Enhancing underwater target localization through proximity-driven recurrent neural networks

Environmental monitoring, ocean research, and underwater exploration are just a few of the marine applications that require precise underwater target localization. This study goes into the field of underwater target localization using Recurrent Neural Networks (RNNs) enhanced with proximity-based approaches, with a focus on mean estimation error as a performance metric. In complex and dynamic underwater environments, conventional localization systems frequently face challenges such as signal degradation, noise interference, and unstable hydrodynamic conditions. This paper presents a novel approach to employing RNNs to increase the accuracy of underwater target localization by exploiting the temporal dynamics of proximity-informed data. This method uses an RNN architecture to track changes in audio emissions from underwater targets sensed by a microphone network. Using the temporal correlations represented in the data, the RNN learns patterns indicative of target localization quickly and correctly. Furthermore, the addition of proximity-based features increases the model's ability to understand the relative distances between hydrophone nodes and the target, resulting in more accurate localization estimates. To evaluate the suggested methodology, thorough simulations and practical experiments were carried out in a variety of underwater environments. The results show that the RNN-based strategy beats conventional methods and works effectively even in difficult settings. The utility of the proximity-aware RNN model is demonstrated, in particular, by considerable reductions in the mean estimate error (MEE), an important performance measure.

Fully automated sinogram-based deep learning model for detection and classification of intracranial hemorrhage

PURPOSE

To propose an automated approach for detecting and classifying Intracranial Hemorrhages (ICH) directly from sinograms using a deep learning framework. This method is proposed to overcome the limitations of the conventional diagnosis by eliminating the time-consuming reconstruction step and minimizing the potential noise and artifacts that can occur during the Computed Tomography (CT) reconstruction process.

METHODS

This study proposes a two-stage automated approach for detecting and classifying ICH from sinograms using a deep learning framework. The first stage of the framework is Intensity Transformed Sinogram Sythesizer, which synthesizes sinograms that are equivalent to the intensity-transformed CT images. The second stage comprises of a cascaded Convolutional Neural Network-Recurrent Neural Network (CNN-RNN) model that detects and classifies hemorrhages from the synthesized sinograms. The CNN module extracts high-level features from each input sinogram, while the RNN module provides spatial correlation of the neighborhood regions in the sinograms. The proposed method was evaluated on a publicly available RSNA dataset consisting of a large sample size of 8652 patients.

RESULTS

The results showed that the proposed method had a notable improvement as high as 27% in patient-wise accuracies when compared to state-of-the-art methods like ResNext-101, Inception-v3 and Vision Transformer. Furthermore, the sinogram-based approach was found to be more robust to noise and offset errors in comparison to CT image-based approaches. The proposed model was also subjected to a multi-label classification analysis to determine the hemorrhage type from a given sinogram. The learning patterns of the proposed model were also examined for explainability using the activation maps.

CONCLUSION

The proposed sinogram-based approach can provide an accurate and efficient diagnosis of ICH without the need for the time-consuming reconstruction step and can potentially overcome the limitations of CT image-based approaches. The results show promising outcomes for the use of sinogram-based approaches in detecting hemorrhages, and further research can explore the potential of this approach in clinical settings.

A systematic review on deep learning-based automated cancer diagnosis models

Deep learning is gaining importance due to its wide range of applications. Many researchers have utilized deep learning (DL) models for the automated diagnosis of cancer patients. This paper provides a systematic review of DL models for automated diagnosis of cancer patients. Initially, various DL models for cancer diagnosis are presented. Five major categories of cancers such as breast, lung, liver, brain and cervical cancer are considered. As these categories of cancers have a very high percentage of occurrences with high mortality rate. The comparative analysis of different types of DL models is drawn for the diagnosis of cancer at early stages by considering the latest research articles from 2016 to 2022. After comprehensive comparative analysis, it is found that most of the researchers achieved appreciable accuracy with implementation of the convolutional neural network model. These utilized the pretrained models for automated diagnosis of cancer patients. Various shortcomings with the existing DL-based automated cancer diagnosis models are also been presented. Finally, future directions are discussed to facilitate further research for automated diagnosis of cancer patients.

A deep learning-based interactive medical image segmentation framework with sequential memory

BACKGROUND AND OBJECTIVE

Image segmentation is an essential component in medical image analysis. The case of 3D images such as MRI is particularly challenging and time consuming. Interactive or semi-automatic methods are thus highly desirable. However, existing methods do not exploit the typical sequentiality of real user interactions. This is due to the interaction memory used in these systems, which discards ordering. In contrast, we argue that the order of the user corrections should be used for training and lead to performance improvements.

METHODS

We contribute to solving this problem by proposing a general multi-class deep learning-based interactive framework for image segmentation, which embeds a base network in a user interaction loop with a user feedback memory. We propose to model the memory explicitly as a sequence of consecutive system states, from which the features can be learned, generally learning from the segmentation refinement process. Training is a major difficulty owing to the network's input being dependent on the previous output. We adapt the network to this loop by introducing a virtual user in the training process, modelled by dynamically simulating the iterative user feedback.

RESULTS

We evaluated our framework against existing methods on the complex task of multi-class semantic instance female pelvis MRI segmentation with 5 classes, including up to 27 tumour instances, using a segmentation dataset collected in our hospital, and on liver and pancreas CT segmentation, using public datasets. We conducted a user evaluation, involving both senior and junior medical personnel in matching and adjacent areas of expertise. We observed an annotation time reduction with 5'56" for our framework against 25' on average for classical tools. We systematically evaluated the influence of the number of clicks on the segmentation accuracy. A single interaction round our framework outperforms existing automatic systems with a comparable setup. We provide an ablation study and show that our framework outperforms existing interactive systems.

CONCLUSIONS

Our framework largely outperforms existing systems in accuracy, with the largest impact on the smallest, most difficult classes, and drastically reduces the average user segmentation time with fast inference at 47.2±6.2 ms per image.

RNN-BiLSTM-CRF based amalgamated deep learning model for electricity theft detection to secure smart grids

Electricity theft presents a substantial threat to distributed power networks, leading to non-technical losses (NTLs) that can significantly disrupt grid functionality. As power grids supply centralized electricity to connected consumers, any unauthorized consumption can harm the grids and jeopardize overall power supply quality. Detecting such fraudulent behavior becomes challenging when dealing with extensive data volumes. Smart grids provide a solution by enabling two-way electricity flow, thereby facilitating the detection, analysis, and implementation of new measures to address data flow issues. The key objective is to provide a deep learning-based amalgamated model to detect electricity theft and secure the smart grid. This research introduces an innovative approach to overcome the limitations of current electricity theft detection systems, which predominantly rely on analyzing one-dimensional (1-D) electric data. These approaches often exhibit insufficient accuracy when identifying instances of theft. To address this challenge, the article proposes an ensemble model known as the RNN-BiLSTM-CRF model. This model amalgamates the strengths of recurrent neural network (RNN) and bidirectional long short-term memory (BiLSTM) architectures. Notably, the proposed model harnesses both one-dimensional (1-D) and two-dimensional (2-D) electricity consumption data, thereby enhancing the effectiveness of the theft detection process. The experimental results showcase an impressive accuracy rate of 93.05% in detecting electricity theft, surpassing the performance of existing models in this domain.

