Deep Learning-based Classification Algorithms Research Articles

Experimental Study: Deep Learning-Based Fall Monitoring among Older Adults with Skin-Wearable Electronics.

Older adults are more vulnerable to falling due to normal changes due to aging, and their falls are a serious medical risk with high healthcare and societal costs. However, there is a lack of automatic fall detection systems for older adults. This paper reports (1) a wireless, flexible, skin-wearable electronic device for both accurate motion sensing and user comfort, and (2) a deep learning-based classification algorithm for reliable fall detection of older adults. The cost-effective skin-wearable motion monitoring device is designed and fabricated using thin copper films. It includes a six-axis motion sensor and is directly laminated on the skin without adhesives for the collection of accurate motion data. To study accurate fall detection using the proposed device, different deep learning models, body locations for the device placement, and input datasets are investigated using motion data based on various human activities. Our results indicate the optimal location to place the device is the chest, achieving accuracy of more than 98% for falls with motion data from older adults. Moreover, our results suggest a large motion dataset directly collected from older adults is essential to improve the accuracy of fall detection for the older adult population.

An intelligent method for pregnancy diagnosis in breeding sows according to ultrasonography algorithms.

Pig breeding management directly contributes to the profitability of pig farms, and pregnancy diagnosis is an important factor in breeding management. Therefore, the need to diagnose pregnancy in sows is emphasized, and various studies have been conducted in this area. We propose a computer-aided diagnosis system to assist livestock farmers to diagnose sow pregnancy through ultrasound. Methods for diagnosing pregnancy in sows through ultrasound include the Doppler method, which measures the heart rate and pulse status, and the echo method, which diagnoses by amplitude depth technique. We propose a method that uses deep learning algorithms on ultrasonography, which is part of the echo method. As deep learning-based classification algorithms, Inception-v4, Xception, and EfficientNetV2 were used and compared to find the optimal algorithm for pregnancy diagnosis in sows. Gaussian and speckle noises were added to the ultrasound images according to the characteristics of the ultrasonography, which is easily affected by noise from the surrounding environments. Both the original and noise added ultrasound images of sows were tested together to determine the suitability of the proposed method on farms. The pregnancy diagnosis performance on the original ultrasound images achieved 0.99 in accuracy in the highest case and on the ultrasound images with noises, the performance achieved 0.98 in accuracy. The diagnosis performance achieved 0.96 in accuracy even when the intensity of noise was strong, proving its robustness against noise.

KCV-B: BOOTSTRAP WITH CROSS-VALIDATION FOR DEEP LEARNING MODEL DEVELOPMENT, ASSESSMENT AND SELECTION

Abstract. This study investigates the inability of two popular data splitting techniques: train/test split and k-fold cross-validation that are to create training and validation data sets, and to achieve sufficient generality for supervised deep learning (DL) methods. This failure is mainly caused by their limited ability of new data creation. In response, the bootstrap is a computer based statistical resampling method that has been used efficiently for estimating the distribution of a sample estimator and to assess a model without having knowledge about the population. This paper couples cross-validation and bootstrap to have their respective advantages in view of data generation strategy and to achieve better generalization of a DL model. This paper contributes by: (i) developing an algorithm for better selection of training and validation data sets, (ii) exploring the potential of bootstrap for drawing statistical inference on the necessary performance metrics (e.g., mean square error), and (iii) introducing a method that can assess and improve the efficiency of a DL model. The proposed method is applied for semantic segmentation and is demonstrated via a DL based classification algorithm, PointNet, through aerial laser scanning point cloud data.

Detecting Out-of-Distribution Data in Wireless Communications Applications of Deep Learning

Deep learning-based classification algorithms offer no performance guarantees when deployed on testing data not generated by the same process as the training data. Such out-of-distribution (OOD) data often cause classification errors that are hard to detect since they do not generate explicit errors in the model. In real-world applications, there is no way to ensure that the testing data and the training data are drawn from the same or sufficiently similar distributions. This problem is especially challenging in wireless communications applications. Because the radio propagation channel is highly dynamic, it is very difficult to ensure that a deep learning model is not tested on OOD data. In this paper, we propose a novel deep learning model called FOOD (Feature representation for detecting OOD data) to detect OOD data in wireless communications applications. FOOD incorporates a new model architecture to detect OOD data accurately and minimizes the instances of normal data being recognized as OOD. We evaluated the performance of FOOD extensively using transmitter classification and modulation recognition tasks, with both experimental datasets and simulation-generated datasets. As far as we know, this is the first systematic study on the impact and detection of OOD data in deep learning-based wireless communications applications.

