Boosting for transfer learning with multiple sources

By Yi Yao, Gianfranco Doretto. 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2010

503 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/a3c2887ccb08d98c69ec1c221888f6abbe96b518

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

Multi-Fidelity Physics-Constrained Neural Network and Its Application in Materials Modeling (2019, background): “Transfer learning [3] is an approach to reduce the required amount of training data ,byreusingthesourcedata (instancetransfer)[4,5],previouslytrainedfeaturesorbasisfunctions(featurerepresentationtransfer)[6,7],orpreviouslytrainedparame-ters(parametertransfer)[8,9].”
Wireless Networks Design in the Era of Deep Learning: Model-Based, AI-Based, or Both? (2019, methodology): “Recently, instance-based transfer learning has proved effective when employed in conjunction with the AdaBoost training algorithm, addressing both classification and regression problems [149], [150].”
A Survey on Negative Transfer (2020, methodology): “Test Performance Target performance [34], [35] Domain classifier [36]–[38]”
A Comprehensive Overview and Comparative Analysis on Deep Learning Models: CNN, RNN, LSTM, GRU (2023, unknown): “Algorithms such as TaskTrAdaBoost [101] and TrAdaBoost.”
Multisource Transfer Learning for Cross-Subject EEG Emotion Recognition (2020, background): “Existing studies have indicated that brute force leveraging of the sources poorly related to the target may decrease the performance, which is referred to as “negative transfer” [31].”
Overcoming Negative Transfer: A Survey (2020, background): “However, the original approach doesn’t consider NT. Yao et al. [12] showed that boosting knowledge from multiple sources to increase robustness.”
Stratified Transfer Learning for Cross-domain Activity Recognition (2017, methodology): “Parameter-based methods [28], [29] first train a model using the labeled source domain, then perform clustering on the”
Transfer Learning Based Visual Tracking with Gaussian Processes Regression (2014, methodology): “In particular, we adopt the task-transfer strategy [38], where the tracking decision using the re-weighted auxiliary samples assists the decision using target samples by fusing these two decisions.”
Towards cross-category knowledge propagation for learning visual concepts (2011, methodology): “The work in [14] proposes a two phase transfer approach by identifying which models from various sources can be reused to improve the target classifier.”
Deep learning for image classification on very small datasets using transfer learning (2019, background): “However, training a new classiﬁer with a small number of samples would dramatically increase the risk of overﬁtting the new data, leading to poor generalization [18].”
Multimodal Sentiment Analysis: A Survey of Methods, Trends, and Challenges (2023, methodology): “TrAdaBoost [36] algorithm was employed to reduce the impact of a low-quality translation corpus.”
Transitive Transfer Learning (2015, unknown): “For instance, the work in [26] extends TrAdaboost [9] by adding a wrapper boosting framework on weighting each source domain.”
Information Bottleneck Learning Using Privileged Information for Visual Recognition (2016, background): “It is different from domain adaptation and transfer learning [1, 48, 32], where the source and target domains are closely related but statistically different.”
Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, methodology): “clustering algorithm17 to identify automatically ”source” domains in a large data set, on which they apply a multi-source extension of the Asymmetric Kernel Transforms (AST) [129]. Yao and Doretto in [265] efﬁciently extends the TrAdaBoost [54] to multiple source domains. Source domain weighting. Wh”
From N to N+1: Multiclass Transfer Incremental Learning (2013, methodology): “Experiments are reported in Section 5, followed by conclusions in Section 6.”
Generalization Bounds for Transfer Learning under Model Shift (2015, background): “Other existing work focuses on ﬁnding the set of sources that are closely re-lated to the target (Crammer et al. (2008)), or a reweighting of sources based on prediction errors (Yao and Doretto, (2010)).”
Learning Categories From Few Examples With Multi Model Knowledge Transfer (2014, methodology): “A large variety of methods have been studied to integrate in different ways the source and target information: boosting approaches [18], [9], KNN [26], Markov logic [27], graphical models [28].”