Adaptive deep learning model selection on embedded systems

Image segmentation is the process of delineating the regions occupied by objects of interest in a given image. This operation is a fundamentally required first step in numerous applications of medical imagery. In the medical imaging field, this activity has a rich literature that spans over 45 years. In spite of numerous advances, including deep learning (DL) networks (DLNs) in recent years, the problem has defied a robust, fail-safe, and satisfactory solution, especially for objects that are manifest with low contrast, are spatially sparse, have variable shape among individuals, or are affected by imaging artifacts, pathology, or post-treatment change in the body. Although image processing techniques, notably DLNs, are uncanny in their ability to harness low-level intensity pattern information on objects, they fall short in the high-level task of identifying and localizing an entire object as a gestalt. This dilemma has been a fundamental unmet challenge in medical image segmentation. In this paper, we demonstrate that by synergistically marrying the unmatched strengths of high-level human knowledge (i.e., natural intelligence (NI)) with the capabilities of DL networks (i.e., artificial intelligence (AI)) in garnering intricate details, these challenges can be significantly overcome. Focusing on the object recognition task, we formulate an anatomy-guided DL object recognition approach named Automatic Anatomy Recognition-Deep Learning (AAR-DL) which combines an advanced anatomy-modeling strategy, model-based non-DL object recognition, and DL object detection networks to achieve expert human-like performance.

Recently, convolutional neural networks (CNNs), which exhibit excellent performance in the field of computer vision, have been in the spotlight. However, as the networks become wider for higher accuracy, the number of parameters and the computational costs increase exponentially. Therefore, it is challenging to use deep learning networks in embedded environments with limited resources, computational performance, and power. Moreover, CNNs consume a great deal of time for inference. To solve this problem, we propose a practical method for filter pruning to provide an optimal network architecture for target capacity and inference acceleration. After revealing the correlation between the inference time and the FLOPs, we proposed a method to generate a network with the desired inference time. Various object detection datasets were used to evaluate the performance of the proposed filter pruning method. The inference time of the pruned network was measured and analyzed using the NVIDIA Jetson Xavier NX platform. As a result of pruning the number of parameters and FLOPs of the YOLOv5 network in the PASCAL VOC dataset by 30%, 40%, and 50%, the mAP decreased by 0.6%, 2.3%, and 2.9%, respectively, while the inference time was improved by 14.3%, 26.4%, and 34.5%, respectively.

Deep learning (also known as deep structured learning, hierarchical learning, or deep machine learning) is a branch of machine learning based on a set of algorithms that attempt to model high-level abstractions in data by using a deep graph with multiple processing layers, composed of multiple linear and nonlinear transformations. Deep learning has been characterized as a class of machine learning algorithms with the following characteristics [257]: • They use a cascade of many layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. The algorithms may be supervised or unsupervised and applications include pattern analysis (unsupervised) and classification (supervised). • They are based on the (unsupervised) learning of multiple levels of features or representations of the data. Higher-level features are derived from lower-level features to form a hierarchical representation. • They are part of the broader machine learning field of learning representations of data. • They learn multiple levels of representations that correspond to different levels of abstraction; the levels form a hierarchy of concepts. These definitions have in common: multiple layers of nonlinear processing units and the supervised or unsupervised learning of feature representations in each layer, with the layers forming a hierarchy from low-level to high-level features. Various deep learning architectures such as deep neural networks, convolutional deep neural networks, deep belief networks (DBN), and recurrent neural networks have been applied to fields like computer vision, automatic speech recognition, natural language processing, audio recognition, and bioinformatics where they have been shown to produce state-of-the-art results on various tasks. In this chapter we start in Section 7.1 with an introduction, giving a brief history of this field, the relevant literature, and its applications. Then we study some basic concepts of deep learning such as convolutional neural networks, recurrent neural networks, backpropagation algorithm, restricted Boltzmann machines, and deep learning networks in Section 7.2. Then we illustrate three examples for Apache Spark implementation for mobile big data (MBD), user moving pattern extraction, and combination with nonparametric Bayesian learning, respectively, in Sections 7.3 through 7.5. Finally, we have summary in Section 7.6.