Optimization of an Intelligent Sorting and Recycling System for Solid Waste Based on Image Recognition Technology

In this paper, the technique of image recognition algorithm is used to conduct an in-depth study and analysis of the intelligent classification and recycling system of solid waste and to optimize the design of its system. The network structure and detection principle of the YOLO target detection algorithm based on convolutional neural nets are analysed, images of construction solid waste are collected as a dataset, and the image dataset is expanded using data enhancement techniques, and the target objects in the dataset are labelled and used to train their own YOLO detection models. To facilitate testing the images and to design a YOLO algorithm-based construction solid waste target detection system. Using the detection system for construction solid waste recognition, the YOLO model can accurately detect the location, class, and confidential information of the target object in the image. Image recognition is a technique to recognize images by capturing real-life images through devices and performing feature extraction, and this technique has been widely used since its inception. The deep learning-based classification algorithm for recyclable solid waste studied in this paper can classify solid waste efficiently and accurately, solving the problem that people do not know how to classify solid waste in daily life. The convolutional layer, pooling layer, and fully connected layer in a convolutional neural network are responsible for feature extraction, reducing the number of parameters, integrating features into high-level features, and finally classifying them by SoftMax classifier in turn. However, the actual situation is intricate and often the result is not obtained as envisioned, and the use of migration learning can be a good way to improve the overfitting phenomenon. In this paper, the combination of lazy optimizer and lookahead can improve the generalization ability and fitting speed as well as greatly improve the accuracy and stability. The experimental results are tested, and it is found that the solid waste classification accuracy can be as high as 95% when the VGG19 model is selected and the optimizer is combined.

Deep Learning-Based Three-Dimensional Oral Conical Beam Computed Tomography for Diagnosis.

In order to deeply study oral three-dimensional cone beam computed tomography (CBCT), the diagnosis of oral and facial surgical diseases based on deep learning was studied. The utility model related to a deep learning-based classification algorithm for oral neck and facial surgery diseases (deep diagnosis of oral and maxillofacial diseases, referred to as DDOM) is brought out; in this method, the DDOM algorithm proposed for patient classification, lesion segmentation, and tooth segmentation, respectively, can effectively process the three-dimensional oral CBCT data of patients and carry out patient-level classification. The segmentation results show that the proposed segmentation method can effectively segment the independent teeth in CBCT images, and the vertical magnification error of tooth CBCT images is clear. The average magnification rate was 7.4%. By correcting the equation of R value and CBCT image vertical magnification rate, the magnification error of tooth image length could be reduced from 7.4. According to the CBCT image length of teeth, the distance R from tooth center to FOV center, and the vertical magnification of CBCT image, the data closer to the real tooth size can be obtained, in which the magnification error is reduced to 1.0%. Therefore, it is proved that the 3D oral cone beam electronic computer based on deep learning can effectively assist doctors in three aspects: patient diagnosis, lesion localization, and surgical planning.

Identification of Human Motion Using Radar Sensor in an Indoor Environment.

In this paper, we propose a method of identifying human motions, such as standing, walking, running, and crawling, using a millimeter wave radar sensor. In our method, two signal processing is performed in parallel to identify the human motions. First, the moment at which a person’s motion changes is determined based on the statistical characteristics of the radar signal. Second, a deep learning-based classification algorithm is applied to determine what actions a person is taking. In each of the two signal processing, radar spectrograms containing the characteristics of the distance change over time are used as input. Finally, we evaluate the performance of the proposed method with radar sensor data acquired in an indoor environment. The proposed method can find the moment when the motion changes with an error rate of 3%, and also can classify the action that a person is taking with more than 95% accuracy.

Amyotrophic Lateral Sclerosis EEG Classification Using Deep Neural Network And TLBO

The prediction and detection of amyotrophic lateral sclerosis (ALS) diseases are a challenging task for the brain-computer interface. The discovery and classification of amyotrophic lateral sclerosis diseases used various neural network models. The limitation of the neural network deals with feature extraction and vector conversion of ALS data. The extraction of features of ALS diseases is very complex due to the recorded signal of the human brain. The recoded signal consists of the actual signal and other brain behaviours of a signal treat as noise. The noise recorded signal degraded the performance of the brain-computer interface. For the betterment of the brain-computer interface, I am using feature selection and reduction process with the neural network model. This paper proposed the deep learning-based classification algorithms for the prediction of ALS disease. The deep learning algorithm is a variant of a multilayer neural network for the extraction of features used discrete wavelet transform function. The discrete wavelet transform function decomposed the electroencephalogram signal in different sub-bands for the processing. The represent bands into varied frequency range. The proposed algorithms are simulated in MATLAB software and used the standard dataset of ALS diseases for the analysis of performance. The proposed algorithm compares with Bayesian-based neural network and ensemble-based machine learning classifiers for ALS disease detection. The performance of the proposed algorithms improved the efficiency of the brain-computer interface system.

