Co-training Model Research Articles

Co-training machine learning enables interpretable discovery of near-infrared phosphors with high performance

Near-infrared (NIR) phosphors based on Cr3+ doped garnets present great potential in the next generation of NIR light sources. Nevertheless, the huge searching space for the garnet composition makes the rapid discovery of NIR phosphors with high performance remain a great challenge for the scientific community. Herein, a generalizable machine learning (ML) strategy is designed to accelerate the exploration of innovative NIR phosphors via establishing the relationship between key parameters and emission peak wavelength (EPW). We propose a semi-supervised co-training model based on kernel ridge regression (KRR) and support vector regression (SVR), which successfully establishes an expanded dataset with unlabeled dataset (previously unidentified garnets), addressing the overfitting issue resulted from a small dataset and greatly improving the model generalization capability. The model is then interpreted to extract valuable insights into the contribution originated from different features. And a new type NIR luminescent material of Lu3Y2Ga3O12: Cr3+ (EPW~750 nm) is efficiently screened, which demonstrates a high internal (external) quantum efficiency of 97.1% (38.8%) and good thermal stability, particularly exhibiting promising application in the NIR phosphor-converted LEDs (pc-LED). These results suggest the strategy proposed in this work could provide new viewpoint and direction for developing NIR luminescence materials.

SC-SSL: Self-Correcting Collaborative and Contrastive Co-Training Model for Semi-Supervised Medical Image Segmentation.

Image segmentation achieves significant improvements with deep neural networks at the premise of a large scale of labeled training data, which is laborious to assure in medical image tasks. Recently, semi-supervised learning (SSL) has shown great potential in medical image segmentation. However, the influence of the learning target quality for unlabeled data is usually neglected in these SSL methods. Therefore, this study proposes a novel self-correcting co-training scheme to learn a better target that is more similar to ground-truth labels from collaborative network outputs. Our work has three-fold highlights. First, we advance the learning target generation as a learning task, improving the learning confidence for unannotated data with a self-correcting module. Second, we impose a structure constraint to encourage the shape similarity further between the improved learning target and the collaborative network outputs. Finally, we propose an innovative pixel-wise contrastive learning loss to boost the representation capacity under the guidance of an improved learning target, thus exploring unlabeled data more efficiently with the awareness of semantic context. We have extensively evaluated our method with the state-of-the-art semi-supervised approaches on four public-available datasets, including the ACDC dataset, M&Ms dataset, Pancreas-CT dataset, and Task_07 CT dataset. The experimental results with different labeled-data ratios show our proposed method's superiority over other existing methods, demonstrating its effectiveness in semi-supervised medical image segmentation.

Maternal and child oral health barriers and solutions: Case studies from South Africa, Brazil, and Indonesia.

The World Federation of Public Health Associations (WFPHA) has introduced a maternal oral health framework and launched its Global Maternal and Child Oral Health Initiative, advocating for maternal and child health as a strategic entry point for oral health to be integrated into primary healthcare systems. This article delves into maternal oral health initiatives in South Africa, Brazil, and Indonesia, exploring challenges and proposing context-specific solutions aligned with the WFPHA maternal oral health framework and its Global Maternal and Child Oral Health Initiative. The apartheid legacy in South Africa has led to the gross health inequalities, including oral health disparities characterized by maldistribution of dental facilities and dental professionals in the public sector. The proposed objectives focus on effective leadership, human resource capacity building, and integration of maternal and child oral health policy into existing health policies. Brazil faces challenges in reaching pregnant women, especially pregnant teens, due to workforce shortages. Brazil's Oral Health National Program mandates all pregnant women undergo at least one dental appointment, but a shortage of oral health teams persists and impedes timely access to dental care during pregnancy. Dental students' outreach to vulnerable pregnant teens can enhance the future oral health workforce's sensitivity to maternal oral health matters. In Indonesia, the Antenatal Care Innovative (iANC) Project has introduced a co-training model for midwives and oral health professionals addressing oral health needs of mothers and their spouses and fostering a comprehensive understanding of the interplay between oral health and maternal well-being. These country cases demonstrate the importance of tailored policies, health promotion, and care delivery interventions in maternal oral health. All cases highlight the importance of workforce training and distribution, fostering cross-disciplinary collaboration and future oral health workforce who can promote health beyond treatment delivery.

