.9	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZLaP*	Accuracy	63.2	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	MSDA	Pearson correlation coefficient (PCC)	0.52	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	SeViLA	Accuracy	72.3	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	M^2-Encoder	Accuracy	80.7	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	FrozenBiLM	Accuracy	51.5	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	CZSL	A-acc	36	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZS3Net	k=10 mIOU	26.3	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	ZSL-KG	Accuracy	88.98	—	Unverified

#	Model	Metric	Claimed	Verified	Status
1	VideoChat2	Accuracy	40.6	—	Unverified