Deep Reconstruction-Classification Networks for Unsupervised Domain Adaptation

By Muhammad Ghifary, W. Kleijn, Mengjie Zhang, David Balduzzi, Wen Li. 2016

906 citations recorded by the source corpus.

In this paper, we propose a novel unsupervised domain adaptation algorithm based on deep learning for visual object recognition. Specifically, we design a new model called Deep Reconstruction-Classification Network (DRCN), which jointly learns a shared encoding representation for two tasks: (i) supervised classification of labeled source data, and (ii) unsupervised reconstruction of unlabeled target data. In this way, the learnt representation not only preserves discriminability, but also encodes useful information from the target domain. Our new DRCN model can be optimized by using backpropagation similarly as the standard neural networks.

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

Adversarial Discriminative Domain Adaptation (2017, background): “In the case of unlabeled target domains (the focus of this paper) the main strategy has been to guide feature learning by minimizing the difference between the source and target feature distributions [11, 12, 5, 6, 8, 9, 13, 14].”
Few-Shot Adversarial Domain Adaptation (2017, background): “LB [60] [34] [15] [58] [27] [38] FADA A →W 61.”
Moment Matching for Multi-Source Domain Adaptation (2018, background): “decoder [3, 10] or a GAN discriminator, such as dualGAN [46], cycle-GAN [50], disco-GAN [16], and CyCADA [15].”
Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks (2016, background): “[17] propose an alternative model in which the task loss for the source domain is combined with a reconstruction loss for the target domain, which results in learning domain-invariant features.”
CyCADA: Cycle-Consistent Adversarial Domain Adaptation (2017, methodology): “Ghifary et al. (2016) uses an additional reconstruction objective in the target domain to encourage alignment in the unsupervised adaptation setting.”
Unified Deep Supervised Domain Adaptation and Generalization (2017, neutral): “Lower Bound [62] [39] [25] [60] [33] CCSA”
Domain Adaptation for Visual Applications: A Comprehensive Survey (2017, methodology): “In contrast to the above methods, the Deep Reconstruction Classification Network36 proposed in [154] combines the standard convolutional network for”
Domain Adaptive Faster R-CNN for Object Detection in the Wild (2018, background): “Domain Adaptation: Domain adaptation has been widely studied for image classification in computer vision [10, 11, 33, 23, 22, 14, 52, 40, 15, 18, 50, 45, 43, 35].”
Asymmetric Tri-training for Unsupervised Domain Adaptation (2017, background): “…(MMD) (Long et al., 2015b), Domain Adversarial Neural Network (DANN) (Ganin & Lempitsky, 2014), Deep Reconstruction Classification Network (DRCN) (Ghifary et al., 2016), Domain Separation Networks (DSN) (Bousmalis et al., 2016), and k-Nearest Neighbor based adaptation (kNN-Ad) (Sener et al.,…”
Collaborative and Adversarial Network for Unsupervised Domain Adaptation (2018, neutral): “In [14], Grifary et al.”
SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud (2018, methodology): “In addition to the task related loss computed from the labeled source data, deep UDA models are usually trained jointly with another loss, such as a discrepancy loss [15], [16], [17], [18], [6], adversarial loss [19], [20], [21], [22], [23], [24], label distribution loss [18] or reconstruction loss ”
Self-ensembling for visual domain adaptation (2017, background): “[1] Ganin & Lempitsky (2015), [2] Ghifary et al. (2016), [3] Sankaranarayanan et al. (2017), [4] Tzeng et al.”
Deep Visual Domain Adaptation: A Survey (2018, methodology): “Reconstructionbased using the data reconstruction as an auxiliary task to ensure feature invariance encoder-decoder reconstruction [4, 26, 24, 114] adversarial reconstruction [105, 113, 50]”
EEG-Based Emotion Recognition Using Regularized Graph Neural Networks (2019, background): “The alignment can be achieved by minimizing divergence [47], maximizing reconstruction [48], or adversarial training [22].”