Micm: Rethinking unsupervised pretraining for enhanced few-shot learning
Humans exhibit a remarkable ability to learn quickly from a limited number of labeled
samples, a capability that starkly contrasts with that of current machine learning systems …
samples, a capability that starkly contrasts with that of current machine learning systems …
A survey of meta-learning for classification tasks
Y Zhang, B Wei, X Li, L Li - 2022 10th International Conference …, 2022 - ieeexplore.ieee.org
The superior performance of deep learning is supported by massive data and powerful
computing engines. Meta-learning is an imitation of human learning ability. Instead of relying …
computing engines. Meta-learning is an imitation of human learning ability. Instead of relying …
DisRot: boosting the generalization capability of few-shot learning via knowledge distillation and self-supervised learning
Few-shot learning (FSL) aims to adapt quickly to new categories with limited samples.
Despite significant progress in utilizing meta-learning for solving FSL tasks, challenges such …
Despite significant progress in utilizing meta-learning for solving FSL tasks, challenges such …
BECLR: Batch Enhanced Contrastive Few-Shot Learning
S Poulakakis-Daktylidis, H Jamali-Rad - arXiv preprint arXiv:2402.02444, 2024 - arxiv.org
Learning quickly from very few labeled samples is a fundamental attribute that separates
machines and humans in the era of deep representation learning. Unsupervised few-shot …
machines and humans in the era of deep representation learning. Unsupervised few-shot …
Contrastive Learning using Random Walk Laplacian Matrix
I Moummad, B Pasdeloup… - Graph Signal …, 2023 - imt-atlantique.hal.science
In recent years, Self-Supervised Learning (SSL) has gained in popularity due to the
availability of unlabeled data. SSL consists in training a neural network encoder capable of …
availability of unlabeled data. SSL consists in training a neural network encoder capable of …