Deep Hashing Network for Unsupervised Domain Adaptation

By Hemanth Venkateswara, José Eusébio, Shayok Chakraborty, Sethuraman Panchanathan. 2017

2,514 citations recorded by the source corpus.

In recent years, deep neural networks have emerged as a dominant machine learning tool for a wide variety of application domains. However, training a deep neural network requires a large amount of labeled data, which is an expensive process in terms of time, labor and human expertise. Domain adaptation or transfer learning algorithms address this challenge by leveraging labeled data in a different, but related source domain, to develop a model for the target domain. Further, the explosive growth of digital data has posed a fundamental challenge concerning its storage and retrieval. Due to its storage and retrieval efficiency, recent years have witnessed a wide application of hashing in a variety of computer vision applications. In this paper, we first introduce a new dataset, Office-Home, to evaluate domain adaptation algorithms. The dataset contains images of a variety of everyday objects from multiple domains. We then propose a novel deep learning framework that can exploit labeled source data and unlabeled target data to learn informative hash codes, to accurately classify unseen target data. To the best of our knowledge, this is the first research effort to exploit the feature learning capabilities of deep neural networks to learn representative hash codes to address the domain adaptation problem. Our extensive empirical studies on multiple transfer tasks corroborate the usefulness of the framework in learning efficient hash codes which outperform existing competitive baselines for unsupervised domain adaptation.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

Citations in the corpus

Break-It-Fix-It: Unsupervised Learning for Program Repair (2021, background): “Such domain shifts typically occur across related but different datasets (Torralba & Efros, 2011; Fang et al., 2013; Venkateswara et al., 2017; Yu et al., 2018b; 2019; Peng et al., 2019; Kamath et al., 2020), as well as from synthetic data to real data (including sim2real) (Wang et al., 2015; Ganin…”
Rescaling Egocentric Vision: Collection, Pipeline and Challenges for EPIC-KITCHENS-100 (2020, methodology): “Im a g e Office [86] 4110 N/A 31 1 3 N/A 2010 Real Office↔ Home [87] 15500 N/A 65 4 1 N/A 2017 Real VisDA-C [88] 280157 N/A 12 3 1 N/A 2017 Real/Syn DomainNet [89] 363534 37706 345 6 1 N/A 2019 Real/Syn”
Prototype-Guided Continual Adaptation for Class-Incremental Unsupervised Domain Adaptation (2022, positive): “To be specific, ProCA achieves the best or comparable performance on all transfer tasks ( e.g., Ar → Cl on Office-Home-CI), which demonstrates the effectiveness of our method.”
On Generalizing Beyond Domains in Cross-Domain Continual Learning (2022, methodology): “Cross-domain continual learning average accuracy numbers for unseen domains with 10-tasks and 5-tasks protocols on the DomainNet [29] and OfficeHome [34] datasets.”
SWAD: Domain Generalization by Seeking Flat Minima (2021, positive): “It takes about 4 hours for PACS and VLCS , 8 hours for OfficeHome , 8.5 hours for TerraIncognita , and 56 hours for DomainNet on average, when using a single NVIDIA Tesla P40 GPU.”
Moment Matching for Multi-Source Domain Adaptation (2018, background): “Dataset Year Images Classes Domains Description Digit-Five - ∼100,000 10 5 digit Office [37] 2010 4,110 31 3 office Office-Caltech [11] 2012 2,533 10 4 office CAD-Pascal [33] 2015 12,000 20 6 animal,vehicle Office-Home [43] 2017 15,500 65 4 office, home PACS [21] 2017 9,991 7 4 animal, stuff Open MI”
In Search of Lost Domain Generalization (2020, methodology): “…loaders for seven multi-domain image classiﬁcation tasks: Colored MNIST [Arjovsky et al., 2019], Rotated MNIST [Ghifary et al., 2015], PACS [Li et al., 2017], VLCS [Fang et al., 2013], Ofﬁce-Home [Venkateswara et al., 2017], Terra Incognita [Beery et al., 2018], and DomainNet [Peng et al., 2019].”
Learning Adaptive Classifiers Synthesis for Generalized Few-Shot Learning (2019, methodology): “We also investigate theOffice-Home (Venkateswara et al. 2017) dataset, which originates from a domain adaptation task.”
WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, neutral): “…on SVHN street signs (LeCun et al., 1998; Yuval et al., 2011; Tzeng et al., 2017; Hoffman et al., 2018), as well as transfers across datasets containing different renditions (e.g., photos, clipart, sketches) in DomainNet (Peng et al., 2019) and the Office-Home dataset (Venkateswara et al., 2017).”
Few-Shot Learning via Embedding Adaptation With Set-to-Set Functions (2018, methodology): “In addition to these, we investigate the OfficeHome [42] dataset to validate the generalization ability of FEAT across domains.”
Unsupervised Domain Adaptation via Structurally Regularized Deep Clustering (2020, background): “This is reasonable since the four domains in Office-Home contain more categories, are visually more different from each other, and have much lower in-domain classification results [55].”
SRM: A Style-Based Recalibration Module for Convolutional Neural Networks (2019, methodology): “We adopt the Office-Home dataset [32] which consists of 15,588 images from 65 categories across 4 heterogeneous domains: Art (Ar), Clip-art (Cl), Product (Pr) and Real-world (Rw).”
Domain Generalization with MixStyle (2021, methodology): “Similar to the results on PACS, here we observe that MixStyle also brings clear improvements to the baseline CNN model and outperforms all general-purpose regularization methods on both Digits-DG and Ofﬁce-Home.”
Unsupervised Multi-Target Domain Adaptation: An Information Theoretic Approach (2018, methodology): “We compare the proposed method with state-of-the-art methods on standard benchmark datasets: a digit classification task that includes 4 datasets: MNIST [44], MNIST-M [45], SVHN [46], USPS [40], Multi-PIE expression recognition dataset [47], and three multi-domain object recognition datasets namely ”
Open Compound Domain Adaptation (2019, background): “The goal is to retain recognition accuracies in new domains without ground truth annotations [41, 46, 49, 38].”
Conditional Adversarial Domain Adaptation (2017, background): “An interpretation is that the four domains in the Office-Home dataset are with more categories, are visually very dissimilar with each other, and are difficult in each domain with much lower in-domain classification accuracy, as described in its original paper (Venkateswara et al., 2017).”
Separate to Adapt: Open Set Domain Adaptation via Progressive Separation (2019, background): “Office-Home [37] is a challenging domain adaptation dataset, crawled through several search engines and online image directories.”
Self-Challenging Improves Cross-Domain Generalization (2020, result): “– Office-Home [28]: 65 object categories over 4 domains (Art, Clipart, Product, and Real-World).”