Semi-Supervised Co-Training Model Using Convolution and Transformer for Hyperspectral Image Classification

Semi-Supervised Co-Training Model Using Convolution and Transformer for Hyperspectral Image Classification

Underwater Noise Target Recognition Based on Sparse Adversarial Co-Training Model with Vertical Line Array

Underwater Noise Target Recognition Based on Sparse Adversarial Co-Training Model with Vertical Line Array

Three-Way Co-Training with Pseudo Labels for Semi-Supervised Learning

The theory of three-way decision has been widely utilized across various disciplines and fields as an efficient method for both knowledge reasoning and decision making. However, the application of the three-way decision theory to partially labeled data has received relatively less attention. In this study, we propose a semi-supervised co-training model based on the three-way decision and pseudo labels. We first present a simple yet effective method for producing two views by assigning pseudo labels to unlabeled data, based on which a heuristic attribute reduction algorithm is developed. The three-way decision is then combined with the concept of entropy to form co-decision rules for classifying unlabeled data into useful, uncertain, or useless samples. Finally, some useful samples are iteratively selected to improve the performance of the co-decision model. The experimental results on UCI datasets demonstrate that the proposed model outperforms other semi-supervised models, exhibiting its potential for partially labeled data.

MP41-13 QUANTITATIVE ASSESSMENT OF MESENCHYMAL DIFFERENTIATION FOR METASTATIC RISK PREDICTION OF CLEAR CELL RENAL CARCINOMA (CCRCC)

You have accessJournal of UrologyCME1 Apr 2023MP41-13 QUANTITATIVE ASSESSMENT OF MESENCHYMAL DIFFERENTIATION FOR METASTATIC RISK PREDICTION OF CLEAR CELL RENAL CARCINOMA (CCRCC) Daniel Vilchez, Bodong Zhang, Alejandro Sanchez, Deepika Sirohi, Tolga Tasdizen, and Beatrice Knudsen Daniel VilchezDaniel Vilchez More articles by this author , Bodong ZhangBodong Zhang More articles by this author , Alejandro SanchezAlejandro Sanchez More articles by this author , Deepika SirohiDeepika Sirohi More articles by this author , Tolga TasdizenTolga Tasdizen More articles by this author , and Beatrice KnudsenBeatrice Knudsen More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000003279.13AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Epithelial-to-mesenchymal transition (EMT) results in poor prognosis sarcomatoid and rhabdoid histologic growth patterns (HGP) of ccRCC. However, only ∼15% of metastatic cases reveal areas of sarcomatoid/rhabdoid differentiation in the primary tumor. We questioned whether a computer can learn salient features of EMT that are associated with risk of metastatic progression across all HGPs of ccRCC. METHODS: The project uses 410 digital H&E slides of ccRCC. We incorporated a novel co-training approach into a machine learning model (MLM) and trained the model to distinguish epithelial (nested) and mesenchymal (sarcomatoid, rhabdoid) HGPs in a Utah cohort. The trained MLM is applied to the TCGA KIRC cohort (n=360) with annotated regions of nested, tubulo-papillary, low grade-spindle, sarcomatoid and rhabdoid HGPs for which the hazard ration of metastatic progression was published. For each case, the MLM outputs a mesenchymal-differentiation (DP-MD) score on a scale of 0 to 1 where closer to 1 is higher likelihood of EMT. RESULTS: A unique machine learning / AI model was developed to quantify sarcomatoid differentiation in ccRCC with no obvious histologic signs of mesenchymal morphology. H&E stains were used as separate inputs of a contrastive co-training model (H/E-CCTM) to force the learning of mesenchymal features. The inclusion of a semi-supervised learning component improved the accuracy of predicting mesenchymal differentiation by >4% over other state-of-the-art co-training algorithms and performed equally well when trained on 10% versus 100% of labeled cancer data. Sarcomatoid, low grade spindle and rhabdoid cases received average DP-MD scores of 0.91, 0.97 and 0.83, respectively in a held-out test set, while cases with nested morphology, including expansile, and high nuclear to cytoplasmic ratio averaged DP-MD scores of 0.08. Tubulo-papillary HGPs without obvious mesenchymal features obtained an average DP-MD score of 0.49. CONCLUSIONS: Computer-assisted morphometric analysis can be used to quantify mesenchymal differentiation and has the potential to improve the prediction of metastatic risk in ccRCC and the selection of patients for adjuvant treatment after nephrectomy. Source of Funding: Huntsman Cancer Foundation, NIH P30CA042014 © 2023 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 209Issue Supplement 4April 2023Page: e560 Advertisement Copyright & Permissions© 2023 by American Urological Association Education and Research, Inc.MetricsAuthor Information Daniel Vilchez More articles by this author Bodong Zhang More articles by this author Alejandro Sanchez More articles by this author Deepika Sirohi More articles by this author Tolga Tasdizen More articles by this author Beatrice Knudsen More articles by this author Expand All Advertisement PDF downloadLoading ...

