SOTAVerified

Self-Supervised Learning

Self-Supervised Learning is proposed for utilizing unlabeled data with the success of supervised learning. Producing a dataset with good labels is expensive, while unlabeled data is being generated all the time. The motivation of Self-Supervised Learning is to make use of the large amount of unlabeled data. The main idea of Self-Supervised Learning is to generate the labels from unlabeled data, according to the structure or characteristics of the data itself, and then train on this unsupervised data in a supervised manner. Self-Supervised Learning is wildly used in representation learning to make a model learn the latent features of the data. This technique is often employed in computer vision, video processing and robot control.

Source: Self-supervised Point Set Local Descriptors for Point Cloud Registration

Image source: LeCun

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 4551–4600 of 5044 papers

TitleStatusHype
Learning Visual Representations for Transfer Learning by Suppressing TextureCode1
Patch2Self: Denoising Diffusion MRI with Self-Supervised LearningCode1
NICT Kyoto Submission for the WMT’20 Quality Estimation Task: Intermediate Training for Domain and Task Adaptation—0
Response Selection for Multi-Party Conversations with Dynamic Topic Tracking—0
Understanding Pre-trained BERT for Aspect-based Sentiment AnalysisCode1
Scene Flow from Point Clouds with or without Learning—0
Self-supervised Representation Learning for Evolutionary Neural Architecture SearchCode0
A Survey on Contrastive Self-supervised Learning—0
Joint Masked CPC and CTC Training for ASRCode1
PAL : Pretext-based Active Learning—0
Combining Self-Training and Self-Supervised Learning for Unsupervised Disfluency DetectionCode1
Pretext-Contrastive Learning: Toward Good Practices in Self-supervised Video Representation LeaningCode1
Multimodal Emotion Recognition with Transformer-Based Self Supervised Feature FusionCode1
Refactoring Policy for Compositional Generalizability using Self-Supervised Object Proposals—0
Multi-object tracking with self-supervised associating network—0
XLVIN: eXecuted Latent Value Iteration Nets—0
Pre-training Text-to-Text Transformers for Concept-centric Common SenseCode1
BARThez: a Skilled Pretrained French Sequence-to-Sequence ModelCode1
LoopReg: Self-supervised Learning of Implicit Surface Correspondences, Pose and Shape for 3D Human Mesh Registration—0
Graph Contrastive Learning with AugmentationsCode1
Self-Supervised Shadow Removal—0
Contrastive Learning with Adversarial Examples—0
Contrastive Self-Supervised Learning for Wireless Power ControlCode0
Self-Alignment Pretraining for Biomedical Entity RepresentationsCode1
Self-supervised Human Activity Recognition by Learning to Predict Cross-Dimensional Motion—0
Self-supervised Graph Learning for RecommendationCode1
Self-Supervised Learning of Part Mobility from Point Cloud SequenceCode0
BYOL works even without batch statisticsCode2
Understanding YouTube Communities via Subscription-based Channel EmbeddingsCode1
Self-supervised Geometric Features Discovery via Interpretable Attention for Vehicle Re-Identification and BeyondCode0
CLAR: Contrastive Learning of Auditory Representations—0
SelfVoxeLO: Self-supervised LiDAR Odometry with Voxel-based Deep Neural Networks—0
From Local Structures to Size Generalization in Graph Neural Networks—0
On the surprising similarities between supervised and self-supervised models—0
Unsupervised Natural Language Inference via Decoupled Multimodal Contrastive LearningCode0
For self-supervised learning, Rationality implies generalization, provablyCode0
Response Selection for Multi-Party Conversations withDynamic Topic Tracking—0
Representation Learning via Invariant Causal MechanismsCode1
Self-Supervised Ranking for Representation Learning—0
Are all negatives created equal in contrastive instance discrimination?—0
Measuring Visual Generalization in Continuous Control from PixelsCode1
SAR: Scale-Aware Restoration Learning for 3D Tumor Segmentation—0
Audio-Visual Self-Supervised Terrain Type Discovery for Mobile Platforms—0
MixCo: Mix-up Contrastive Learning for Visual RepresentationCode1
Self-Supervised Multi-View Synchronization Learning for 3D Pose Estimation—0
MS^2L: Multi-Task Self-Supervised Learning for Skeleton Based Action RecognitionCode1
Rethinking supervised learning: insights from biological learning and from calling it by its name—0
Learning 3D Face Reconstruction with a Pose Guidance Network—0
A Human Ear Reconstruction Autoencoder—0
Pathological Visual Question Answering—0
Show:102550
← PrevPage 92 of 101Next →
All datasetsDABSSTL-10CIFAR10cifar100ImageNet-100 (TEMI Split)TinyImageNetCIFAR-10CIFAR-100CREMA-DTiny ImageNet

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1Pretraining: NoneImages & Text57.5—Unverified
2Pretraining: ShEDImages & Text54.3—Unverified
3Pretraining: e-MixImages & Text48.9—Unverified
#ModelMetricClaimedVerifiedStatus
1ResNet50Accuracy91.7—Unverified
2ResNet18Accuracy91.02—Unverified
3MV-MRAccuracy89.67—Unverified
#ModelMetricClaimedVerifiedStatus
1ResNet50average top-1 classification accuracy93.89—Unverified
2ResNet18average top-1 classification accuracy92.58—Unverified
#ModelMetricClaimedVerifiedStatus
1ResNet50average top-1 classification accuracy72.51—Unverified
2ResNet18average top-1 classification accuracy69.31—Unverified
#ModelMetricClaimedVerifiedStatus
1CorInfomax (ResNet50)Top-1 Accuracy82.64—Unverified
2CorInfomax (ResNet18)Top-1 Accuracy80.48—Unverified
#ModelMetricClaimedVerifiedStatus
1ResNet50average top-1 classification accuracy51.84—Unverified
2ResNet18average top-1 classification accuracy51.67—Unverified
#ModelMetricClaimedVerifiedStatus
1CorInfomax (ResNet18)Top-1 Accuracy93.18—Unverified
#ModelMetricClaimedVerifiedStatus
1CorInfomax (ResNet18)Top-1 Accuracy71.61—Unverified
#ModelMetricClaimedVerifiedStatus
1Hybrid BYOL-S/CvTAccuracy67.2—Unverified
#ModelMetricClaimedVerifiedStatus
1CorInfomax (ResNet50)Top-1 Accuracy54.86—Unverified