Status and Prospects of Research on Deep Learning-based De Novo Generation of Drug Molecules

Traditional molecular de novo generation methods, such as evolutionary algorithms, generate new molecules mainly by linking existing atomic building blocks. The challenging issues in these methods include difficulty in synthesis, failure to achieve desired properties, and structural optimization requirements. Advances in deep learning offer new ideas for rational and robust de novo drug design. Deep learning, a branch of machine learning, is more efficient than traditional methods for processing problems, such as speech, image, and translation. This study provides a comprehensive overview of the current state of research in de novo drug design based on deep learning and identifies key areas for further development. Deep learning-based de novo drug design is pivotal in four key dimensions. Molecular databases form the basis for model training, while effective molecular representations impact model performance. Common DL models (GANs, RNNs, VAEs, CNNs, DMs) generate drug molecules with desired properties. The evaluation metrics guide research directions by determining the quality and applicability of generated molecules. This abstract highlights the foundational aspects of DL-based de novo drug design, offering a concise overview of its multifaceted contributions. Consequently, deep learning in de novo molecule generation has attracted more attention from academics and industry. As a result, many deep learning-based de novo molecule generation types have been actively proposed.

Optimizing RNNs for EMG Signal Classification: A Novel Strategy Using Grey Wolf Optimization

Accurate classification of electromyographic (EMG) signals is vital in biomedical applications. This study evaluates different architectures of recurrent neural networks for the classification of EMG signals associated with five movements of the right upper extremity. A Butterworth filter was implemented for signal preprocessing, followed by segmentation into 250 ms windows, with an overlap of 190 ms. The resulting dataset was divided into training, validation, and testing subsets. The Grey Wolf Optimization algorithm was applied to the gated recurrent unit (GRU), long short-term memory (LSTM) architectures, and bidirectional recurrent neural networks. In parallel, a performance comparison with support vector machines (SVMs) was performed. The results obtained in the first experimental phase revealed that all the RNN networks evaluated reached a 100% accuracy, standing above the 93% achieved by the SVM. Regarding classification speed, LSTM ranked as the fastest architecture, recording a time of 0.12 ms, followed by GRU with 0.134 ms. Bidirectional recurrent neural networks showed a response time of 0.2 ms, while SVM had the longest time at 2.7 ms. In the second experimental phase, a slight decrease in the accuracy of the RNN models was observed, standing at 98.46% for LSTM, 96.38% for GRU, and 97.63% for the bidirectional network. The findings of this study highlight the effectiveness and speed of recurrent neural networks in the EMG signal classification task.

An integrated feature selection approach to high water stress yield prediction

The timely and precise prediction of winter wheat yield plays a critical role in understanding food supply dynamics and ensuring global food security. In recent years, the application of unmanned aerial remote sensing has significantly advanced agricultural yield prediction research. This has led to the emergence of numerous vegetation indices that are sensitive to yield variations. However, not all of these vegetation indices are universally suitable for predicting yields across different environments and crop types. Consequently, the process of feature selection for vegetation index sets becomes essential to enhance the performance of yield prediction models. This study aims to develop an integrated feature selection method known as PCRF-RFE, with a focus on vegetation index feature selection. Initially, building upon prior research, we acquired multispectral images during the flowering and grain filling stages and identified 35 yield-sensitive multispectral indices. We then applied the Pearson correlation coefficient (PC) and random forest importance (RF) methods to select relevant features for the vegetation index set. Feature filtering thresholds were set at 0.53 and 1.9 for the respective methods. The union set of features selected by both methods was used for recursive feature elimination (RFE), ultimately yielding the optimal subset of features for constructing Cubist and Recurrent Neural Network (RNN) yield prediction models. The results of this study demonstrate that the Cubist model, constructed using the optimal subset of features obtained through the integrated feature selection method (PCRF-RFE), consistently outperformed the RNN model. It exhibited the highest accuracy during both the flowering and grain filling stages, surpassing models constructed using all features or subsets derived from a single feature selection method. This confirms the efficacy of the PCRF-RFE method and offers valuable insights and references for future research in the realms of feature selection and yield prediction studies.

Optimization of Deep Learning Parameters for Magneto-Impedance Sensor in Metal Detection and Classification

Deep learning technology is generally applied to analyze periodic data, such as the data of electromyography (EMG) and acoustic signals. Conversely, its accuracy is compromised when applied to the anomalous and irregular nature of the data obtained using a magneto-impedance (MI) sensor. Thus, we propose and analyze a deep learning model based on recurrent neural networks (RNNs) optimized for the MI sensor, such that it can detect and classify data that are relatively irregular and diverse compared to the EMG and acoustic signals. Our proposed method combines the long short-term memory (LSTM) and gated recurrent unit (GRU) models to detect and classify metal objects from signals acquired by an MI sensor. First, we configured various layers used in RNN with a basic model structure and tested the performance of each layer type. In addition, we succeeded in increasing the accuracy by processing the sequence length of the input data and performing additional work in the prediction process. An MI sensor acquires data in a non-contact mode; therefore, the proposed deep learning approach can be applied to drone control, electronic maps, geomagnetic measurement, autonomous driving, and foreign object detection.

Performance evaluation of metaheuristics-tuned recurrent neural networks for electroencephalography anomaly detection

Electroencephalography (EEG) serves as a diagnostic technique for measuring brain waves and brain activity. Despite its precision in capturing brain electrical activity, certain factors like environmental influences during the test can affect the objectivity and accuracy of EEG interpretations. Challenges associated with interpretation, even with advanced techniques to minimize artifact influences, can significantly impact the accurate interpretation of EEG findings. To address this issue, artificial intelligence (AI) has been utilized in this study to analyze anomalies in EEG signals for epilepsy detection. Recurrent neural networks (RNNs) are AI techniques specifically designed to handle sequential data, making them well-suited for precise time-series tasks. While AI methods, including RNNs and artificial neural networks (ANNs), hold great promise, their effectiveness heavily relies on the initial values assigned to hyperparameters, which are crucial for their performance for concrete assignment. To tune RNN performance, the selection of hyperparameters is approached as a typical optimization problem, and metaheuristic algorithms are employed to further enhance the process. The modified hybrid sine cosine algorithm has been developed and used to further improve hyperparameter optimization. To facilitate testing, publicly available real-world EEG data is utilized. A dataset is constructed using captured data from healthy and archived data from patients confirmed to be affected by epilepsy, as well as data captured during an active seizure. Two experiments have been conducted using generated dataset. In the first experiment, models were tasked with the detection of anomalous EEG activity. The second experiment required models to segment normal, anomalous activity as well as detect occurrences of seizures from EEG data. Considering the modest sample size (one second of data, 158 data points) used for classification models demonstrated decent outcomes. Obtained outcomes are compared with those generated by other cutting-edge metaheuristics and rigid statistical validation, as well as results' interpretation is performed.