Detecting subtle signals from small earthquakes triggered by transient stresses from the surface waves of large magnitude earthquakes can contribute to a more general understanding of how earthquakes nucleate and interact with each other. However, searching for signals from such small earthquakes in thousands of seismograms is overwhelming, and discriminating them from a miscellany of noise is challenging. Here, we explore how we can automate the detection of such dynamically triggered earthquakes using a simple, diagnostic signal-to-noise ratio (SNR) threshold as well as a convolutional deep metric learning network. Our analysis shows that the deep learning network was more reliable at detecting small earthquakes than the SNR method.

To construct the risk prediction model of acute exacerbation of chronic obstructive pulmonary disease (AECOPD) and verify its effectiveness based on deep learning and back propagation algorithm neural network (BP neural network). Based on the relevant data of 1 326 patients with chronic obstructive pulmonary disease (COPD) in the team's previous clinical study, the acute exacerbation, and its risk factors during the stable period and 6 months of follow-up were recorded and analyzed. Combined with previous clinical research data and expert questionnaire results, the independent risk factors of AECOPD after screening and optimization by multivariate Logistic regression including gender, body mass index (BMI) classification, number of acute exacerbation, duration of acute exacerbation and forced expiratory volume in one second (FEV1) were used to build the BP neural network by Python 3.6 programming language and Tensorflow 1.12 deep learning framework. The patients were randomly selected according to the ratio of 4:1 to generate the training group and the test group, of which, the training group had 1 061 sample data while the test group had 265 pieces of sample data. The training group was used to establish the prediction model of neural network, and the test group was used for back-substitution test. When using the training group data to construct the neural network model, the training group was randomly divided into training set and verification set according to the ratio of 4:1. There were 849 training samples in the training set and 212 verification samples in the verification set. The optimal model was screened by adjusting the parameters of the neural network and combining the area under the receiver operator characteristic curve (AUC), and the sample data of the test group was substituted into the model for verification. The independent risk factors including gender, BMI classification, number of acute exacerbation, duration of acute exacerbation and FEV1 were collected from the team's previous clinical research, and the AECOPD risk prediction model was constructed based on deep learning and BP neural network. After 10 000 training sessions, the accuracy of the AECOPD risk prediction model in the validation set of the training group was 83.09%. When the number of training times reached 8 000, the accuracy basically tended to be stable and the prediction ability reached the upper limit. The AECOPD risk prediction model trained for 10 000 times was used to predict the risk of the validation set data, and the receiver operator characteristic curve (ROC curve) analysis showed that the AUC was 0.803. When using this model to predict the risk of the data of the test group, the accuracy rate was 81.69%. The risk prediction model based on deep learning and BP neural network has a medium level of prediction efficiency for acute exacerbation within 6 months in COPD patients, which can evaluate the risk of AECOPD and assist the clinic in making accurate treatment decisions.

Big data is playing a significant role in the current computing revolution. Industries and organizations are utilizing their insights for Business Intelligence by using Deep Learning Networks (DLN). However, dynamic characteristics of BD introduce many critical issues for DLN; Concept Drift (CD) is one of them. CD issue appears frequently in Online Supervised Learning environments in which data trends change over time. The problem may even worsen in a BD environment due to the veracity and variability factors. The CD issue may render the DLN inapplicable by degrading the accuracy of classification results in DLN which is a very serious issue that needs to be addressed. Therefore, these DLN need to quickly adapt to changes for maintaining the accuracy level of the results. To overcome classification accuracy, we need some dynamical changes in the existing DLN. Therefore, in this paper, we examine some of the existing Shallow Learning and Deep Learning models and their behavior before and after the Concept Drift (in experiment 1) and validate the pre-trained Deep Learning network (ResNet-50). In future work, this experiment will examine the most recent pre-trained DLN (Alex Net, VGG16, VGG19) and identify their suitability to overcome Concept Drift using fine-tuning and transfer learning approaches.

Technical note: Impact of tissue section thickness on accuracy of cell classification with a deep learning network.

