6 个仓库
Awesome GitHub RepositoriesSelf Training Methods
Frameworks that iteratively generate pseudo-labels to adapt models to target domains.
Explore 6 awesome GitHub repositories matching part of an awesome list · Self Training Methods. Refine with filters or upvote what's useful.
Awesome Self Training Methods GitHub Repositories
用 AI 发现最棒的仓库。我们将通过 AI 为您搜索最匹配的仓库。
- thuml/transfer-learning-library
thuml/Transfer-Learning-Library
3,917在 GitHub 上查看这是一个用于计算机视觉中迁移学习和领域自适应的综合库。它作为一个框架，用于对齐源数据集和目标数据集之间的特征分布，也是一个用于领域泛化的工具包，以及一个利用少量标记数据集和大量未标记数据集进行半监督学习的库。 该库为无监督领域自适应提供了专门的功能，包括使用对抗网络、基于差异的架构以及图像到图像的转换来减少分布不匹配。它还包括用于领域泛化的工具，通过风格混合和不变风险最小化来确保模型在未见过的目标域上的可靠性。 该项目涵盖了广泛的功能面，包括任务自适应和使用专门正则化的微调、通过伪标签和一致性学习进行的半监督训练，以及使用可迁移性指标进行迁移学习模型选择。它还包括一个用于自动化获取和准备标准化视觉基准的数据集管理器。 该库包含用于监控和可观测性的工具，例如 t-SNE 可视化和 A-distance 指标，用于分析特征分布和领域差异。
Generates pseudo-labels for unlabeled data using confidence thresholds to guide the training process.
Python - bupt-ai-cz/meta-selflearning
bupt-ai-cz/Meta-SelfLearning
205在 GitHub 上查看Meta Self-learning for Multi-Source Domain Adaptation： A Benchmark
Uses meta-learning to improve self-training for multi-source adaptation.
Python - bupt-ai-cz/iast-eccv2020
bupt-ai-cz/IAST-ECCV2020
IAST: Instance Adaptive Self-training for Unsupervised Domain Adaptation (ECCV 2020) https://teacher.bupt.edu.cn/zhuchuang/en/index.htm
Adapts self-training instances to improve unsupervised performance.
Python - ljjcoder/probabilistic-contrastive-learning
ljjcoder/Probabilistic-Contrastive-Learning
Probabilistic Contrastive Learning for Domain Adaptation
Integrates probabilistic contrastive learning into self-training loops.
Python - ybzh/auxselftrain
YBZh/AuxSelfTrain
Gradual Domain Adaptation via Self-Training of Auxiliary Models
Uses auxiliary models to guide gradual self-training.
Python - reddyav1/unite
reddyav1/unite
CVPR 2024 Code for UNITE, an unsupervised approach for video domain adaptation (https://arxiv.org/abs/2312.02914)
Combines masked pre-training with collaborative self-training for video.
Python