DLSDHMS: Design of a deep learning-based analysis model for secure and distributed hospital management using context-aware sidechains

Designing an efficient hospital management solution requires the integration of multidomain operations that include secure storage, alert system modelling, infrastructure management, staff management, report analysis, and feedback-based learning tasks. Existing hospital management models are either highly complex or do not incorporate comprehensive deep learning analysis, which limits their deployment capabilities. Moreover, most of these models use mutable storage solutions, which restricts their trust levels under multi-patient to multi-doctor mapping scenarios. To overcome these issues, this article proposes the design of a novel deep Learning-based analysis model for secure and distributed hospital management via context-aware sidechains. The model initially collects large-scale information sets from different hospital entities via an IoT-based network and stores the information on context-sensitive sidechains. These context-sensitive sidechains store information sets related to Medicine Management, Doctor Management, Insurance and Billing Management, and Appointment Management operations. These chains are optimized via an Iterative Genetic Algorithm (IGA) that assists in improving storage and retrieval performance via intelligent merging and splitting operations. Information stored on these chains is processed via a combination of Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs), that assist in identifying patient-level diseases and issues. The information obtained from these classifiers is updated on the central repository and assists in the pre-emption of diseases for other patients. Due to these integrations, the proposed model is capable of reducing computational delay by 3.5 % and reducing storage cost by 8.3 % when compared to other blockchain-based deployments. The model is also able to pre-empt patient issues with 9.3 % higher accuracy and 4.8 % higher precision, which makes it useful for real-time clinical deployments.

Atmospheric Turbulence Degraded Video Restoration with Recurrent GAN (ATVR-GAN)

Atmospheric turbulence (AT) can change the path and direction of light during video capturing of a target in space due to the random motion of the turbulent medium, a phenomenon that is most noticeable when shooting videos at long ranges, resulting in severe video dynamic distortion and blur. To mitigate geometric distortion and reduce spatially and temporally varying blur, we propose a novel Atmospheric Turbulence Video Restoration Generative Adversarial Network (ATVR-GAN) with a specialized Recurrent Neural Network (RNN) generator, which is trained to predict the scene's turbulent optical flow (OF) field and utilizes a recurrent structure to catch both spatial and temporal dependencies. The new architecture is trained using a newly combined loss function that counts for the spatiotemporal distortions, specifically tailored to the AT problem. Our network was tested on synthetic and real imaging data and compared against leading algorithms in the field of AT mitigation and image restoration. The proposed method outperformed these methods for both synthetic and real data examined.

Accelerated NLRP3 inflammasome-inhibitory peptide design using a recurrent neural network model and molecular dynamics simulations

Anomalous NLRP3 inflammasome responses have been linked to multiple health issues, including but not limited to atherosclerosis, diabetes, metabolic syndrome, cardiovascular disease, and neurodegenerative disease. Thus, targeting NLRP3 and modulating its associated immune response might be a promising strategy for developing new anti-inflammatory drugs. Herein, we report a computational method for de novo peptide design for targeting NLRP3 inflammasomes. The described method leverages a long-short-term memory (LSTM) network based on a recurrent neural network (RNN) to model a valuable latent space of molecules. The resulting classifiers are utilized to guide the selection of molecules generated by the model based on circular dichroism spectra and physicochemical features derived from high-throughput molecular dynamics simulations. Of the experimentally tested sequences, 60% of the peptides showed NLRP3-mediated inhibition of IL-1β and IL-18. One peptide displayed high potency against NLRP3-mediated IL-1β inhibition. However, NLRC4 and AIM2 inflammasome-mediated IL-1β secretion was uninterrupted by this peptide, demonstrating its selectivity toward the NLRP3 inflammasome. Overall, these results indicate that deep learning and molecular dynamics can accelerate the discovery of NLRP3 inhibitors with potent and selective activity.

Toward an autism-friendly environment based on mobile apps user feedback analysis using deep learning and machine learning models

Autistic people are often disadvantaged in employment, education, etc. In fact, autistic students/employees face several challenges navigating and communicating with their superiors and colleagues. Mobile applications for people with Autism Spectrum Disorder (ASD apps for short) have been increasingly being adapted to help autistic people manage their conditions and daily activities. User feedback analysis is an effective method that can be used to improve ASD apps' services. In this article, we investigate the usage of ASD apps to improve the quality of life for autistic students/employees based on user feedback analysis. For this purpose, we analyze user reviews suggested on highly ranked ASD apps for college students, and workers. A total of 97,051 reviews have been collected from 13 ASD apps available on Google Play and Apple App stores. The collected reviews have been classified into negative, positive, and neutral opinions. This analysis has been performed using machine learning and deep learning models. The best performances were provided by combining RNN and LSTM models with an accuracy of 96.58% and an AUC of 99.41%. Finally, we provide some recommendations to improve ASD apps to assist developers in upgrading the main services provided by their apps.

Actual rating calculation of the zoom cloud meetings app using user reviews on google play store with sentiment annotation of BERT and hybridization of RNN and LSTM

The recent outbreaks of the COVID-19 forced people to work from home. All the educational institutes run their academic activities online. The online meeting app the "Zoom Cloud Meeting" provides the most entire supports for this purpose. For providing proper functionalities require in this situation of online supports the developers need the frequent release of new versions of the application. Which makes the chances to have lots of bugs during the release of new versions. To fix those bugs introduce developer needs users' feedback based on the new release of the application. But most of the time the ratings and reviews are created contraposition between them because of the users' inadvertent in giving ratings and reviews. And it has been the main problem to fix those bugs using user ratings for software developers. For this reason, we conduct this average rating calculation process based on the sentiment of user reviews to help software developers. We use BERT-based sentiment annotation to create unbiased datasets and hybridize RNN with LSTM to find calculated ratings based on the unbiased reviews dataset. Out of four models trained on four different datasets, we found promising performance in two datasets containing a necessarily large amount of unbiased reviews. The results show that the reviews have more positive sentiments than the actual ratings. Our results found an average of 3.60 stars rating, where the actual average rating found in dataset is 3.08 stars. We use reviews of more than 250 apps from the Google Play app store. The results of our can provide more promising if we can use a large dataset only containing the reviews of the Zoom Cloud Meeting app.

Graph MADDPG with RNN for multiagent cooperative environment

Multiagent systems face numerous challenges due to environmental uncertainty, with scalability being a critical issue. To address this, we propose a novel multi-agent cooperative model based on a graph attention network. Our approach considers the relationship between agents and continuous action spaces, utilizing graph convolution and recurrent neural networks to define these relationships. Graph convolution is used to define the relationship between agents, while recurrent neural networks define continuous action spaces. We optimize and model the multiagent system by encoding the interaction weights among agents using the graph neural network and the weights between continuous action spaces using the recurrent neural network. We evaluate the performance of our proposed model by conducting experimental simulations using a 3D wargame engine that involves several unmanned air vehicles (UAVs) acting as attackers and radar stations acting as defenders, where both sides have the ability to detect each other. The results demonstrate that our proposed model outperforms the current state-of-the-art methods in terms of scalability, robustness, and learning efficiency.

