Abstract:Existing semi-supervised learning (SSL) methods assume that labeled and unlabeled data share the same class space. However, in real-world applications, unlabeled data always contain classes not present in the labeled set, which may cause classification performance degradation of known classes. Therefore, open-world SSL approaches are researched to handle the presence of multiple unknown classes in the unlabeled data, which aims to accurately classify known classes while fine-grained distinguishing different unknown classes. To address this challenge, in this paper, we propose an open-world SSL method for Self-learning Open-world Classes (SSOC), which can explicitly self-learn multiple unknown classes. Specifically, SSOC first defines class center tokens for both known and unknown classes and autonomously learns token representations according to all samples with the cross-attention mechanism. To effectively discover novel classes, SSOC further designs a pairwise similarity loss in addition to the entropy loss, which can wisely exploit the information available in unlabeled data from instances' predictions and relationships. Extensive experiments demonstrate that SSOC outperforms the state-of-the-art baselines on multiple popular classification benchmarks. Specifically, on the ImageNet-100 dataset with a novel ratio of 90%, SSOC achieves a remarkable 22% improvement.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is dealing with the existence of unknown classes in semi - supervised learning (SSL). Specifically, existing semi - supervised learning methods assume that the labeled data and the unlabeled data share the same class space, that is, the unlabeled data does not contain new unknown classes. However, in practical applications, unlabeled data often contains new classes that are not in the labeled data set, which may lead to a decline in the classification performance of known classes. Therefore, this paper proposes an open - world semi - supervised learning method (Open - World SSL), aiming to accurately classify known classes and finely distinguish different unknown classes simultaneously. ### Main contributions of the paper 1. **Proposed a new open - world semi - supervised learning method**: - This method is called Self - Supervised Open - World Class (SSOC), which can explicitly self - learn multiple unknown classes. - Use the Cross - Attention Mechanism to explicitly model class concepts and autonomously learn multiple unknown classes. 2. **Designed pairwise similarity loss**: - In addition to entropy loss, pairwise similarity loss is also designed to intelligently utilize the information in unlabeled data and discover new classes through instance prediction and relationship identification. 3. **Experimental verification**: - Extensive experiments were carried out on the CIFAR - 10, CIFAR - 100 and ImageNet - 100 data sets, showing the significant robustness and effectiveness of SSOC in the case of limited labeled data and a large number of unknown classes. - Especially in the ImageNet - 100 data set, when the proportion of new - type classes reaches 90%, SSOC has achieved a significant improvement of 22%. ### Method overview 1. **Model architecture**: - SSOC uses a pre - trained deep neural network to extract image features and dynamically adjusts the class centers through the cross - attention mechanism. - The cross - attention mechanism captures the correlation between image features and class centers, thereby achieving explicit modeling of class information. 2. **Optimization objectives**: - **Cross - entropy loss**: Used to select unlabeled data with high confidence and maximize the use of label information. - **Pairwise similarity loss**: Used to select pairs of related samples with high confidence, reduce noise interference and enhance the robustness of the model. - **Maximum entropy regularization**: Increase the uncertainty of model predictions and promote a more uniform class distribution. ### Experimental results - The experimental results on multiple data sets show that SSOC performs excellently in handling open - world semi - supervised learning tasks, especially in the case of dealing with a large number of unknown classes, significantly outperforming existing methods. ### Summary This paper proposes a new open - world semi - supervised learning method SSOC. By explicitly modeling class concepts and utilizing the information in unlabeled data, it effectively solves the performance degradation problem of traditional semi - supervised learning methods when dealing with unknown classes. The experimental results verify the effectiveness and robustness of this method.

Robust Semi-Supervised Learning for Self-learning Open-World Classes

Label Distribution-based Open-world Semi-supervised Learning

Open World Semi-supervised Learning Based on Multi-scale Enhanced Feature

Robust Semi-Supervised Learning when Not All Classes have Labels

Open-World Semi-Supervised Learning for Node Classification

Open-world Barely-Supervised Learning Via Augmented Pseudo Labels

Robust Semi-supervised Learning by Wisely Leveraging Open-set Data

Exploration and Exploitation of Unlabeled Data for Open-Set Semi-Supervised Learning

Promote knowledge mining towards open-world semi-supervised learning

Towards Realistic Long-tailed Semi-supervised Learning in an Open World

SemiRS-COC: Semi-Supervised Classification for Complex Remote Sensing Scenes With Cross-Object Consistency

Rethinking Open-World Semi-Supervised Learning: Distribution Mismatch and Inductive Inference

Learning Where to Learn in Cross-View Self-Supervised Learning

Open-Set Semi-Supervised Learning for 3D Point Cloud Understanding

Self-Supervised Features Improve Open-World Learning

Multi-Task Curriculum Framework for Open-Set Semi-Supervised Learning

Open-world Semi-supervised Novel Class Discovery

A Graph-Theoretic Framework for Understanding Open-World Semi-Supervised Learning

OCI-SSL: Open Class-Imbalanced Semi-Supervised Learning with Contrastive Learning

S2OSC: A Holistic Semi-Supervised Approach for Open Set Classification

Enhancing Semi-Supervised Learning with Cross-Modal Knowledge