Determining feature extractors for unsupervised learning on satellite images

Automated detection of leukemia by pretrained deep neural networks and transfer learning: A comparison

The evolution and improvements of deep learning are being used to tackle any research obstacles that could be converted into classification problems in all spheres of life. Each Deep convolutional neural network (DCNN) design's output is determined by the depth and value of the hyperparameters, which explains why so many of them have been proposed. These DCNN architectures must be created entirely from scratch, and they can only be used for the applications for which they were intended. Transfer learning may be used to modify these pre-trained networks so they are more appropriate for particular purposes. This article aims to evaluate the empirical performance of the applicability of pre-trained DCNN models to identify human face presentation threats (FPAD). Human FPAD is one of the most significant and crucial areas of research right now because of the introduction of ambient computing, which necessitates contact-free identification of persons with the help of their biometric traits. Six pre-trained DCNN models are taken into account for an experimental evaluation in human FPAD alias VGG19, VGG16, DensNet121, MobileNet, Xception, and InceptionV3. The investigation makes use of the NUAA and Replay-Attack benchmark FPAD datasets. Thepade's sorted block truncation coding (SBTC) 10-ary features are merged with deep learning features produced from the finest performing finetuned DCNNs to enhance the FPAD capabilities of analyzed machine learning (ML) classifiers. The integration of features of Thepade's SBTC 10-ary and DCNN has considerably increased the FPAD accuracy of ML classifiers with slightly more computations of feature extraction.

Female accounts for approximately 50% of the total population worldwide and many of them had breast cancer. Computer-aided diagnosis frameworks could reduce the number of needless biopsies and the workload of radiologists. This research aims to detect benign and malignant tumors automatically using breast ultrasound (BUS) images. Accordingly, two pretrained deep convolutional neural network (CNN) models were employed for transfer learning using BUS images like AlexNet and DenseNet201. A total of 697 BUS images containing benign and malignant tumors are preprocessed and performed classification tasks using the transfer learning-based CNN models. The classification accuracy of the benign and malignant tasks is completed and achieved 92.8% accuracy using the DensNet201 model. The results thus achieved compared in state of the art using benchmark data set and concluded proposed model outperforms in accuracy from first stage breast tumor diagnosis. Finally, the proposed model could help radiologists diagnose benign and malignant tumors swiftly by screening suspected patients.

In this paper, we propose a method to reduce the number of parameters based on a pre-trained deep convolutional neural network (DNN) for a facial expression recognition (FER) task. In order to maintain the high level of accuracy of the pre-trained network and save computational resources through reducing the number of parameters, we first build and train a high accuracy DNN model and extract the layers which contain the most representative facial expression features; then, we connect these layers with lightweight depthwise separable convolutions and global max pooling layers as the re-training network. We train and compare the re-training network with different DNN architectures on two FER datasets (FER2013, AffectNet). The results demonstrate that our method retains its pre-trained high accuracy with fewer model parameters than current state-of-the-art DNN methods.

Assess the efficacy of deep convolutional neural networks (DCNNs) in detection of critical enteric feeding tube malpositions on radiographs. 5475 de-identified HIPAA compliant frontal view chest and abdominal radiographs were obtained, consisting of 174 x-rays of bronchial insertions and 5301 non-critical radiographs, including normal course, normal chest, and normal abdominal x-rays. The ground-truth classification for enteric feeding tube placement was performed by two board-certified radiologists. Untrained and pretrained deep convolutional neural network models for Inception V3, ResNet50, and DenseNet 121 were each employed. The radiographs were fed into each deep convolutional neural network, which included untrained and pretrained models. The Tensorflow framework was used for Inception V3, ResNet50, and DenseNet. Images were split into training (4745), validation (630), and test (100). Both real-time and preprocessing image augmentation strategies were performed. Receiver operating characteristic (ROC) and area under the curve (AUC) on the test data were used to assess the models. Statistical differences among the AUCs were obtained. p < 0.05 was considered statistically significant. The pretrained Inception V3, which had an AUC of 0.87 (95 CI; 0.80–0.94), performed statistically significantly better (p < .001) than the untrained Inception V3, with an AUC of 0.60 (95 CI; 0.52–0.68). The pretrained Inception V3 also had the highest AUC overall, as compared with ResNet50 and DenseNet121, with AUC values ranging from 0.82 to 0.85. Each pretrained network outperformed its untrained counterpart. (p < 0.05). Deep learning demonstrates promise in differentiating critical vs. non-critical placement with an AUC of 0.87. Pretrained networks outperformed untrained ones in all cases. DCNNs may allow for more rapid identification and communication of critical feeding tube malpositions.