Combined CNN and RNN Neural Networks for GPR Detection of Railway Subgrade Diseases

Vehicle-mounted ground-penetrating radar (GPR) has been used to non-destructively inspect and evaluate railway subgrade conditions. However, existing GPR data processing and interpretation methods mostly rely on time-consuming manual interpretation, and limited studies have applied machine learning methods. GPR data are complex, high-dimensional, and redundant, in particular with non-negligible noises, for which traditional machine learning methods are not effective when applied to GPR data processing and interpretation. To solve this problem, deep learning is more suitable to process large amounts of training data, as well as to perform better data interpretation. In this study, we proposed a novel deep learning method to process GPR data, the CRNN network, which combines convolutional neural networks (CNN) and recurrent neural networks (RNN). The CNN processes raw GPR waveform data from signal channels, and the RNN processes features from multiple channels. The results show that the CRNN network achieves a higher precision at 83.4%, with a recall of 77.3%. Compared to the traditional machine learning method, the CRNN is 5.2 times faster and has a smaller size of 2.6 MB (traditional machine learning method: 104.0 MB). Our research output has demonstrated that the developed deep learning method improves the efficiency and accuracy of railway subgrade condition evaluation.

Amplified cortical neural responses as animals learn to use novel activity patterns

Cerebral cortex supports representations of the world in patterns of neural activity, used by the brain to make decisions and guide behavior. Past work has found diverse, or limited, changes in the primary sensory cortex in response to learning, suggesting that the key computations might occur in downstream regions. Alternatively, sensory cortical changes may be central to learning. We studied cortical learning by using controlled inputs we insert: we trained mice to recognize entirely novel, non-sensory patterns of cortical activity in the primary visual cortex (V1) created by optogenetic stimulation. As animals learned to use these novel patterns, we found that their detection abilities improved by an order of magnitude or more. The behavioral change was accompanied by large increases in V1 neural responses to fixed optogenetic input. Neural response amplification to novel optogenetic inputs had little effect on existing visual sensory responses. A recurrent cortical model shows that this amplification can be achieved by a small mean shift in recurrent network synaptic strength. Amplification would seem to be desirable to improve decision-making in a detection task; therefore, these results suggest that adult recurrent cortical plasticity plays a significant role in improving behavioral performance during learning.

Multivariate Time Series Information Bottleneck

Time series (TS) and multiple time series (MTS) predictions have historically paved the way for distinct families of deep learning models. The temporal dimension, distinguished by its evolutionary sequential aspect, is usually modeled by decomposition into the trio of "trend, seasonality, noise", by attempts to copy the functioning of human synapses, and more recently, by transformer models with self-attention on the temporal dimension. These models may find applications in finance and e-commerce, where any increase in performance of less than 1% has large monetary repercussions, they also have potential applications in natural language processing (NLP), medicine, and physics. To the best of our knowledge, the information bottleneck (IB) framework has not received significant attention in the context of TS or MTS analyses. One can demonstrate that a compression of the temporal dimension is key in the context of MTS. We propose a new approach with partial convolution, where a time sequence is encoded into a two-dimensional representation resembling images. Accordingly, we use the recent advances made in image extension to predict an unseen part of an image from a given one. We show that our model compares well with traditional TS models, has information-theoretical foundations, and can be easily extended to more dimensions than only time and space. An evaluation of our multiple time series-information bottleneck (MTS-IB) model proves its efficiency in electricity production, road traffic, and astronomical data representing solar activity, as recorded by NASA's interface region imaging spectrograph (IRIS) satellite.

In-City Rain Mapping from Commercial Microwave Links-Challenges and Opportunities

Obtaining accurate rainfall measurements is highly important in urban areas, having a significant impact on different aspects in city life. Opportunistic rainfall sensing utilizing measurements collected by existing microwave and mmWave-based wireless networks has been researched in the last two decades and can be considered as an opportunistic integrated sensing and communication (ISAC) approach. In this paper, we compare two methods that utilize received signal level (RSL) measurements obtained by an existing smart-city wireless network deployed in the city of Rehovot, Israel, for rain estimation. The first method is a model-based approach using the RSL measurements from short links, in which two design parameters are calibrated empirically. This method is combined with a known wet/dry classification method, which is based on the rolling standard deviation of the RSL. The second method is a data-driven approach, based on a recurrent neural network (RNN), which is trained to estimate rainfall and classify wet/dry periods. We compare the results of rainfall classification and estimation from both methods and show that the data-driven approach slightly outperforms the empirical model and that the improvement is most significant for light rainfall events. Furthermore, we apply both methods to construct high-resolution 2D maps of accumulated rainfall in the city of Rehovot. The ground-level rainfall maps constructed over the city area are compared for the first time with weather radar rainfall maps obtained from the Israeli Meteorological Service (IMS). The rain maps generated by the smart-city network are found to be in agreement with the average rainfall depth obtained from the radar, demonstrating the potential of using existing smart-city networks as a source for constructing 2D high-resolution rainfall maps.

An Ensembled Framework for Human Breast Cancer Survivability Prediction Using Deep Learning

Breast cancer is categorized as an aggressive disease, and it is one of the leading causes of death. Accurate survival predictions for both long-term and short-term survivors, when delivered on time, can help physicians make effective treatment decisions for their patients. Therefore, there is a dire need to design an efficient and rapid computational model for breast cancer prognosis. In this study, we propose an ensemble model for breast cancer survivability prediction (EBCSP) that utilizes multi-modal data and stacks the output of multiple neural networks. Specifically, we design a convolutional neural network (CNN) for clinical modalities, a deep neural network (DNN) for copy number variations (CNV), and a long short-term memory (LSTM) architecture for gene expression modalities to effectively handle multi-dimensional data. The independent models' results are then used for binary classification (long term > 5 years and short term < 5 years) based on survivability using the random forest method. The EBCSP model's successful application outperforms models that utilize a single data modality for prediction and existing benchmarks.

Evaluation of Electrical Impedance Spectra by Long Short-Term Memory to Estimate Nitrate Concentrations in Soil

Monitoring the nitrate concentration in soil is crucial to guide the use of nitrate-based fertilizers. This study presents the characteristics of an impedance sensor used to estimate the nitrate concentration in soil based on the sensitivity of the soil dielectric constant to ion conductivity and on electrical double layer effects at electrodes. The impedance of synthetic sandy soil samples with nitrate nitrogen concentrations ranging from 0 to 15 mg/L was measured at frequencies between 20 Hz and 5 kHz and noticeable conductance and susceptance effects were observed. Long short-term memory (LSTM), a variant of recurrent artificial neural networks (RNN), was investigated with respect to its suitability to extract nitrate concentrations from the measured impedance spectra and additional physical properties of the soils, such as mass density and water content. Both random forest and LSTM were tested as feature selection methods. Then, numerous LSTMs were trained to estimate the nitrate concentrations in the soils. To increase estimation accuracy, hyperparameters were optimized with Bayesian optimization. The resulting optimal regression model showed coefficients of determination between true and predicted nitrate concentrations as high as 0.95. Thus, it could be demonstrated that the system has the potential to monitor nitrate concentrations in soils in real time and in situ.

