Curriculum Graph Co-Teaching for Multi-Target Domain Adaptation

Roy, Subhankar; Krivosheev, Evgeny; Zhong, Zhun; Sebe, Nicu; Ricci, Elisa

doi:10.1109/cvpr46437.2021.00531

Cited by 55 publications

(41 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The formal CL concept is first proposed by Bengio et al [7] with experiments on supervised visual and language learning [8][9][10][11][12]. After them, many methods are proposed to pursue generalization improvement or convergence speedup with the spirit of training from a sequence of easy-to-hard data.…”

Section: Related Work 21 Curriculum Learning (Cl)mentioning

confidence: 99%

Mixed-order self-paced curriculum learning for universal lesion detection

Li¹,

Huang²,

Zhou³

2023

Preprint

View full text Add to dashboard Cite

Self-paced curriculum learning (SCL) has demonstrated its great potential in computer vision, natural language processing, etc. During training, it implements easy-to-hard sampling based on online estimation of data difficulty. Most SCL methods commonly adopt a loss-based strategy of estimating data difficulty and deweight the 'hard' samples in the early training stage. While achieving success in a variety of applications, SCL stills confront two challenges in a medical image analysis task, such as universal lesion detection, featuring insufficient and highly class-imbalanced data: (i) the loss-based difficulty measurer is inaccurate; ii) the hard samples are under-utilized from a deweighting mechanism. To overcome these challenges, in this paper we propose a novel mixed-order self-paced curriculum learning (Mo-SCL) method. We integrate both uncertainty and loss to better estimate difficulty online and mix both hard and easy samples in the same mini-batch to appropriately alleviate the problem of under-utilization of hard samples. We provide a theoretical investigation of our method in the context of stochastic gradient descent optimization and extensive experiments based on the DeepLesion benchmark dataset for universal lesion detection (ULD). When applied for two state-of-the-art ULD methods, the proposed mixed-order SCL method can provide a free boost to lesion detection accuracy without extra special network designs.

show abstract

Section: Related Work 21 Curriculum Learning (Cl)mentioning

confidence: 99%

Mixed-order self-paced curriculum learning for universal lesion detection

Li¹,

Huang²,

Zhou³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…In multi-target domain adaptation (MTDA) the goal is learning from a single labeled source domain with the aim of performing well on multiple target domains -at the same time. To tackle MTDA within an image classification context, standard UDA approaches can be directly extended to multiple targets [54,32,155,137].…”

Section: Multi-target Dasismentioning

confidence: 99%

Unsupervised Domain Adaptation for Semantic Image Segmentation: a Comprehensive Survey

Volpi¹,

Chidlovskii²

2021

Preprint

View full text Add to dashboard Cite

Semantic segmentation plays a fundamental role in a broad variety of computer vision applications, providing key information for the global understanding of an image. Yet, the state-of-the-art models rely on large amount of annotated samples, which are more expensive to obtain than in tasks such as image classification. Since unlabelled data is instead significantly cheaper to obtain, it is not surprising that Unsupervised Domain Adaptation reached a broad success within the semantic segmentation community.This survey is an effort to summarize five years of this incredibly rapidly growing field, which embraces the importance of semantic segmentation itself and a critical need of adapting segmentation models to new environments. We present the most important semantic segmentation methods; we provide a comprehensive survey on domain adaptation techniques for semantic segmentation; we unveil newer trends such as multi-domain learning, domain generalization, test-time adaptation or source-free domain adaptation; we conclude this survey by describing datasets and benchmarks most widely used in semantic segmentation research. We hope that this survey will provide researchers across academia and industry with a comprehensive reference guide and will help them in fostering new research directions in the field.1 See as example the https://github.com/errolPereira/ Semi-Automatic-Image-Annotation-Tool 2 We refer to the dedicated page https://github.com/ heartexlabs/awesome-data-labeling for an up-to-date collection of annotation tools.

show abstract

“…This is a very difficult problem to solve as one has to overcome both the category and domain gap. Graphs Neural Networks (GNN) based approaches [3][4][5][6] have gained traction in this field due to their ability to find relations in unstructured data. It also helps to represent the domains in a unified subspace.…”

Section: Introductionmentioning

confidence: 99%

“…Recent methods [4][5][6] involving GNNs have gained popularity owing to their transductive ability for semantic propagation of related samples among multiple domains. Roy et al [3] proposes using graph convolution networks in 1SmT setting, along with co-curriculum teaching to handle noisy pseudo-labels. In our novel setting, we propose the use of attentional graph neural network [50] by passing messages between the source and target domains to aggregate them in a unified space.…”

Section: Introductionmentioning

confidence: 99%

Open-Set Multi-Source Multi-Target Domain Adaptation

Lal¹,

Arihant²,

Iyer³

et al. 2023

Preprint

View full text Add to dashboard Cite

Single-Source Single-Target Domain Adaptation (1S1T) aims to bridge the gap between a labelled source domain and an unlabelled target domain. Despite 1S1T being a well-researched topic, they are typically not deployed to the real world. Methods like Multi-Source Domain Adaptation and Multi-Target Domain Adaptation have evolved to model real-world problems but still do not generalise well. The fact that most of these methods assume a common label-set between source and target is very restrictive. Recent Open-Set Domain Adaptation methods handle unknown target labels but fail to generalise in multiple domains. To overcome these difficulties, first, we propose a novel generic domain adaptation (DA) setting named Open-Set Multi-Source Multi-Target Domain Adaptation (OS-nSmT), with n and m being number of source and target domains respectively. Next, we propose a graph attention based framework named DEGAA which can capture information from multiple source and target domains without knowing the exact label-set of the target. We argue that our method, though offered for multiple sources and multiple targets, can also be agnostic to various other DA settings. To check the robustness and versatility of DEGAA, we put forward ample experiments and ablation studies.35th Pre-registration workshop (NeurIPS 2021), Remote.

show abstract

Curriculum Graph Co-Teaching for Multi-Target Domain Adaptation

Cited by 55 publications

References 30 publications

Mixed-order self-paced curriculum learning for universal lesion detection

Mixed-order self-paced curriculum learning for universal lesion detection

Unsupervised Domain Adaptation for Semantic Image Segmentation: a Comprehensive Survey

Open-Set Multi-Source Multi-Target Domain Adaptation

Contact Info

Product

Resources

About