95,208 Hits in 7.2 sec

An Empirical Study and Analysis on Open-Set Semi-Supervised Learning [article]

Huixiang Luo, Hao Cheng, Fanxu Meng, Yuting Gao, Ke Li, Mengdan Zhang, Xing Sun
2021 arXiv   pre-print
Pseudo-labeling (PL) and Data Augmentation-based Consistency Training (DACT) are two approaches widely used in Semi-Supervised Learning (SSL) methods.  ...  These methods exhibit great power in many machine learning tasks by utilizing unlabeled data for efficient training.  ...  This setting is firstly introduced by , and is named as "Open-Set Semi-Supervised Learning" (open-set SSL, illustrated in Figure 1 ).  ... 
arXiv:2101.08237v2 fatcat:pw6beprzyzhpjlz7xzerxplvqy

Enhancing educational data mining techniques on online educational resources with a semi-supervised learning approach

Vincent Tam, Edmund Y. Lam, S. T. Fung, W. W. T. Fok, Allan H. K. Yuen
2015 2015 IEEE International Conference on Teaching, Assessment, and Learning for Engineering (TALE)  
In this work, we propose a systematic framework of a semi-supervised learning approach in which a concept-based classifier is co-trained with an explicit semantic analysis (ESA) classifier to derive a  ...  EDM is generally more interested in automated methods for discovery within the educational data while LA is relatively keen on applying human-led methods to understand the involved learning processes.  ...  ACKNOWLEDGMENT The authors are grateful to Professor Daniel Churchill, Dr. Kinshuk and Professor Henry Chan for their valuable inputs.  ... 
doi:10.1109/tale.2015.7386044 dblp:conf/tale/TamLFFY15 fatcat:dmerijugrbcy3j66yzfuzk2gly

OpenMatch: Open-set Consistency Regularization for Semi-supervised Learning with Outliers [article]

Kuniaki Saito, Donghyun Kim, Kate Saenko
2021 arXiv   pre-print
To address this problem, we propose a novel Open-set Semi-Supervised Learning (OSSL) approach called OpenMatch.  ...  Semi-supervised learning (SSL) is an effective means to leverage unlabeled data to improve a model's performance.  ...  We utilize the training split of ImageNet-30 for training and test one for evaluation.  ... 
arXiv:2105.14148v2 fatcat:xt5hthd3unc6jemfn55bwr3d5e

Feature-enhanced Adversarial Semi-supervised Semantic Segmentation Network for Pulmonary Embolism Annotation [article]

Ting-Wei Cheng, Jerry Chang, Ching-Chun Huang, Chin Kuo, Yun-Chien Cheng
2022 arXiv   pre-print
We used the labeled open-source dataset and the unlabeled National Cheng Kung University Hospital (NCKUH) (IRB number: B-ER-108-380) dataset to train the semi-supervised learning model, and the resulting  ...  In current studies, all of the PE CTPA image segmentation methods are trained by supervised learning.  ...  Open source dataset supervised model results The pre-trained model for semi-supervised learning were trained by open source dataset.  ... 
arXiv:2204.04217v1 fatcat:2leh2smjnzerre2upymsbmnfxa

Trash to Treasure: Harvesting OOD Data with Cross-Modal Matching for Open-Set Semi-Supervised Learning [article]

Junkai Huang, Chaowei Fang, Weikai Chen, Zhenhua Chai, Xiaolin Wei, Pengxu Wei, Liang Lin, Guanbin Li
2021 arXiv   pre-print
Open-set semi-supervised learning (open-set SSL) investigates a challenging but practical scenario where out-of-distribution (OOD) samples are contained in the unlabeled data.  ...  While the mainstream technique seeks to completely filter out the OOD samples for semi-supervised learning (SSL), we propose a novel training mechanism that could effectively exploit the presence of OOD  ...  Open-set Semi-supervised Learning. Recently, researchers have gradually focused on settling the open-set SSL problem.  ... 
arXiv:2108.05617v1 fatcat:e3hkqoboq5ewnodvphe5unkl3a

OSSGAN: Open-Set Semi-Supervised Image Generation [article]

Kai Katsumata and Duc Minh Vo and Hideki Nakayama
2022 arXiv   pre-print
We introduce a challenging training scheme of conditional GANs, called open-set semi-supervised image generation, where the training dataset consists of two parts: (i) labeled data and (ii) unlabeled data  ...  Unlike the existing semi-supervised image generation task, where unlabeled data only contain closed-set samples, our task is more general and lowers the data collection cost in practice by allowing open-set  ...  Supplementary Material for OSSGAN: Open-Set Semi-Supervised Image Generation I.  ... 
arXiv:2204.14249v1 fatcat:fngomvxnq5hdhh4yktjlezum44

A*HAR: A New Benchmark towards Semi-supervised learning for Class-imbalanced Human Activity Recognition [article]

Govind Narasimman, Kangkang Lu, Arun Raja, Chuan Sheng Foo, Mohamed Sabry Aly, Jie Lin, Vijay Chandrasekhar
2021 arXiv   pre-print
We evaluate state-of-the-art semi-supervised learning method on A*HAR, by combining Mean Teacher and Convolutional Neural Network.  ...  These findings lead to an interesting open problem, i.e., development of semi-supervised HAR algorithms that are class-imbalance aware without any prior knowledge on the class distribution for unlabeled  ...  Consistency-based method has been one of the most widely used semi-supervised learning methods.  ... 
arXiv:2101.04859v1 fatcat:pwxk7wyl7jbm5e3s44d7552ecm

OpenCoS: Contrastive Semi-supervised Learning for Handling Open-set Unlabeled Data [article]