The neural machine translation models for the low-resource Kazakh-English language pair

The development of the machine translation field was driven by people's need to communicate with each other globally by automatically translating words, sentences, and texts from one language into another. The neural machine translation approach has become one of the most significant in recent years. This approach requires large parallel corpora not available for low-resource languages, such as the Kazakh language, which makes it difficult to achieve the high performance of the neural machine translation models. This article explores the existing methods for dealing with low-resource languages by artificially increasing the size of the corpora and improving the performance of the Kazakh-English machine translation models. These methods are called forward translation, backward translation, and transfer learning. Then the Sequence-to-Sequence (recurrent neural network and bidirectional recurrent neural network) and Transformer neural machine translation architectures with their features and specifications are concerned for conducting experiments in training models on parallel corpora. The experimental part focuses on building translation models for the high-quality translation of formal social, political, and scientific texts with the synthetic parallel sentences from existing monolingual data in the Kazakh language using the forward translation approach and combining them with the parallel corpora parsed from the official government websites. The total corpora of 380,000 parallel Kazakh-English sentences are trained on the recurrent neural network, bidirectional recurrent neural network, and Transformer models of the OpenNMT framework. The quality of the trained model is evaluated with the BLEU, WER, and TER metrics. Moreover, the sample translations were also analyzed. The RNN and BRNN models showed a more precise translation than the Transformer model. The Byte-Pair Encoding tokenization technique showed better metrics scores and translation than the word tokenization technique. The Bidirectional recurrent neural network with the Byte-Pair Encoding technique showed the best performance with 0.49 BLEU, 0.51 WER, and 0.45 TER.

Crowd-Aware Mobile Robot Navigation Based on Improved Decentralized Structured RNN via Deep Reinforcement Learning

Efficient navigation in a socially compliant manner is an important and challenging task for robots working in dynamic dense crowd environments. With the development of artificial intelligence, deep reinforcement learning techniques have been widely used in the robot navigation. Previous model-free reinforcement learning methods only considered the interactions between robot and humans, not the interactions between humans and humans. To improve this, we propose a decentralized structured RNN network with coarse-grained local maps (LM-SRNN). It is capable of modeling not only Robot-Human interactions through spatio-temporal graphs, but also Human-Human interactions through coarse-grained local maps. Our model captures current crowd interactions and also records past interactions, which enables robots to plan safer paths. Experimental results show that our model is able to navigate efficiently in dense crowd environments, outperforming state-of-the-art methods.

Human induced pluripotent stem cell formation and morphology prediction during reprogramming with time-lapse bright-field microscopy images using deep learning methods

BACKGROUND AND OBJECTIVE

Human induced pluripotent stem cells (hiPSCs) represent an ideal source for patient specific cell-based regenerative medicine; however, efficiency of hiPSC formation from reprogramming cells is low. We use several deep-learning results from time-lapse brightfield microscopy images during culture, to early detect the cells potentially reprogramming into hiPSCs and predict the colony morphology of these cells for improving efficiency of culturing a new hiPSC line.

METHODS

Sets of time-lapse bright-field images are taken to track reprogramming process of CD34+ cells biologically identified as just beginning reprogramming. Prior the experiment, 9 classes of templates with distinct cell features clipped from microscopy images at various reprogramming stages are used to train a CNN model. The CNN is then used to classify a microscopy image as probability images of these classes. Probability images of some class are used to train a densely connected convolutional network for extracting regions of this class on a microscopy image. A U-net is trained to segment cells on the time-lapse images in early reprogramming stage during culture. The segmented cells are classified by the extracted regions to count various types of cells appearing in the early reprogramming stage for predicting the identified cells potentially forming hiPSCs. The probability images of hiPSC classes are also used to train a spatiotemporal RNN for predicting the future hiPSC colony morphology of the potential cells.

RESULTS

Experimental results show the prediction (before 7 days after of beginning of the reprogramming) achieved 0.8 accuracy, and 66% of the identified cells under different culture conditions, predicted as forming, finally formed hiPSCs. The predicted hiPSC images and extracted colonies on the images show the prediction for future 1.5 days achieved high accuracy of hiPSC colony areas and image similarity.

CONCLUSIONS

Our study proposes a method using several deep learning models to efficiently select the reprogramming cells possibly forming hiPSCs and to predict the shapes of growing hiPSC colonies. The proposed method is expected to improve the efficiency when establishing a new hiPSC line culture.

Molecular Property Prediction of Modified Gedunin Using Machine Learning

Images of molecules are often utilized in education and synthetic exploration to predict molecular characteristics. Deep learning (DL) has also had an influence on drug research, such as the interpretation of cellular images as well as the development of innovative methods for the synthesis of organic molecules. Although research in these areas has been significant, a comprehensive review of DL applications in drug development would be beyond the scope of a single Account. In this study, we will concentrate on a single major area where DL has influenced molecular design: the prediction of molecular properties of modified gedunin using machine learning (ML). AI and ML technologies are critical in drug research and development. In these other words, deep learning (DL) algorithms and artificial neural networks (ANN) have changed the field. In short, advances in AI and ML present a good potential for rational drug design and exploration, which will ultimately benefit humanity. In this paper, long short-term memory (LSTM) was used to convert a modified gedunin SMILE into a molecular image. The 2D molecular representations and their immediately visible highlights should then provide adequate data to predict the subordinate characteristics of atom design. We aim to find the properties of modified gedunin using K-means clustering; RNN-like ML tools. To support this postulation, neural network (NN) clustering based on the AI picture is used and evaluated in this study. The novel chemical developed via profound learning has long been predicted on characteristics. As a result, LSTM with RNNs allow us to predict the properties of molecules of modified gedunin. The total accuracy of the suggested model is 98.68%. The accuracy of the molecular property prediction of modified gedunin research is promising enough to evaluate extrapolation and generalization. The model suggested in this research requires just seconds or minutes to calculate, making it faster as well as more effective than existing techniques. In short, ML can be a useful tool for predicting the properties of modified gedunin molecules.