CNN-Based Broad Learning for Cross-Domain Emotion Classification

Cross-domain emotion classification aims to leverage useful information in a source domain to help predict emotion polarity in a target domain in a unsupervised or semi-supervised manner. Due to the domain discrepancy, an emotion classifier trained on source domain may not work well on target domain. Many researchers have focused on traditional cross-domain sentiment classification, which is coarse-grained emotion classification. However, the problem of emotion classification for cross-domain is rarely involved. In this paper, we propose a method, called convolutional neural network (CNN) based broad learning, for cross-domain emotion classification by combining the strength of CNN and broad learning. We first utilized CNN to extract domain-invariant and domain-specific features simultaneously, so as to train two more efficient classifiers by employing broad learning. Then, to take advantage of these two classifiers, we designed a co-training model to boost together for them. Finally, we conducted comparative experiments on four datasets for verifying the effectiveness of our proposed method. The experimental results show that the proposed method can improve the performance of emotion classification more effectively than those baseline methods.

Multi-source domain adaptation method for textual emotion classification using deep and broad learning

Existing domain adaptation methods for classifying textual emotions have the propensity to focus on single-source domain exploration rather than multi-source domain adaptation. The efficacy of emotion classification is hampered by the restricted information and volume from a single source domain. Thus, to improve the performance of domain adaptation, we present a novel multi-source domain adaptation approach for emotion classification, by combining broad learning and deep learning in this article. Specifically, we first design a model to extract domain-invariant features from each source domain to the same target domain by using BERT and Bi-LSTM, which can better capture contextual features. Then we adopt broad learning to train multiple classifiers based on the domain-invariant features, which can more effectively conduct multi-label classification tasks. In addition, we design a co-training model to boost these classifiers. Finally, we carry out several experiments on four datasets by comparison with the baseline methods. The experimental results show that our proposed approach can significantly outperform the baseline methods for textual emotion classification.

Self-paced multi-label co-training

Multi-label learning aims to solve classification problems where instances are associated with a set of labels. In reality, it is generally easy to acquire unlabeled data but expensive or time-consuming to label them, and this situation becomes more serious in multi-label learning as an instance needs to be annotated with several labels. Hence, semi-supervised multi-label learning approaches emerge as they are able to exploit unlabeled data to help train predictive models. This work proposes a novel approach called Self-paced Multi-label Co-Training (SMCT). It leverages the well-known co-training paradigm to iteratively train two classifiers on two views of a dataset and communicate one classifier’s predictions on unlabeled data to augment the other’s training set. As pseudo labels may be false in iterative training, self-paced learning is integrated into SMCT to rectify false pseudo labels and avoid error accumulation. Concretely, the multi-label co-training model in SMCT is formulated as an optimization problem by introducing latent weight variables of unlabeled instances. It is then solved via an alternative convex optimization algorithm. Experimental evaluations are carried out based on six benchmark multi-label datasets and three metrics. The results demonstrate that SMCT is very competitive in each setting when compared with five state-of-the-art methods.

