Exploiting weakly-labeled Web images to improve object classification: a domain adaptation approach

By Alessandro Bergamo, Lorenzo Torresani. 2010

261 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning

How researchers see this paper

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

In Representation & Self-Supervised Learning

Citations in the corpus

Revisiting Unreasonable Effectiveness of Data in Deep Learning Era (2017, background): “In order to overcome the bottleneck, there have been recent efforts on visual representation learning using web-supervision [2, 5, 6, 9, 21, 23, 24, 27] or unsupervised [10, 11, 31, 32, 34, 42, 43] paradigms.”
Learning Video Representations from Textual Web Supervision (2020, methodology): “The most commonly-used approach is to use image search results, and to label each image with the query that was used to ﬁnd it [65, 15, 56, 6, 11, 13, 21, 10, 18, 35].”
Webly Supervised Learning of Convolutional Networks (2015, background): “This tends to offer less biased training pairs, but unfortunately such an association is often too weak and hard to capture, since visual knowledge is usually regarded as common sense knowledge and too obvious to be mentioned in the text [9].”
Few-Shot Adversarial Domain Adaptation (2017, methodology): “Finally, some use regularization to improve the fit on the target domain [4, 1, 62, 10, 3, 8].”
Simultaneous Deep Transfer Across Domains and Tasks (2015, background): “For the supervised adaptation scenario, when a limited amount of labeled data is available in the target domain, some approaches have been proposed to learn a target classifier regularized against the source classifier [32, 2, 1].”
Deep Domain Confusion: Maximizing for Domain Invariance (2014, background): “For the supervised adaptation scenario, when a limited amount of labeled data is available in the target domain, some approaches have been proposed to learn a target classifier regularized against the source classifier [33, 2, 1].”
Discriminative virtual views for cross-view action recognition (2012, background): “See [5, 25] for more details.”
Large-Scale Weakly-Supervised Pre-Training for Video Action Recognition (2019, background): “Dataset sources: Many prior approaches use Internet as a natural source of content [6, 12, 13, 18, 24, 25, 34, 37, 44, 46, 51, 58, 67] and the associated search queries, hashtags, or user-comments as labels to construct datasets.”
Unified Deep Supervised Domain Adaptation and Generalization (2017, background): “The third one regularizes a classifier trained on a source distribution to work well on a target distribution [4, 1, 66, 15, 3, 12].”
Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, methodology): “In the same spirit, webly supervised approaches [82, 242, 189, 14, 62, 37, 93] are also related to DA and HDA as is these approaches rely on collected Web data (source) data used to refine the target model.”
Learning from Noisy Labels with Distillation (2017, background): “also proposed to exploit small manually labeled dataset to learn with text based image search in the framework of domain adaptation [1].”
Revisiting Batch Normalization For Practical Domain Adaptation (2016, methodology): “Caltech-Bing (Bergamo and Torresani 2010) is a much larger domain adaptation dataset, which contains 30,607 and 121,730 images in 256 categories from two domains Caltech-256(C) and Bing(B).”
Automatic Feature Decomposition for Single View Co-training (2011, methodology): “With this method, Bergamo & Torresani (2010) report an improvement of 65% (27.1% compared to 16.7”
Unsupervised Domain Adaptation by Domain Invariant Projection (2013, background): “Existing domain adaptation methods can be divided into two categories: Semi-supervised approaches [12, 3, 26] that assume that a small number of labeled examples from the target domain are available during training, and unsupervised approaches [15, 14, 16, 21] that do not require any labels from the”
DLID: Deep Learning for Domain Adaptation by Interpolating between Domains (2013, background): “Another set of approaches in this fixed representation view of the problem treats domain adaptation as a conventional semi-supervised learning (Bergamo & Torresani, 2010; Dai et al., 2007; Yang et al., 2007; Duan et al., 2012).”
Connecting the Dots with Landmarks: Discriminatively Learning Domain-Invariant Features for Unsupervised Domain Adaptation (2013, methodology): “When some labeled data from the target is accessible, the problem is similar to semi-supervised learning and is referred to as semi-supervised domain adaptation (Daumé et al., 2010; Bergamo & Torresani, 2010; Saenko et al., 2010).”
WebVision Database: Visual Learning and Understanding from Web Data (2017, background): “Berg-amo and Torresani [2] studied the domain adaptation issue between web images and existing visual recognition benchmark datasets.”