Cardiac Magnetic Resonance Image Segmentation Method Based on Multi-Scale Feature Fusion and Sequence Relationship Learning

Accurate segmentation of the left atrial structure using magnetic resonance images provides an important basis for the diagnosis of atrial fibrillation (AF) and its treatment using robotic surgery. In this study, an image segmentation method based on sequence relationship learning and multi-scale feature fusion is proposed for 3D to 2D sequence conversion in cardiac magnetic resonance images and the varying scales of left atrial structures within different slices. Firstly, a convolutional neural network layer with an attention module was designed to extract and fuse contextual information at different scales in the image, to strengthen the target features using the correlation between features in different regions within the image, and to improve the network's ability to distinguish the left atrial structure. Secondly, a recurrent neural network layer oriented to two-dimensional images was designed to capture the correlation of left atrial structures in adjacent slices by simulating the continuous relationship between sequential image slices. Finally, a combined loss function was constructed to reduce the effect of positive and negative sample imbalance and improve model stability. The Dice, IoU, and Hausdorff distance values reached 90.73%, 89.37%, and 4.803 mm, respectively, based on the LASC2013 (left atrial segmentation challenge in 2013) dataset; the corresponding values reached 92.05%, 89.41% and 9.056 mm, respectively, based on the ASC2018 (atrial segmentation challenge at 2018) dataset.

DeepZ: A Deep Learning Approach for Z-DNA Prediction

Here we describe an approach that uses deep learning neural networks such as CNN and RNN to aggregate information from DNA sequence; physical, chemical, and structural properties of nucleotides; and omics data on histone modifications, methylation, chromatin accessibility, and transcription factor binding sites and data from other available NGS experiments. We explain how with the trained model one can perform whole-genome annotation of Z-DNA regions and feature importance analysis in order to define key determinants for functional Z-DNA regions.

Effective treatment of imbalanced datasets in health care using modified SMOTE coupled with stacked deep learning algorithms

One of the prominent uses of Predictive Analytics is Health care for more accurate predictions based on proper analysis of cumulative datasets. Often times the datasets are quite imbalanced and sampling techniques like Synthetic Minority Oversampling Technique (SMOTE) give only moderate accuracy in such cases. To overcome this problem, a two-step approach has been proposed. In the first step, SMOTE is modified to reduce the class imbalance in terms of Distance-based SMOTE (D-SMOTE) and Bi-phasic SMOTE (BP-SMOTE) which were then coupled with selective classifiers for prediction. An increase in accuracy is noted for both BP-SMOTE and D-SMOTE compared to basic SMOTE. In the second step, Machine learning, Deep Learning and Ensemble algorithms were used to develop a Stacking Ensemble Framework which showed a significant increase in accuracy for Stacking compared to individual machine learning algorithms like Decision Tree, Naïve Bayes, Neural Networks and Ensemble techniques like Voting, Bagging and Boosting. Two different methods have been developed by combing Deep learning with Stacking approach namely Stacked CNN and Stacked RNN which yielded significantly higher accuracy of 96-97% compared to individual algorithms. Framingham dataset is used for data sampling, Wisconsin Hospital data of Breast Cancer study is used for Stacked CNN and Novel Coronavirus 2019 dataset relating to forecasting COVID-19 cases, is used for Stacked RNN.

Cough Audio Analysis for COVID-19 Diagnosis

Humanity has suffered catastrophically due to the COVID-19 pandemic. One of the most reliable diagnoses of COVID-19 is RT-PCR (reverse-transcription polymer chain reaction) testing. This method, however, has its limitations. It is time consuming and requires scalability. This research work carries out a preliminary prognosis of COVID-19, which is scalable and less time consuming. The research carried out a competitive analysis of four machine-learning models namely, Multilayer Perceptron, Convolutional Neural Networks, Recurrent Neural Networks with Long Short-Term Memory, and VGG-19 with Support Vector Machines. Out of these models, Multilayer Perceptron outperformed with higher specificity of 94.5% and accuracy of 96.8%. The results show that Multilayer Perceptron was able to distinguish between positive and negative COVID-19 coughs by a robust feature embedding technique.

Identifying molecular functional groups of organic compounds by deep learning of NMR data

We preprocess the raw nuclear magnetic resonance (NMR) spectrum and extract key features by using two different methodologies, called equidistant sampling and peak sampling for subsequent substructure pattern recognition. We also provide a strategy to address the imbalance issue frequently encountered in statistical modeling of NMR data set and establish two conventional support vector machine (SVM) and K-nearest neighbor (KNN) models to assess the capability of two feature selections, respectively. Our results in this study show that the models using the selected features of peak sampling outperform those using equidistant sampling. Then we build the recurrent neural network (RNN) model trained by data collected from peak sampling. Furthermore, we illustrate the easier optimization of hyperparameters and the better generalization ability of the RNN deep learning model by detailed comparison with traditional machine learning SVM and KNN models.

SARS-CoV-2 Morphometry Analysis and Prediction of Real Virus Levels Based on Full Recurrent Neural Network Using TEM Images

The SARS-CoV-2 virus is responsible for the rapid global spread of the COVID-19 disease. As a result, it is critical to understand and collect primary data on the virus, infection epidemiology, and treatment. Despite the speed with which the virus was detected, studies of its cell biology and architecture at the ultrastructural level are still in their infancy. Therefore, we investigated and analyzed the viral morphometry of SARS-CoV-2 to extract important key points of the virus's characteristics. Then, we proposed a prediction model to identify the real virus levels based on the optimization of a full recurrent neural network (RNN) using transmission electron microscopy (TEM) images. Consequently, identification of virus levels depends on the size of the morphometry of the area (width, height, circularity, roundness, aspect ratio, and solidity). The results of our model were an error score of training network performance 3.216 × 10^-11 at 639 epoch, regression of -1.6 × 10^-9, momentum gain (Mu) 1 × 10^-9, and gradient value of 9.6852 × 10^-8, which represent a network with a high ability to predict virus levels. The fully automated system enables virologists to take a high-accuracy approach to virus diagnosis, prevention of mutations, and life cycle and improvement of diagnostic reagents and drugs, adding a point of view to the advancement of medical virology.

An End-to-End Recurrent Neural Network for Radial MR Image Reconstruction

Recent advances in deep learning have contributed greatly to the field of parallel MR imaging, where a reduced amount of k-space data are acquired to accelerate imaging time. In our previous work, we have proposed a deep learning method to reconstruct MR images directly from k-space data acquired with Cartesian trajectories. However, MRI utilizes various non-Cartesian trajectories, such as radial trajectories, with various numbers of multi-channel RF coils according to the purpose of an MRI scan. Thus, it is important for a reconstruction network to efficiently unfold aliasing artifacts due to undersampling and to combine multi-channel k-space data into single-channel data. In this work, a neural network named 'ETER-net' is utilized to reconstruct an MR image directly from k-space data acquired with Cartesian and non-Cartesian trajectories and multi-channel RF coils. In the proposed image reconstruction network, the domain transform network converts k-space data into a rough image, which is then refined in the following network to reconstruct a final image. We also analyze loss functions including adversarial and perceptual losses to improve the network performance. For experiments, we acquired k-space data at a 3T MRI scanner with Cartesian and radial trajectories to show the learning mechanism of the direct mapping relationship between the k-space and the corresponding image by the proposed network and to demonstrate the practical applications. According to our experiments, the proposed method showed satisfactory performance in reconstructing images from undersampled single- or multi-channel k-space data with reduced image artifacts. In conclusion, the proposed method is a deep-learning-based MR reconstruction network, which can be used as a unified solution for parallel MRI, where k-space data are acquired with various scanning trajectories.