A semi-supervised co-training model for predicting passenger flow change in expanding subways

Subways in many big cities are experiencing rapid development with new lines being planned, built and put into operations. Understanding passenger flow change in expanding subways is of significant importance, because it is a crucial piece of information for the new networks’ planning and operation. Here, we employ large-scale smart card data to simulate the change of section passenger flow in an actual expanding subway. We find that the geographic properties of new subway lines can influence the passenger flows of existing sections. Based on the empirical findings, we develop a semi-supervised co-training (S-MLR-XGBoost) model to solve the problem of insufficient training samples and predict passenger flow change in the expanding subway. The proposed S-MLR-XGBoost model captures the relationship between the geographic properties of new subway lines and the passenger flow change in existing lines and achieves higher prediction accuracy comparing with several benchmark models. Finally, sensitivity analysis of the proposed model and possible reasons for passenger flow change in expanding subways are discussed.

Exploring the Status of Co-training as a Professional Development Tool (Omani English Teacher-trainers’ Perspectives)

This study investigates the status of co-training as a tool for the PD of English teacher-trainers (TTs) in Oman. Since these TTs adopt this tool even though they have received no training on it, it is significant to explore their knowledge and practice to explore if there are any discrepancies in their performance. An open-ended questionnaire is used to collect data. Fourteen (14) English TTs from 9 governorates responded to the questionnaire. Their responses were analysed, categorised and presented. The findings revealed both, performance discrepancies and the causes of these discrepancies. The study showed some weakness in the practice and orientation of co-training as a tool for PD among the participants. The findings revealed that there is a fluctuation in the value that TTs place on co-training and its implementation. These also highlighted different practices of co-training reflecting a lack of understanding of how to implement it. Moreover, some inappropriate practices were identified which put the adoption of this collaborative tool at risk. The findings showed that these discrepancies existed because of a lack of knowledge, inappropriate practices and negative perceptions towards co-training. Participants’ suggestions for better future practices of co-training highlighted the strong need to improve the situation by clarifying the ambiguity and setting out conditions to follow. This study recommended constructing a co-training model and involving TTs in a training programme to raise their knowledge of co-training and foster their perception of it as a tool for their PD.

How and when to stop the co-training process

Co-training is a semi-supervised learning approach used when only a small set of the data that is available for training is labeled. By using multiple classifiers, the co-training process utilizes the small set of labeled data in order to label an additional set of samples. During this process, the classifiers gradually augment the training data in an iterative process in which a new co-training model is derived and used for labeling the unlabeled samples in each iteration. A few of the newly labeled samples are added in each iteration to the training dataset to improve the performance of the classifiers. The main challenge in applying co-training is to make sure that the co-trainer assigns accurate labels to the unlabeled samples. Many empirical studies showed that the performance (accuracy) of the co-trainer could not be further improved when a certain number of iterations was reached, and in some cases, the performance even declined if the process (i.e., labeling) continued. Despite this, no general solution has been suggested for identifying the optimal final co-training model or number of iterations before this decline. In this work, we propose a novel method aimed at selecting the near-optimal final co-training model among all models created in the various iterations according to a predefined measurement based solely on the unlabeled data. Experiments on nine open, publicly available and real-life datasets demonstrate that the proposed method outputs a near-optimal final co-training model compared to other co-training models created in the various iterations.

Co-Training for Visual Object Recognition Based on Self-Supervised Models Using a Cross-Entropy Regularization.

Automatic recognition of visual objects using a deep learning approach has been successfully applied to multiple areas. However, deep learning techniques require a large amount of labeled data, which is usually expensive to obtain. An alternative is to use semi-supervised models, such as co-training, where multiple complementary views are combined using a small amount of labeled data. A simple way to associate views to visual objects is through the application of a degree of rotation or a type of filter. In this work, we propose a co-training model for visual object recognition using deep neural networks by adding layers of self-supervised neural networks as intermediate inputs to the views, where the views are diversified through the cross-entropy regularization of their outputs. Since the model merges the concepts of co-training and self-supervised learning by considering the differentiation of outputs, we called it Differential Self-Supervised Co-Training (DSSCo-Training). This paper presents some experiments using the DSSCo-Training model to well-known image datasets such as MNIST, CIFAR-100, and SVHN. The results indicate that the proposed model is competitive with the state-of-art models and shows an average relative improvement of 5% in accuracy for several datasets, despite its greater simplicity with respect to more recent approaches.