BackgroundTo determine if mammographic features from deep learning networks can be applied in breast cancer to identify groups at interval invasive cancer risk due to masking beyond using traditional breast density measures.MethodsFull-field digital screening mammograms acquired in our clinics between 2006 and 2015 were reviewed. Transfer learning of a deep learning network with weights initialized from ImageNet was performed to classify mammograms that were followed by an invasive interval or screen-detected cancer within 12 months of the mammogram. Hyperparameter optimization was performed and the network was visualized through saliency maps. Prediction loss and accuracy were calculated using this deep learning network. Receiver operating characteristic (ROC) curves and area under the curve (AUC) values were generated with the outcome of interval cancer using the deep learning network and compared to predictions from conditional logistic regression with errors quantified through contingency tables.ResultsPre-cancer mammograms of 182 interval and 173 screen-detected cancers were split into training/test cases at an 80/20 ratio. Using Breast Imaging-Reporting and Data System (BI-RADS) density alone, the ability to correctly classify interval cancers was moderate (AUC = 0.65). The optimized deep learning model achieved an AUC of 0.82. Contingency table analysis showed the network was correctly classifying 75.2% of the mammograms and that incorrect classifications were slightly more common for the interval cancer mammograms. Saliency maps of each cancer case found that local information could highly drive classification of cases more than global image information.ConclusionsPre-cancerous mammograms contain imaging information beyond breast density that can be identified with deep learning networks to predict the probability of breast cancer detection.

Raman spectroscopy has attracted much attention due to its wide applications in drug detection and many other fields. However, Raman spectra often contain background noise, which poses significant challenges for subsequent analysis and processing. Although various methods for removing background noise have been proposed and have improved analysis accuracy to some extent, these mathematical model-based methods often rely on parameter adjustments based on spectral data to achieve the desired de-noising effect. The introduction of deep learning technology has provided new ideas to solve this problem, breaking through the limitations of traditional methods in parameter dependency. Different deep learning structures exhibit unique advantages when processing different types of data. In this study, we propose a static dropout triangular deep convolutional network (SD-TDCN). This deep learning network can maintain superior performance while significantly reducing the size of model parameters by statically discarding some convolutional blocks in the deep learning network. In addition, this deep learning network also lays an experimental foundation for the subsequent development of deep learning structures with adaptive dropout mechanisms.

Raman spectroscopy is an efficient method for detection of explosives even in small quantities. A laser can be combined with a Coded Aperture Snapshot Spectral Imaging (CASSI) system to collect Raman spectra from a surface at stand-off distances. The CASSI-system decrease the data collection time but instead increase the reconstruction time for the Raman image. Reconstruction of Raman spectra from an ensemble of compressed sensing measurements using standard reconstruction methods such as Total Variation (TV) is rather time consuming and limits the application domain for the technique. Novel machine learning approaches such as Deep Learning (DL) has lately been applied to reconstruction problems. We evaluate our earlier developed DL approach for reconstruction of Raman spectra from an ensemble of measurements formulated as a regression problem. The DL network is trained by minimizing a loss-function which is composed of two components: a reconstruction error and a re-projection error. The evaluated method is trained on simulated data where the training data has been generated using a transfer function. The transfer function has been developed to mimic the optical properties of a CASSI system. The DL network has been trained on different training sets with different levels of background noise, different number of materials in the scene and different spatial configurations of the materials in the scene. The reconstruction results using the DL network has been qualitatively evaluated on simulated data and the results are also compared to the Two-Step Iterative Shrinkage/Thresholding (TwIST) algorithm in terms of reconstruction quality and computation time. The reconstruction time for the DL are orders of magnitude lower than for TwIST without reducing the quality of the reconstructed Raman spectra.

