Deep transfer metric learning

By Junlin Hu, Jiwen Lu, Yap-Peng Tan, Jie Zhou. 2015

278 citations recorded by the source corpus.

Subfields: Biomedical Engineering, 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 Biomedical Engineering

In Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

Citations in the corpus

A Closer Look at Few-shot Classification (2019, methodology): “The importance of reducing intra-class variations of features has been highlighted in deep metric learning Hu et al. (2015) and few-shot classification methods Gidaris & Komodakis (2018). The training procedure of Baseline++ is the same as the original Baseline model except for the classifier design”
Towards Explainable Deep Neural Networks (xDNN) (2019, methodology): “This concept postulates that knowledge in the form of a model architecture learned in one context can be re-used and useful in another context [8].”
A Baseline for Few-Shot Image Classification (2019, methodology): “Initialization for fine-tuning: We use recent ideas from the deep metric learning literature (Hu et al., 2015; Movshovitz-Attias et al., 2017; Qi et al., 2018; Chen et al., 2018; Gidaris & Komodakis, 2018) to initialize the meta-trained model for fine-tuning.”
Transferring GANs: generating images from limited data (2018, background): “Finally, domain transfer has also been studied for networks that learn metrics [18].”
See the Forest for the Trees: Joint Spatial and Temporal Recurrent Neural Networks for Video-Based Person Re-identification (2017, unknown): “And more importantly, deep neural networks have shown to be effective for person re-identification, and have achieved much better performance than traditional methods [1, 6, 9, 12, 13, 18, 25, 36, 38].”
Amalgamating Knowledge towards Comprehensive Classification (2018, background): “It can be roughly categorized into cross-domain (Long et al. 2013; Huang, Huang, and Krähenbühl 2018; Hu, Lu, and Tan 2015; Ding et al. 2018) and cross-task transfer learning (Hong et al. 2016; Cui et al. 2018; Gholami, Rudovic, and Pavlovic 2017).”
Deep Visual Domain Adaptation: A Survey (2018, methodology): “training loss, L = − ∑N i=0 yi log ŷi (ŷi are the softmax predictions of the model, which represent class probabilities) [95, 40, 34, 70, 45, 100].”
ForensicTransfer: Weakly-supervised Domain Adaptation for Forgery Detection (2018, background): “In the literature, the importance of reducing intra-class variances has been demonstrated in [22] for transfer learning and in [9] for few-shot problems.”
A Unified Framework for Metric Transfer Learning (2017, methodology): “Recently, a metric learning method based on deep learning was proposed [32], namely Deep Transfer Metric Learning (DTML).”
Generalizing a Person Retrieval Model Hetero- and Homogeneously (2018, background): “These approaches aim at reducing the gap between source domain and target domain on either the image-level space [1,6] or feature-level space [7,32,33], while overlook the image style variations caused by diﬀerent cameras in target domain.”
Person Re-Identification by Camera Correlation Aware Feature Augmentation (2017, methodology): “Differing significantly from existing deep learning based reid approaches [26], [27], [28], [29] that typically learn directly from the small person re-id training data, we instead utilize large auxiliary less-related image data.”
Customizing Student Networks From Heterogeneous Teachers via Adaptive Knowledge Amalgamation (2019, background): “It contains two main research directions: cross-domain transfer learning [22, 12, 10, 4] and cross-task one [9, 3, 5, 35].”