Vision -

I-JEPA: Image-based Joint Embedding Predictive Architecture

Posted on October 9, 2025 | 113 words

A non-generative, self-supervised framework predicting high-level feature representations of masked regions from visible context, enabling scalable and efficient visual pretraining. [Read More]

MaskFeat: Masked Feature Prediction for Self-Supervised Visual Pre-Training

Posted on October 9, 2025 | 1026 words

Predict handcrafted features (e.g., HOG) of masked regions instead of raw pixels. [Read More]

SSL Vision Representation Learning Masked Image Modeling

BYOL: Bootstrap Your Own Latent

Posted on October 8, 2025 | 952 words

Learn representations by predicting one network’s output from another’s, without using negative samples. [Read More]

SSL Vision Representation Learning Joint Embedding Distillation Methods

DINO: Self-Distillation with No Labels

Posted on October 8, 2025 | 1257 words

A student network learns from a teacher network using self-distillation, producing emergent semantic attention maps. [Read More]

SSL Vision Representation Learning Joint Embedding Distillation Methods

MAE: Masked Autoencoders Are Scalable Vision Learners

Posted on October 8, 2025 | 1330 words

Randomly mask image patches and reconstruct the missing ones to learn context-aware visual representations. [Read More]

SSL Vision Representation Learning Masked Image Modeling

SwAV: Swapping Assignments between Views

Posted on October 8, 2025 | 1276 words

Simultaneously cluster the data and learn visual representations by enforcing consistency between cluster assignments, or ‘codes’, generated from different augmented views of the same image. [Read More]

SSL Vision Representation Learning Joint Embedding Clustering Methods

MoCo: Momentum Contrast for Unsupervised Visual Representation Learning

Posted on October 7, 2025 | 1055 words

It stabilizes and scales contrastive learning by maintaining a dynamic dictionary with momentum-based updates, becoming a cornerstone for modern SSL methods. [Read More]

SSL Vision Representation Learning Joint Embedding Contrastive Methods

SimCLR: A Simple Framework for Contrastive Learning of Visual Representations

Posted on October 7, 2025 | 1029 words • Other languages: CH

Learn invariant representations by maximizing similarity between augmented views of the same image while contrasting with others. [Read More]

SSL Vision Representation Learning Joint Embedding Contrastive Methods