Grey theory–based BP-NN co-training for dense sequence long-term tendency prediction

PurposeThe purpose of this paper is to solve the problems existing in topic popularity prediction in online social networks and advance a fine-grained and long-term prediction model for lack of sufficient data.Design/methodology/approachBased on GM(1,1) and neural networks, a co-training model for topic tendency prediction is proposed in this paper. The interpolation based on GM(1,1) is employed to generate fine-grained prediction values of topic popularity time series and two neural network models are considered to achieve convergence by transmitting training parameters via their loss functions.FindingsThe experiment results indicate that the integrated model can effectively predict dense sequence with higher performance than other algorithms, such as NN and RBF_LSSVM. Furthermore, the Markov chain state transition probability matrix model is used to improve the prediction results.Practical implicationsFine-grained and long-term topic popularity prediction, further improvement could be made by predicting any interpolation in the time interval of popularity data points.Originality/valueThe paper succeeds in constructing a co-training model with GM(1,1) and neural networks. Markov chain state transition probability matrix is deployed for further improvement of popularity tendency prediction.

A Co-Training Model with Label Propagation on a Bipartite Graph to Identify Online Users with Disabilities

Collecting data from representative users with disabilities for accessibility research is time and resource consuming. With the proliferation of social media websites, many online spaces have emerged for people with disabilities. The information accumulated in such places is of great value for data collection and participant recruiting. However, there are also many active non-representative users in such online spaces such as medical practitioners, caretakers, or family members. In this work, we introduce a novel co-training model based on the homophily phenomenon observed among online users with the same disability. The model combines a variational label propagation algorithm and a naive Bayes classifier to identify online users who have the same disability. We evaluated this model on a dataset collected from Reddit and the results show improvements over traditional models.

A Remote Sensing Ship Recognition Method Based on Co-Training Model

Aiming at detecting sea targets efficiently, an approach using optical remote sensing data based on co-training model is proposed. Firstly, using size, texture, shape, moment invariants features and ratio codes, feature extraction is realized. Secondly, based on rough set theory, the common discernibility degree is used to select valid recognition features automatically. Finally, a co-training model for classification is introduced. Firstly, two diverse ruducts are generated, and then the model employs them to train two base classifiers on labeled dada, and makes two base classifiers teach each other on unlabeled data to boot their performance iteratively. Experimental results show the proposed approach can get better performance than K-Nearest Neighbor (KNN), Support Vector Machines (SVM), traditional hierarchical discriminant regression (HDR).

A Neighborhood Rough Sets-Based Co-Training Model for Classification

A Neighborhood Rough Sets-Based Co-Training Model for Classification

Supervised learning and Co-training

Co-training under the Conditional Independence Assumption is among the models which demonstrate how radically the need for labeled data can be reduced if a huge amount of unlabeled data is available. In this paper, we explore how much credit for this saving must be assigned solely to the extra assumptions underlying the Co-training Model. To this end, we compute general (almost tight) upper and lower bounds on the sample size needed to achieve the success criterion of PAC-learning in the realizable case within the model of Co-training under the Conditional Independence Assumption in a purely supervised setting. The upper bounds lie significantly below the lower bounds for PAC-learning without Co-training. Thus, Co-training saves labeled data even when not combined with unlabeled data. On the other hand, the saving is much less radical than the known savings in the semi-supervised setting.

Diverse reduct subspaces based co-training for partially labeled data

Rough set theory is an effective supervised learning model for labeled data. However, it is often the case that practical problems involve both labeled and unlabeled data, which is outside the realm of traditional rough set theory. In this paper, the problem of attribute reduction for partially labeled data is first studied. With a new definition of discernibility matrix, a Markov blanket based heuristic algorithm is put forward to compute the optimal reduct of partially labeled data. A novel rough co-training model is then proposed, which could capitalize on the unlabeled data to improve the performance of rough classifier learned only from few labeled data. The model employs two diverse reducts of partially labeled data to train its base classifiers on the labeled data, and then makes the base classifiers learn from each other on the unlabeled data iteratively. The classifiers constructed in different reduct subspaces could benefit from their diversity on the unlabeled data and significantly improve the performance of the rough co-training model. Finally, the rough co-training model is theoretically analyzed, and the upper bound on its performance improvement is given. The experimental results show that the proposed model outperforms other representative models in terms of accuracy and even compares favorably with rough classifier trained on all training data labeled.