China
Africa
Explore chapters and articles related to this topic
The Role of Artificial Intelligence for Intelligent Mobile Apps
Published in Anirbid Sircar, Gautami Tripathi, Namrata Bist, Kashish Ara Shakil, Mithileysh Sathiyanarayanan, Emerging Technologies for Sustainable and Smart Energy, 2022
Mohamed Yousuff, Anusha, Vijayashree, Jayashree
The anteroposterior and posteroanterior chest radiographs from CIED implanted patients over three years from 2016 to 2018 are obtained. The collected images are categorized based on device manufacturers like Abbott, Biotronik, Medtronic and Boston Scientific. Raw X-rays are resized to 400 × 400 pel red, green, blue (RGB) singe-channel files. Various data augmentation techniques, including random cropping, contrast variation, brightness adjustment, vertical and horizontal flipping, are performed. Mobile phone snapshots are also included to integrate artifact deviations. The images captured through mobile phones are subjected to diversified ambient lighting effects. Tensorflow (Martín et al., 2015) and Keras (Chollet, 2015) libraries of Python are utilized for the development of the model. K-fold cross-validation with all unique datapoints in a ratio of 7:2:1 is taken for training, validation and testing the model. Performance of image classification task is evaluated using was Tensorflow analysis modules. Sensitivity and specificity values are computed from confusion matrix parameters. Statistical significance is assessed using Pearson’s chi-squared test (X2) (Weinreich et al., 2019).
Deep Learning in Brain Segmentation
Published in Saravanan Krishnan, Ramesh Kesavan, B. Surendiran, G. S. Mahalakshmi, Handbook of Artificial Intelligence in Biomedical Engineering, 2021
While having access to more training data improves the performance of a machine learning algorithm, collecting new data may be expensive and sometimes infeasible for the case of medical images. Data augmentation is a powerful training technique that extends the existing dataset at hand. It is widely adopted in DNNs and machine learning algorithms in general due to its effectiveness in preventing overfitting and improving model performance. The technique creates small perturbation by performing geometric and color transformation to the input image. For instance, randomly cropping small portions of the image 10 times effectively creates 10 different images to the dataset. Other forms of data augmentation include image translation, random flipping, color jittering, and random affine transformation.
Deep Learning Approach to Predict and Grade Glaucoma from Fundus Images through Constitutional Neural Networks
Published in K. Gayathri Devi, Mamata Rath, Nguyen Thi Dieu Linh, Artificial Intelligence Trends for Data Analytics Using Machine Learning and Deep Learning Approaches, 2020
Kishore Balasubramanian, N. B. Ananthamoorthy
The developed CNN architecture is trained to classify images into normal and glaucoma and, further, glaucomatous cases in terms of severity using the local and public image sub-groups. From the acquired dataset, the training datasets are provided as inputs to the network. The images are then preprocessed, augmented, downsized, and fed to the CNN. One hundred images are selected for testing. The rest are selected for training the network. Many iterations are carried out randomly to generalize the performance. Stochastic gradient descent momentum training, also known as the steep descent, with a batch size of 30 samples is used to reduce the entropy loss. The momentum set is 0.9. The system is iterated with learning rates of 0.1, 0.01, 0.001, 0.0001, and 0.00001 with 300 epochs. The training rate must be optimal, as too low or high may cause long computational time or error respectively. Data augmentation is employed to prevent the problem of overfitting with the aim to artificially enlarge the training data set. Horizontal and vertical flipping is carried out randomly in training and image rotation is done via +30° to −30° [40].
Abnormal driving behavior detection based on an improved ant colony algorithm
Published in Applied Artificial Intelligence, 2023
Xiaodi Huang, Po Yun, Shuhui Wu, Zhongfeng Hu
Data augmentation is used in machine learning to increase dataset size and diversity. The technique involves creating new examples from existing data to improve the model’s generalization ability by exposing it to more diverse and representative examples. There are various data augmentation techniques depending on the data type and problem being addressed. In our proposed method for detecting abnormal driving behavior using pheromones and an improved ant colony algorithm, we utilized data augmentation techniques to increase dataset diversity and improve the generalization ability of our method. We generated additional driving data points by applying slight variations to existing driving data points, such as adjusting the vehicle’s speed, acceleration, or direction. This helped to create a more diverse set of examples for the method to learn from, which ultimately enhanced its performance on new and unseen datasets.
Automated segmentation of standard scanning planes to measure biometric parameters in foetal ultrasound images – a survey
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
U. B. Balagalla, J.V.D. Jayasooriya, C. de Alwis, A. Subasinghe
A larger training dataset increases the accuracy of DLA-based image segmentation (Sinclair et al. 2018). However, using a large clinical dataset in research is not feasible. To address the issue, data augmentation and cross-domain training are suggested. Data augmentation can be performed by using functions such as flipping, cropping, and rotation. However, generating more than ten images from one original image is not acceptable in medical image segmentation. Training FCN in cross-domain provides improved output. Nevertheless, the model fails to improve the performance if the gap between the two domains is significant (Chen et al. 2015). For example, when considering the medical images and natural images, the model complexities of natural images limit the application as it exceeds the required data limit (Wu et al. 2017).
Automatic Speech Recognition Using Limited Vocabulary: A Survey
Published in Applied Artificial Intelligence, 2022
Jean Louis K. E Fendji, Diane C. M. Tala, Blaise O. Yenke, Marcellin Atemkeng
DL can be implemented using various tools. However, Tensor Flow seems to be one of the best application methods currently available (Dhankar 2017). Figure 4 gives the steps of DL in ASR. Data augmentation helps to improve the performance of the model by generalizing better and thereby reducing overfitting (Salamon and Bello 2017). Data augmentation creates a rich, diverse set of data from a small amount of data. Data augmentation can be applied as a pre-processing step before training the model or later, directly in real-time. Different augmentation policies can be applied to audio data such as Time warping, Frequency masking, and Time masking. Recently, a new augmentation method called SpecAugment has been proposed by Park et al. in (Park 2019) for the ASR system. They combined the warping of the features and the masking of blocks of frequency channels, as well as the blocks of time steps. To ease the augmentation process, a recent free MATLAB Toolbox called Audiogmenter has been proposed (Maguolo et al. 2019).