Jongjin Park, Sukmin Yun, Jongheon Jeong, Jinwoo Shin
2022 arXiv   pre-print
Semi-supervised learning (SSL) is one of the most promising paradigms to circumvent the expensive labeling cost for building a high-performance model.  ...  Our extensive experimental results show the effectiveness of OpenCoS, fixing up the state-of-the-art semi-supervised methods to be suitable for diverse scenarios involving open-set unlabeled data.  ...  Algorithm 1 1 OpenCoS: A general framework for open-set semi-supervised learning (SSL).  ... 
arXiv:2107.08943v2 fatcat:yeimauemc5dd5ezaho2qz2hgxu

Semi-supervised Learning of Galaxy Morphology using Equivariant Transformer Variational Autoencoders [article]

Mizu Nishikawa-Toomey, Lewis Smith, Yarin Gal
2020 arXiv   pre-print
However, by leveraging the information present in the ever growing set of unlabelled images, semi-supervised learning could be an effective way of reducing the required labelling and increasing classification  ...  We show that this novel architecture leads to improvements in accuracy when used for the galaxy morphology classification task on the Galaxy Zoo data set.  ...  This opens up many avenues for architectures for semi-supervised learning compared to other generative models such as Generative Adversarial Networks (GANs) [7] that only model the generative distribution  ... 
arXiv:2011.08714v1 fatcat:nhfd3kkjlngr3c56w26g3trdla

Open set learning with augmented category by exploiting unlabelled data (open-LACU) [article]

Emile R. Engelbrecht, Johan A. du Preez
2022 arXiv   pre-print
This research is the first to generalize between observed and unobserved novel categories within a new learning policy called open-set learning with augmented category by exploiting unlabeled data or open-LACU  ...  In contrast, open-sets define novel categories as those unobserved during during training, but present during testing.  ...  learning, OSR is open-set recognition, SSL is semi-supervised learning, MSSL is mismatched semi-supervised learning with USSL being universal semi-supervised learning, LACU is learning with augmented category  ... 
arXiv:2002.01368v4 fatcat:pbyuj7lbcjc3jjldp7oih3slye

Learning with Limited Annotations: A Survey on Deep Semi-Supervised Learning for Medical Image Segmentation [article]

Rushi Jiao, Yichi Zhang, Le Ding, Rong Cai, Jicong Zhang
2022 arXiv   pre-print
Semi-supervised learning has emerged as an appealing strategy and been widely applied to medical image segmentation tasks to train deep models with limited annotations.  ...  In this paper, we present a comprehensive review of recently proposed semi-supervised learning methods for medical image segmentation and summarized both the technical novelties and empirical results.  ...  ACKNOWLEDGMENTS This paper was supported in part by the National Science Foundation under Grant 32000687, and in part by the University Synergy Innovation Program of Anhui Province under Grant GXXT-2019  ... 
arXiv:2207.14191v1 fatcat:k47z5cbqbvhp7lzzhhfxtpt2wa

SODA10M: A Large-Scale 2D Self/Semi-Supervised Object Detection Dataset for Autonomous Driving [article]

Jianhua Han, Xiwen Liang, Hang Xu, Kai Chen, Lanqing Hong, Jiageng Mao, Chaoqiang Ye, Wei Zhang, Zhenguo Li, Xiaodan Liang, Chunjing Xu
2021 arXiv   pre-print
Aiming at facilitating a real-world, ever-evolving and scalable autonomous driving system, we present a large-scale dataset for standardizing the evaluation of different self-supervised and semi-supervised  ...  Experiments show that SODA10M can serve as a promising pre-training dataset for different self-supervised learning methods, which gives superior performance when fine-tuning with different downstream tasks  ...  We thank our two data suppliers, named Testin 3 and Speechocean 4 (collected from King-IM-055), for helping us collect and annotate SODA10M dataset.  ... 
arXiv:2106.11118v3 fatcat:wtypjknlrbc4vp7yrygit4qxna

AuxMix: Semi-Supervised Learning with Unconstrained Unlabeled Data [article]

Amin Banitalebi-Dehkordi, Pratik Gujjar, Yong Zhang
2022 arXiv   pre-print
Semi-supervised learning (SSL) has seen great strides when labeled data is scarce but unlabeled data is abundant.  ...  We also propose to regularize learning by maximizing the predicted entropy for dissimilar auxiliary samples.  ...  In phase two, the positive set D + a is used in semi-supervised learning based on consistency regularization across augmentations [37] .  ... 
arXiv:2206.06959v1 fatcat:droef274kvbu7gmdetioofv4dq


E. Bousias Alexakis, C. Armenakis
2021 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
The approach is based on the Mean Teacher method, a semi-supervised approach, successfully applied for image classification and for sematic segmentation of medical images.  ...  In this work we experiment with the implementation of a semi-supervised training approach in an attempt to improve the image semantic segmentation performance of models trained using a small number of  ...  The Change Detection Dataset was provided by Lebedev et al., 2018: to-nHrNs9  ... 
doi:10.5194/isprs-archives-xliii-b3-2021-829-2021 fatcat:de3tnxo6sndlzayzc2zi5kctme

Semi-Supervised Representation Learning via Triplet Loss Based on Explicit Class Ratio of Unlabeled Data

2022 IEICE transactions on information and systems  
Combined with a consistency regularization-based method, our method achieves more accurate semi-supervised learning.  ...  In this paper, we propose a semi-supervised triplet loss function that realizes semi-supervised representation learning in a novel manner.  ...  That is, our method utilizes the unlabeled samples for representation learning. Related Work A number of semi-supervised learning methods based on deep learning have been proposed [19] .  ... 
doi:10.1587/transinf.2021edp7073 fatcat:c6eu6fmr4zf7hakdz47mbsnsv4
« Previous Showing results 1 — 15 out of 95,208 results