Due to the particularity of spatial structure in geographic data and the inadequacy of labeled data, the deep learning features can not reflect the spatial information and scale changes of objects in land-use/land-cover (LULC) images. To overcome these limitations, a feature fusion method is proposed to obtain well-discriminative and low-dimensional features, which can effectively fuse the features extracted from different levels. Given the inadequacy of labeled data, a pre-trained deep convolution neural network is proposed and used as a feature extractor to obtain the semantic features of images. To overcome the variety of scale and rotation in LULC images, the scale-invariant feature transform feature is introduced to represent the spatial structure information of the LULC images. Then different features are fused for LULC classification. In order to effectively distinguish easily-confused categories, an improved discriminant correlation analysis method based on the distance of scatter matrix between different categories is designed. The distance coefficient minimizes the correlation among categories. The proposed methods are applied on the UC Merced land use dataset, and achieve outstanding classification performance with dozens of feature dimensions under the case of limited training samples. The results demonstrated that the proposed method outperform several state-of-the-art methods.

This study focuses on the detection of COVID-19 using chest X-ray images. With the rapid global spread of the highly contagious disease, early and accurate detection is crucial to prevent further transmission. We propose a multi-view approach that leverages different deep-learning methods to detect COVID-19 based on chest X-ray images. The framework presented in this study aims to capture both complementary and correlative information from multiple views. By using pre-trained deep convolutional neural network (CNN) models such as ResNet50 and VGG16 to extract deep features from the X-ray images. Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) algorithms are then employed to classify the X-ray images based on the learned feature representations. The single-view models using VGG16 and ResNet50 with SVM achieved accuracy scores of 97% and 98% respectively. Similarly, using KNN, the single-view models achieved accuracy scores of 89% and 95%. Recognizing the potential of multi-view learning to improve generalization performance. We utilized early fusion to integrate the outputs of the pre-trained models and fed them into SVM and KNN classifiers. The multi-view model achieved accuracy scores of 98% and 92% respectively. The experimental results demonstrate that the multi-view deep learning methods outperformed the single-view deep models in detecting COVID-19 from chest X-ray images. The findings of this study have practical implications as they can assist experts in early diagnosis of COVID-19 cases, highlighting the effectiveness of the proposed methods.

Malignant melanoma is the deadliest form of skin cancer and, in recent years, is rapidly growing in terms of the incidence worldwide rate. The most effective approach to targeted treatment is early diagnosis. Deep learning algorithms, specifically convolutional neural networks, represent a methodology for the image analysis and representation. They optimize the features design task, essential for an automatic approach on different types of images, including medical. In this paper, we adopted pretrained deep convolutional neural networks architectures for the image representation with purpose to predict skin lesion melanoma. Firstly, we applied a transfer learning approach to extract image features. Secondly, we adopted the transferred learning features inside an ensemble classification context. Specifically, the framework trains individual classifiers on balanced subspaces and combines the provided predictions through statistical measures. Experimental phase on datasets of skin lesion images is performed and results obtained show the effectiveness of the proposed approach with respect to state-of-the-art competitors.

The human tongue has been long believed to be a window to provide important insights into a patient's health in medicine. The present study introduced a novel approach to predict patient age, gender, and weight inferences based on tongue images using pretrained deep convolutional neural networks (CNNs). Our results demonstrated that the deep CNN models (e.g., ResNeXt) trained on dorsal tongue images produced excellent results for age prediction with a Pearson correlation coefficient of 0.71 and a mean absolute error (MAE) of 8.5 years. We also obtained an excellent classification of gender, with a mean accuracy of 80% and an AUC (area under the receiver operating characteristic curve) of 88%. ResNeXt model also obtained a moderate level of accuracy for weight prediction, with a Pearson correlation coefficient of 0.39 and a MAE of 9.06 kg. These findings support our hypothesis that the human tongue contains crucial information about a patient. This study demonstrated the feasibility of using the pretrained deep CNNs along with a large tongue image dataset to develop computational models to predict patient medical conditions for noninvasive, convenient, and inexpensive patient health monitoring and diagnosis.