Deep learning networks were employed to predict the maximum differential deflection of stiffened and waffle rafts due to reactive soil movements, Δmax. Four deep learning networks were used to predict Δmax, these are (1) stiffened rafts on shrinking soil, (2) stiffened rafts on swelling soil, (3) waffle rafts on shrinking soil, and (4) waffle rafts on swelling soil. The deep learning models were used to create design lines, which showed that both soil and structural features strongly influence the stiffened rafts. In contrast, waffle rafts showed a strong dependence on soil features in shrinking soils and beam depth in swelling soils. This demonstrates that the finite element-based deep learning networks captured the effect of the embedment of the beams. The results of the deep learning models led to non-linear design curves, which are disparate from the suggested standard Australian design. These results suggest that increasing the value of beam depth can have a positive or negative impact on the global residential slab depending on the type of substructure and whether the founding reactive soil is shrinking or swelling. Global sensitivity analyses of the deep learning models showed that for stiffened rafts on shrinking soil, the slab length, slab width and active depth zone of reactive soil had the most significant influence on Δmax, whilst for stiffened rafts on swelling soil, the primary drivers are ground movement, beam depth, and slab width. The prediction of Δmax for waffle rafts on shrinking soil was driven by the surface characteristic and mound movements, and the active depth zone, whilst waffle rafts on swelling soil was driven by the beam depth. Overall, the finite element-based deep learning showed the capacity to estimate Δmax in both shrinking and swelling design scenarios for different types of residential footing systems to further understand the characteristic behaviour of shallow residential slab foundations on reactive soils leading to improved designs.

Introduction. The significance of machine learning under the conditions of digital transformation of industry, and methods of implementing deep learning to provide the performance of trust management systems are considered. The necessity of using convolutional artificial neural networks for deep machine learning is determined. Various technologies and architectures for the implementation of artificial neural networks are briefly considered; a comparative analysis of their performance is carried out. The work objective is to study the need to develop new approaches to the architecture of computing machines for solving problems of deep machine learning in the trust management system implementation.Materials and Methods. In the context of digital transformation, the use of artificial intelligence reaches a new level. The technical implementation of artificial neural systems with deep machine learning is based on the use of one of three basic technologies: high performance computing (HPC) with parallel data processing, neuromorphic computing (NC), and quantum computing (QC).Results. Implementation models for deep machine learning, basic technologies and architecture of computing machines, as well as requirements for trust assurance in control systems using deep machine learning are analyzed. The problem of shortage of computation power for solving such problems is identified. None of the currently existing technologies can solve the full range of learning and impedance problems. The current level of technology does not provide information security and reliability of neural networks. The practical implementation of trust management systems with deep machine learning based on existing technologies for a significant part of the tasks does not provide a sufficient level of performance.Discussion and Conclusions. The study made it possible to identify the challenge of the computation power shortage for solving problems of deep machine learning. Through the analysis of the requirements for trust management systems, the external challenges of their implementation on the basis of existing technologies, and the need to develop new approaches to the computer architecture are determined.

Cardiac motion correction with a deep learning network for perfusion defect assessment in single-photon emission computed tomography myocardial perfusion imaging

Deep learning belongs to the field of artificial intelligence, where machines perform tasks that typically require some kind of human intelligence. Deep learning tries to achieve this by drawing inspiration from the learning of a human brain. Similar to the basic structure of a brain, which consists of (billions of) neurons and connections between them, a deep learning algorithm consists of an artificial neural network, which resembles the biological brain structure. Mimicking the learning process of humans with their senses, deep learning networks are fed with (sensory) data, like texts, images, videos or sounds. These networks outperform the state-of-the-art methods in different tasks and, because of this, the whole field saw an exponential growth during the last years. This growth resulted in way over 10,000 publications per year in the last years. For example, the search engine PubMed alone, which covers only a sub-set of all publications in the medical field, provides already over 11,000 results in Q3 2020 for the search term ‘deep learning’, and around 90% of these results are from the last three years. Consequently, a complete overview over the field of deep learning is already impossible to obtain and, in the near future, it will potentially become difficult to obtain an overview over a subfield. However, there are several review articles about deep learning, which are focused on specific scientific fields or applications, for example deep learning advances in computer vision or in specific tasks like object detection. With these surveys as a foundation, the aim of this contribution is to provide a first high-level, categorized meta-survey of selected reviews on deep learning across different scientific disciplines and outline the research impact that they already have during a short period of time. The categories (computer vision, language processing, medical informatics and additional works) have been chosen according to the underlying data sources (image, language, medical, mixed). In addition, we review the common architectures, methods, pros, cons, evaluations, challenges and future directions for every sub-category.

Deep learning neural network approach to thermal-wave imaging of damage in solids with application to diffusivity measurements of a green (unsintered) metal powder compact slab