Transfer Learning for Visual Categorization: A Survey

By Ling Shao, Fan Zhu, Xuelong Li. IEEE Transactions on Neural Networks and Learning Systems, 2015

838 citations recorded by the source corpus.

Regular machine learning and data mining techniques study the training data for future inferences under a major assumption that the future data are within the same feature space or have the same distribution as the training data. However, due to the limited availability of human labeled training data, training data that stay in the same feature space or have the same distribution as the future data cannot be guaranteed to be sufficient enough to avoid the over-fitting problem. In real-world applications, apart from data in the target domain, related data in a different domain can also be included to expand the availability of our prior knowledge about the target future data. Transfer learning addresses such cross-domain learning problems by extracting useful information from data in a related domain and transferring them for being used in target tasks. In recent years, with transfer learning being applied to visual categorization, some typical problems, e.g., view divergence in action recognition tasks and concept drifting in image classification tasks, can be efficiently solved. In this paper, we survey state-of-the-art transfer learning algorithms in visual categorization applications, such as object recognition, image classification, and human action recognition.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/ae20c630759847cd88655631d0041b0be127fd0c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Transfer Learning for Semisupervised Collaborative Recommendation (2016, background): “2007], visual categorization [Shao et al. 2015], and social recommendation [Guo et al.”
Visual and Semantic Knowledge Transfer for Large Scale Semi-Supervised Object Detection (2016, background): “Transfer learning (TL) [38] aims to transfer knowledge across different domains or tasks.”
Cosaliency Detection Based on Intrasaliency Prior Transfer and Deep Intersaliency Mining (2016, background): “the auxiliary data sets would yield a significant performance improvement [23], [24].”
Measuring the tendency of CNNs to Learn Surface Statistical Regularities (2017, background): “The problems of transfer learning [44, 37, 38] and domain adaptation [35, 10] both investigate the generalization ability of deep neural networks.”
Plant Diseases recognition on images using Convolutional Neural Networks: A Systematic Review (2020, background): “Thus, the objective is to borrow labeled or extract knowledge from some related ﬁelds to obtain the highest possible performance in the area of interest[23, 24].”
Salient Band Selection for Hyperspectral Image Classification via Manifold Ranking (2016, methodology): “Traditional methods primarily focus on the spectral feature with adjusted classifier, such as support vector machines (SVMs) [32] and k-nearest neighborhood (kNN) [33], [34].”
A Survey of Unsupervised Deep Domain Adaptation (2018, background): “Transfer learning is a broader topic to cover, thus surveys provide minimal coverage and comparison of the deep learning methods that have been designed for unsupervised domain adaptation [152, 180, 216, 232, 252, 273], or they focus on tasks such as activity recognition [45] or reinforcement learni”
Cross-Modality Submodular Dictionary Learning for Information Retrieval (2014, background): “Thus, asasub-topicoftransferlearning [ 35 ] , thecross-modality scenario is proposed to ﬁll such a gap, where the database contains data in one modality and the query instances are expressed in the other modality / modalities.”
Learning to Rank Using User Clicks and Visual Features for Image Retrieval (2015, background): “One feasible solution to address this problem is to integrate visual information of images [42]–[44], [46], [48]–[50] into the rank learning framework [45], [47].”
Deep Transfer Learning for Image‐Based Structural Damage Recognition (2018, background): “Thus, usually in CNN, TL plays the similar roles of feature representation transfer and parameter transfer (Shao et al., 2014).”
Deep Visual Domain Adaptation: A Survey (2018, methodology): “[80] categorized TL techniques into feature-representation-level knowledge transfer and classifier-level knowledge transfer.”
A Comprehensive Survey on Transfer Learning (2019, background): “Besides, a number of surveys focus on specific application scenarios including activity recognition [25], visual categorization [26], collaborative recommendation [27], computer vision [24], and sentiment analysis [28].”
Local structure preserving discriminative projections for RGB-D sensor-based scene classification (2015, background): “However, differences in viewing angles and lighting variations limit successful scene classification [36,54,58].”
Recent Advances in Open Set Recognition: A Survey (2018, background): “[3]–[5], domain adaptation [6], [7], zero-shot [8]–[10], oneshot (few-shot) [11]–[20] recognition/learning and open set recognition/classification [21]–[23], and so forth.”
Transfer learning: a friendly introduction (2022, background): “This limit can be solved using the trained models with large annotated legacy datasets on new datasets with different domains, and trained models will require the clinical setting [47].”