Predicting respiratory motion using a novel patient specific dual deep recurrent neural networks

Real-time tracking of a target volume is a promising solution for reducing the planning margins and both dosimetric and geometric uncertainties in the treatment of thoracic and upper-abdomen cancers. Respiratory motion prediction is an integral part of real-time tracking to compensate for the latency of tracking systems. The purpose of this work was to develop a novel method for accurate respiratory motion prediction using dual deep recurrent neural networks (RNNs). The respiratory motion data of 111 patients were used to train and evaluate the method. For each patient, two models (Network1 and Network2) were trained on 80% of the respiratory wave, and the remaining 20% was used for evaluation. The first network (Network 1) is a "coarse resolution" prediction of future points and second network (Network 2) provides a "fine resolution" prediction to interpolate between the future predictions. The performance of the method was tested using two types of RNN algorithms : Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). The accuracy of each model was evaluated using the root mean square error (RMSE) and mean absolute error (MAE). Overall, the RNN model with GRU- function had better accuracy than the RNN model with LSTM-function (RMSE (mm): 0.4±0.2 vs. 0.6±0.3; MAE (mm): 0.4±0.2 vs. 0.6±0.2). The GRU was able to predict the respiratory motion accurately (<1 mm) up to the latency period of 440 ms, and LSTM's accuracy was acceptable only up to 240 ms. The proposed method using GRU function can be used for respiratory-motion prediction up to a latency period of 440 ms.

A Deep Sequence Learning Framework for Action Recognition in Small-Scale Depth Video Dataset

Depth video sequence-based deep models for recognizing human actions are scarce compared to RGB and skeleton video sequences-based models. This scarcity limits the research advancements based on depth data, as training deep models with small-scale data is challenging. In this work, we propose a sequence classification deep model using depth video data for scenarios when the video data are limited. Unlike summarizing the frame contents of each frame into a single class, our method can directly classify a depth video, i.e., a sequence of depth frames. Firstly, the proposed system transforms an input depth video into three sequences of multi-view temporal motion frames. Together with the three temporal motion sequences, the input depth frame sequence offers a four-stream representation of the input depth action video. Next, the DenseNet121 architecture is employed along with ImageNet pre-trained weights to extract the discriminating frame-level action features of depth and temporal motion frames. The extracted four sets of feature vectors about frames of four streams are fed into four bi-directional (BLSTM) networks. The temporal features are further analyzed through multi-head self-attention (MHSA) to capture multi-view sequence correlations. Finally, the concatenated genre of their outputs is processed through dense layers to classify the input depth video. The experimental results on two small-scale benchmark depth datasets, MSRAction3D and DHA, demonstrate that the proposed framework is efficacious even for insufficient training samples and superior to the existing depth data-based action recognition methods.

HVIOnet: A deep learning based hybrid visual-inertial odometry approach for unmanned aerial system position estimation

Sensor fusion is used to solve the localization problem in autonomous mobile robotics applications by integrating complementary data acquired from various sensors. In this study, we adopt Visual-Inertial Odometry (VIO), a low-cost sensor fusion method that integrates inertial data with images using a Deep Learning (DL) framework to predict the position of an Unmanned Aerial System (UAS). The developed system has three steps. The first step extracts features from images acquired from a platform camera and uses a Convolutional Neural Network (CNN) to project them to a visual feature manifold. Next, temporal features are extracted from the Inertial Measurement Unit (IMU) data on the platform using a Bidirectional Long Short Term Memory (BiLSTM) network and are projected to an inertial feature manifold. The final step estimates the UAS position by fusing the visual and inertial feature manifolds via a BiLSTM-based architecture. The proposed approach is tested with the public EuRoC (European Robotics Challenge) dataset and simulation environment data generated within the Robot Operating System (ROS). The result of the EuRoC dataset shows that the proposed approach achieves successful position estimations comparable to previous popular VIO methods. In addition, as a result of the experiment with the simulation dataset, the UAS position is successfully estimated with 0.167 Mean Square Error (RMSE). The obtained results prove that the proposed deep architecture is useful for UAS position estimation.

Novel Internet of Things based approach toward diabetes prediction using deep learning models

The integration of the Internet of Things with machine learning in different disciplines has benefited from recent technological advancements. In medical IoT, the fusion of these two disciplines can be extremely beneficial as it allows the creation of a receptive and interconnected environment and offers a variety of services to medical professionals and patients. Doctors can make early decisions to save a patient's life when disease forecasts are made early. IoT sensor captures the data from the patients, and machine learning techniques are used to analyze the data and predict the presence of the fatal disease i.e., diabetes. The goal of this research is to make a smart patient's health monitoring system based on machine learning that helps to detect the presence of a chronic disease in patient early and accurately. For the implementation, the diabetic dataset has been used. In order to detect the presence of the fatal disease, six different machine learning techniques are used i.e., Support Vector Machine (SVM), Logistic Regression, Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and Long Short-Term Memory (LSTM). The performance of the proposed model is evaluated by using four evaluation metrics i.e., accuracy, precision, recall, and F1-Score. The RNN outperformed remaining algorithms in terms of accuracy (81%), precision (75%), and F1-Score (65%). However, the recall (56%) for ANN was higher as compared to SVM and logistic regression, CNN, RNN, and LSTM. With the help of this proposed patient's health monitoring system, doctors will be able to diagnose the presence of the disease earlier.

Designing optimized drug candidates with Generative Adversarial Network

Drug design is an important area of study for pharmaceutical businesses. However, low efficacy, off-target delivery, time consumption, and high cost are challenges and can create barriers that impact this process. Deep Learning models are emerging as a promising solution to perform de novo drug design, i.e., to generate drug-like molecules tailored to specific needs. However, stereochemistry was not explicitly considered in the generated molecules, which is inevitable in targeted-oriented molecules. This paper proposes a framework based on Feedback Generative Adversarial Network (GAN) that includes optimization strategy by incorporating Encoder-Decoder, GAN, and Predictor deep models interconnected with a feedback loop. The Encoder-Decoder converts the string notations of molecules into latent space vectors, effectively creating a new type of molecular representation. At the same time, the GAN can learn and replicate the training data distribution and, therefore, generate new compounds. The feedback loop is designed to incorporate and evaluate the generated molecules according to the multiobjective desired property at every epoch of training to ensure a steady shift of the generated distribution towards the space of the targeted properties. Moreover, to develop a more precise set of molecules, we also incorporate a multiobjective optimization selection technique based on a non-dominated sorting genetic algorithm. The results demonstrate that the proposed framework can generate realistic, novel molecules that span the chemical space. The proposed Encoder-Decoder model correctly reconstructs 99% of the datasets, including stereochemical information. The model's ability to find uncharted regions of the chemical space was successfully shown by optimizing the unbiased GAN to generate molecules with a high binding affinity to the Kappa Opioid and Adenosine [Formula: see text] receptor. Furthermore, the generated compounds exhibit high internal and external diversity levels 0.88 and 0.94, respectively, and uniqueness.

