Semi-Supervised Learning for Fine-Grained Classification With Self-Training

Nartey, Obed Tettey; Yang, Guowu; Wu, Jinzhao; Asare, Sarpong Kwadwo

doi:10.1109/access.2019.2962258

Cited by 34 publications

(16 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This process continues until some pre-defined termination condition is reached. Self-training has previously been applied to several problems ranging from fine-grained classification [86] to parsing [87]. However, its capability has not been explored in the context of MOOCs analytics.…”

Section: Self-trainingmentioning

confidence: 99%

On the Effectiveness of Self-Training in MOOC Dropout Prediction

Goel

Goyal

2020

Open Computer Science

View full text Add to dashboard Cite

Massive open online courses (MOOCs) have gained enormous popularity in recent years and have attracted learners worldwide. However, MOOCs face a crucial challenge in the high dropout rate, which varies between 91%-93%. An interplay between different learning analytics strategies and MOOCs have emerged as a research area to reduce dropout rate. Most existing studies use click-stream features as engagement patterns to predict at-risk students. However, this study uses a combination of click-stream features and the influence of the learner’s friends based on their demographics to identify potential dropouts. Existing predictive models are based on supervised learning techniques that require the bulk of hand-labelled data to train models. In practice, however, scarcity of massive labelled data makes training difficult. Therefore, this study uses self-training, a semi-supervised learning model, to develop predictive models. Experimental results on a public data set demonstrate that semi-supervised models attain comparable results to state-ofthe-art approaches, while also having the flexibility of utilizing a small quantity of labelled data. This study deploys seven well-known optimizers to train the self-training classifiers, out of which, Stochastic Gradient Descent (SGD) outperformed others with the value of F1 score at 94.29%, affirming the relevance of this exposition.

show abstract

Section: Self-trainingmentioning

confidence: 99%

On the Effectiveness of Self-Training in MOOC Dropout Prediction

Goel

Goyal

2020

Open Computer Science

View full text Add to dashboard Cite

show abstract

“…It can be observed that the formulation in Equation 7was similar to the work proposed in [44]. However, the difference between the formulation in Equation 7and the optimization flow in [44] was that we introduced a class-wise bias by normalizing class-wise confidence levels compared to the use of an L 1 regularizer to prevent most pseudo-labels from being ignored by serving as a negative sparse term. The authors in [44] solved the pseudo-label framework optimizer by utilizing the solver in Equation (8)…”

Section: Self-training With Self-paced Curriculum Learningmentioning

confidence: 99%

“…From Equation (10), the pseudo-label generation and selection were not dependent on the prediction confidence level output as given in the solver by [44]. Instead, this was dependent on the class-wise normalized output p n (c|w,I u ) exp(−k c ) .…”

Section: Self-training With Self-paced Curriculum Learningmentioning

confidence: 99%

“…Fortunately, through semi-supervised and weakly-supervised learning, a robust semi-supervised traffic sign recognition can to some extent alleviate the costly and laboriousannotations by utilizing unlabeled images. Techniques such as those in [40][41][42][43][44] use self-training or similar concepts to utilize unlabeled samples for semi-supervised learning. A greedy unsupervised criterion has been used to generate and select the pseudo-labeled data for the retraining process of models.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Robust Semi-Supervised Traffic Sign Recognition via Self-Training and Weakly-Supervised Learning

Nartey

Yang

Asare

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

Traffic sign recognition is a classification problem that poses challenges for computer vision and machine learning algorithms. Although both computer vision and machine learning techniques have constantly been improved to solve this problem, the sudden rise in the number of unlabeled traffic signs has become even more challenging. Large data collation and labeling are tedious and expensive tasks that demand much time, expert knowledge, and fiscal resources to satisfy the hunger of deep neural networks. Aside from that, the problem of having unbalanced data also poses a greater challenge to computer vision and machine learning algorithms to achieve better performance. These problems raise the need to develop algorithms that can fully exploit a large amount of unlabeled data, use a small amount of labeled samples, and be robust to data imbalance to build an efficient and high-quality classifier. In this work, we propose a novel semi-supervised classification technique that is robust to small and unbalanced data. The framework integrates weakly-supervised learning and self-training with self-paced learning to generate attention maps to augment the training set and utilizes a novel pseudo-label generation and selection algorithm to generate and select pseudo-labeled samples. The method improves the performance by: (1) normalizing the class-wise confidence levels to prevent the model from ignoring hard-to-learn samples, thereby solving the imbalanced data problem; (2) jointly learning a model and optimizing pseudo-labels generated on unlabeled data; and (3) enlarging the training set to satisfy the hunger of deep learning models. Extensive evaluations on two public traffic sign recognition datasets demonstrate the effectiveness of the proposed technique and provide a potential solution for practical applications.

show abstract

“…Moreover, compared to the process of obtaining well-labeled data, unlabeled data is rather inexpensive and abundant. Semisupervised learning algorithms have been adopted in some works mentioned in the literature for some classification tasks [ 27 , 29 – 34 ].…”

Section: Introductionmentioning

confidence: 99%

A Semisupervised Learning Scheme with Self-Paced Learning for Classifying Breast Cancer Histopathological Images

Asare

You

Nartey

2020

Computational Intelligence and Neuroscience

Self Cite

View full text Add to dashboard Cite

The unavailability of large amounts of well-labeled data poses a significant challenge in many medical imaging tasks. Even in the likelihood of having access to sufficient data, the process of accurately labeling the data is an arduous and time-consuming one, requiring expertise skills. Again, the issue of unbalanced data further compounds the abovementioned problems and presents a considerable challenge for many machine learning algorithms. In lieu of this, the ability to develop algorithms that can exploit large amounts of unlabeled data together with a small amount of labeled data, while demonstrating robustness to data imbalance, can offer promising prospects in building highly efficient classifiers. This work proposes a semisupervised learning method that integrates self-training and self-paced learning to generate and select pseudolabeled samples for classifying breast cancer histopathological images. A novel pseudolabel generation and selection algorithm is introduced in the learning scheme to generate and select highly confident pseudolabeled samples from both well-represented classes to less-represented classes. Such a learning approach improves the performance by jointly learning a model and optimizing the generation of pseudolabels on unlabeled-target data to augment the training data and retraining the model with the generated labels. A class balancing framework that normalizes the class-wise confidence scores is also proposed to prevent the model from ignoring samples from less represented classes (hard-to-learn samples), hence effectively handling the issue of data imbalance. Extensive experimental evaluation of the proposed method on the BreakHis dataset demonstrates the effectiveness of the proposed method.

show abstract

Semi-Supervised Learning for Fine-Grained Classification With Self-Training

Cited by 34 publications

References 33 publications

On the Effectiveness of Self-Training in MOOC Dropout Prediction

On the Effectiveness of Self-Training in MOOC Dropout Prediction

Robust Semi-Supervised Traffic Sign Recognition via Self-Training and Weakly-Supervised Learning

A Semisupervised Learning Scheme with Self-Paced Learning for Classifying Breast Cancer Histopathological Images

Contact Info

Product

Resources

About