Major Depressive Disorder (MDD) is a widespread global mental disease. The Effectiveness of Selective Serotonin Reuptake Inhibitors (SSRIs) antidepressants prescribed for MDD patients is limited and pre-treatment assessment of treatment outcome is a vital task. In this study, a hybrid model based on Transfer Learning (TL) of powerful pre-trained deep Convolutional Neural Networks (CNNs) empowered with bidirectional Long Short-term Memory (BLSTM) cells and attention mechanism were developed to classify Responders (R) and Non-Responders (NR) to SSRI antidepressants, using raw data images of pre-treatment EEG signal obtained from 30 MDD patients. TL-LSTM-Attention models based on VGG16, Xception, and Densenet121 models were created for the classification of R/NR. Results show that the highest performance of hybrid TL-LSTM-Attention models was achieved by VGG16-LSTM-Attention with an accuracy of 98.21%, sensitivity of 96.22%, and specificity of 99.67%. An Ensemble model based on weighted majority voting among hybrid models is constructed to surpass the performance of each single hybrid model. Our proposed ensemble model gained accuracy, sensitivity, and specificity of 98.84%, 97.80%, and 99.60%, respectively. Hence, the proposed model which is an ensemble of TL-LSTM-Attention models fed by raw data images of EEG signal leads to high confidence in the prediction of antidepressants treatment outcome.

Automatic facial expression classification is very demanding research field because of its application in the field of health, safety and human machine interfaces. Many attempts by the researchers have been made in developing methodologies which can interpret, decode facial expression and obtain important features from the facial images to achieve better classification result. With the advancement in the data capturing techniques and various deep learning architectures it is possible to achieve higher accuracy in the computer vision task like facial expression classification. The aim of this research paper is to propose Custom-CNN architecture for the facial expression classification and performance of the model is compared with other standard pre-trained deep convolutional neural networks. Kaggle dataset comprises 35,900 is utilized to train, validate and test CNN models.

Breast cancer is a prevalent and life-threatening disease affecting many women worldwide. Early diagnosis plays a crucial role in improving survival rates and allowing effective treatment. Mammography is recognized as the most effective imaging modality for breast cancer diagnosis. To achieve early diagnosis, researchers have developed intelligent systems based on Deep Learning (DL) models, specifically Convolutional Neural Networks (CNNs), which are well-suited for analyzing medical images. Training deep CNN models for breast cancer diagnosis poses challenges, mainly due to the limited size of publicly available mammography datasets commonly used by researchers. This scarcity of data can lead to overfitting issues. Obtaining a large-scale breast mammogram dataset is a time-consuming and expensive process in clinical practice. Researchers have successfully employed Transfer Learning (TL) to overcome this issue. In this scenario, models are initially trained on large datasets from other domains, generating pre-trained models that capture general image representations. These models are then fine-tuned on a breast cancer image dataset. This approach leverages the learned knowledge from the first dataset to improve the performance of DL models in the breast cancer diagnosis task. This survey aims to present the latest research findings in the realm of TL architectures applied to breast cancer diagnosis using limited mammography datasets. It covers the following areas: (i) The structure of CNNs, (ii) Essential background knowledge on TL, (iii) Diverse strategies for implementing TL, (iv) Commonly employed pre-trained CNN architectures, (v) Well-known publicly available mammography datasets along with their key characteristics and strengths, and (vi) A summary of the current state-of-the-art pretrained CNNs specifically applied to breast cancer diagnosis. The survey focuses on the performance and key findings of these models, showcasing their effectiveness in enhancing diagnostic accuracy. This survey seeks to summarize the current trends in TL for breast cancer diagnosis using mammography images, intending to inspire researchers to actively contribute to the progress of transfer learning in the mammogram image analysis field.

International audience

Investigational procedures and medical diagnosis can be greatly improved by opting for detecting automatically abnormalities and anatomical landmarks in medical images. However, this remains a challenging task and still unexplored field. This paper aims to investigate the capabilities of a pretrained deep convolutional neural network VGG-16 model for images categorization with transfer learning containing anatomical landmarks, pathological finding and endoscopic procedures. Data augmentation is also performed to highlight the importance of data size for deep models. The accuracies achieved before and after data augmentation are 96.9% and 98.8% respectively.

Automatically generating summaries of user-generated videos is very useful but challenging. User-generated videos are unedited and usually only contain a long single shot which makes traditional video temporal segmentation methods such as shot boundary detection ineffective in producing meaningful video segments for summarization. To address this issue, we propose a novel temporal segmentation framework based on the clustering of joint visual and semantic affinity graph of the video frames. Based on a pre-trained deep convolutional neural network (CNN), we extract deep visual features of the frames to construct the visual affinity graph. We then construct the semantic affinity graph of the frames based on word embedding of the frames' semantic tags generated from an automatic image tagging algorithm. A dense neighbor method is then used to cluster the joint visual and semantic affinity graph to divide the video into sub-shot level segments and from which a summary of the video can be generated. Experimental results show that our approach outperforms state-of-the-art methods. Furthermore, we show that the method achieves results that are similar to those performed manually.