High-Resolution Remote Sensing Imagery Classification of Imbalanced Data Using Multistage Sampling Method and Deep Neural Networks

Class imbalance is a key issue for the application of deep learning for remote sensing image classification because a model generated by imbalanced samples training has low classification accuracy for minority classes. In this study, an accurate classification approach using the multistage sampling method and deep neural networks was proposed to classify imbalanced data. We first balance samples by multistage sampling to obtain the training sets. Then, a state-of-the-art model is adopted by combining the advantages of atrous spatial pyramid pooling (ASPP) and Encoder-Decoder for pixel-wise classification, which are two different types of fully convolutional networks (FCNs) that can obtain contextual information of multiple levels in the Encoder stage. The details and spatial dimensions of targets are restored using such information during the Decoder stage. We employ four deep learning-based classification algorithms (basic FCN, FCN-8S, ASPP, and Encoder-Decoder with ASPP of our approach) on multistage training sets (original, MUS1, and MUS2) of WorldView-3 images in southeastern Qinghai-Tibet Plateau and GF-2 images in northeastern Beijing for comparison. The experiments show that, compared with existing sets (original, MUS1, and identical) and existing method (cost weighting), the MUS2 training set of multistage sampling significantly enhance the classification performance for minority classes. Our approach shows distinct advantages for imbalanced data.

Vulnerable community identification using hate speech detection on social media

With the rapid development in mobile computing and Web technologies, online hate speech has been increasingly spread in social network platforms since it's easy to post any opinions. Previous studies confirm that exposure to online hate speech has serious offline consequences to historically deprived communities. Thus, research on automated hate speech detection has attracted much attention. However, the role of social networks in identifying hate-related vulnerable community is not well investigated. Hate speech can affect all population groups, but some are more vulnerable to its impact than others. For example, for ethnic groups whose languages have few computational resources, it is a challenge to automatically collect and process online texts, not to mention automatic hate speech detection on social media. In this paper, we propose a hate speech detection approach to identify hatred against vulnerable minority groups on social media. Firstly, in Spark distributed processing framework, posts are automatically collected and pre-processed, and features are extracted using word n-grams and word embedding techniques such as Word2Vec. Secondly, deep learning algorithms for classification such as Gated Recurrent Unit (GRU), a variety of Recurrent Neural Networks (RNNs), are used for hate speech detection. Finally, hate words are clustered with methods such as Word2Vec to predict the potential target ethnic group for hatred. In our experiments, we use Amharic language in Ethiopia as an example. Since there was no publicly available dataset for Amharic texts, we crawled Facebook pages to prepare the corpus. Since data annotation could be biased by culture, we recruit annotators from different cultural backgrounds and achieved better inter-annotator agreement. In our experimental results, feature extraction using word embedding techniques such as Word2Vec performs better in both classical and deep learning-based classification algorithms for hate speech detection, among which GRU achieves the best result. Our proposed approach can successfully identify the Tigre ethnic group as the highly vulnerable community in terms of hatred compared with Amhara and Oromo. As a result, hatred vulnerable group identification is vital to protect them by applying automatic hate speech detection model to remove contents that aggravate psychological harm and physical conflicts. This can also encourage the way towards the development of policies, strategies, and tools to empower and protect vulnerable communities.

An adaptive deep Q-learning strategy for handwritten digit recognition

Handwritten digits recognition is a challenging problem in recent years. Although many deep learning-based classification algorithms are studied for handwritten digits recognition, the recognition accuracy and running time still need to be further improved. In this paper, an adaptive deep Q-learning strategy is proposed to improve accuracy and shorten running time for handwritten digit recognition. The adaptive deep Q-learning strategy combines the feature-extracting capability of deep learning and the decision-making of reinforcement learning to form an adaptive Q-learning deep belief network (Q-ADBN). First, Q-ADBN extracts the features of original images using an adaptive deep auto-encoder (ADAE), and the extracted features are considered as the current states of Q-learning algorithm. Second, Q-ADBN receives Q-function (reward signal) during recognition of the current states, and the final handwritten digits recognition is implemented by maximizing the Q-function using Q-learning algorithm. Finally, experimental results from the well-known MNIST dataset show that the proposed Q-ADBN has a superiority to other similar methods in terms of accuracy and running time.