Ensemble Machine Learning Model for Accurate Air Pollution Detection Using Commercial Gas Sensors

This paper presents the results on developing an ensemble machine learning model to combine commercial gas sensors for accurate concentration detection. Commercial gas sensors have the low-cost advantage and become key components of IoT devices in atmospheric condition monitoring. However, their native coarse resolution and poor selectivity limit their performance. Thus, we adopted recurrent neural network (RNN) models to extract the time-series concentration data characteristics and improve the detection accuracy. Firstly, four types of RNN models, LSTM and GRU, Bi-LSTM, and Bi-GRU, were optimized to define the best-performance single weak models for CO, O₃, and NO₂ gases, respectively. Next, ensemble models which integrate multiple single weak models with a dynamic model were defined and trained. The testing results show that the ensemble models perform better than the single weak models. Further, a retraining procedure was proposed to make the ensemble model more flexible to adapt to environmental conditions. The significantly improved determination coefficients show that the retraining helps the ensemble models maintain long-term stable sensing performance in an atmospheric environment. The result can serve as an essential reference for the applications of IoT devices with commercial gas sensors in environment condition monitoring.

Comparison of neural basis expansion analysis for interpretable time series (N-BEATS) and recurrent neural networks for heart dysfunction classification

OBJECTIVE

The primary purpose of this work is to analyze the ability of N-BEATS architecture for the problem of prediction and classification of electrocar- diogram (ECG) signals. To achieve this, performance comparison with various types of other SotA (state-of-the-art) recurrent neural network architectures commonly used for such problems is conducted.

APPROACH

Four architectures (N-BEATS, LSTM, LSTM with peepholes, GRU) were tested for performance and dimension reduction problems for different number of leads (2, 3, 4, 6, 12), both in variants consisting of blended branches, allowing retaining ac- curacy while reducing the computational capacity needed. The analysis was performed on datasets and using metrics from Challenges in Cardiology (CinC) 2021 competition.

MAIN RESULTS

Best results were achieved for LSTM with peepholes, then LSTM, GRU and the worst for N-BEATS (challenge metrics respectively: 0.42, 0.40, 0.39, 0.35; for times: 0.0395 s, 0.0036 s, 0.0027 s, 0.0002 s). Commonly used LSTM outperforms N- BEATS in terms of multi-label classification, data set resilience, and obtained challenge metrics. Still, N-BEATS can obtain acceptable results for 2 lead classification (metric of 0.35 for N-BEATS and 0.38 for other networks) and outperforms other solutions in terms of complexity and speed.

SIGNIFICANCE

This paper features a novel approach of using the N-BEATS, which was previously used only for forecasting, to classify ECG signals with success. While N- BEATS multi-label classification capacity is lower than LSTM, its speed obtaining results with a reduced number of leads (faster by one to two degrees of magnitude) allows for arrhythmias detection and classification while using off-the-shelf wearable devices (Holter monitors, sport bands, etc.).

DPAI: A Data-driven simulation-assisted-Physics learned AI model for transient ultrasonic wave propagation

In this paper, we propose a deep neural network model to simulate the transient ultrasonic wave propagation in the 2D domain by implementing the Data driven-simulation-assisted-Physics learned AI (DPAI) model. The DPAI model consists of modified convolutional long short-term memory (ConvLSTM) with an encoder-decoder structure, which learns the representation of spatio-temporal dependence from input sequence data. The DPAI uses the data-driven approach to understand the underlying physics of elastic wave propagation in a medium. This model is trained with simulation-assisted finite element simulation datasets consisting of distributed single and multi-point excitation sources in the medium. The effectiveness of the proposed approach is demonstrated by modeling a wide range of scenarios in elastodynamic physics, such as multiple point sources, varying excitation parameters, and wave propagation in a large 2D domain. The trained DPAI model is tested and compared against FE modeling with respect to accuracy and computational time.

Time series predicting of COVID-19 based on deep learning

COVID-19 was declared a global pandemic by the World Health Organisation (WHO) on 11th March 2020. Many researchers have, in the past, attempted to predict a COVID outbreak and its effect. Some have regarded time-series variables as primary factors which can affect the onset of infectious diseases like influenza and severe acute respiratory syndrome (SARS). In this study, we have used public datasets provided by the European Centre for Disease Prevention and Control for developing a prediction model for the spread of the COVID-19 outbreak to and throughout Malaysia, Morocco and Saudi Arabia. We have made use of certain effective deep learning (DL) models for this purpose. We assessed some specific major features for predicting the trend of the existing COVID-19 outbreak in these three countries. In this study, we also proposed a DL approach that includes recurrent neural network (RNN) and long short-term memory (LSTM) networks for predicting the probable numbers of COVID-19 cases. The LSTM models showed a 98.58% precision accuracy while the RNN models showed a 93.45% precision accuracy. Also, this study compared the number of coronavirus cases and the number of resulting deaths in Malaysia, Morocco and Saudi Arabia. Thereafter, we predicted the number of confirmed COVID-19 cases and deaths for a subsequent seven days. In this study, we presented their predictions using the data that was available up to December 3rd, 2020.

Predicting continuous ground reaction forces from accelerometers during uphill and downhill running: a recurrent neural network solution

BACKGROUND

Ground reaction forces (GRFs) are important for understanding human movement, but their measurement is generally limited to a laboratory environment. Previous studies have used neural networks to predict GRF waveforms during running from wearable device data, but these predictions are limited to the stance phase of level-ground running. A method of predicting the normal (perpendicular to running surface) GRF waveform using wearable devices across a range of running speeds and slopes could allow researchers and clinicians to predict kinetic and kinematic variables outside the laboratory environment.

PURPOSE

We sought to develop a recurrent neural network capable of predicting continuous normal (perpendicular to surface) GRFs across a range of running speeds and slopes from accelerometer data.

METHODS

Nineteen subjects ran on a force-measuring treadmill at five slopes (0°, ±5°, ±10°) and three speeds (2.5, 3.33, 4.17 m/s) per slope with sacral- and shoe-mounted accelerometers. We then trained a recurrent neural network to predict normal GRF waveforms frame-by-frame. The predicted versus measured GRF waveforms had an average ± SD RMSE of 0.16 ± 0.04 BW and relative RMSE of 6.4 ± 1.5% across all conditions and subjects.

RESULTS

The recurrent neural network predicted continuous normal GRF waveforms across a range of running speeds and slopes with greater accuracy than neural networks implemented in previous studies. This approach may facilitate predictions of biomechanical variables outside the laboratory in near real-time and improves the accuracy of quantifying and monitoring external forces experienced by the body when running.

Has Artificial Intelligence Impacted Drug Discovery?

Artificial intelligence (AI) tools find increasing application in drug discovery supporting every stage of the Design-Make-Test-Analyse (DMTA) cycle. The main focus of this chapter is the application in molecular generation with the aid of deep neural networks (DNN). We present a historical overview of the main advances in the field. We analyze the concepts of distribution and goal-directed learning and then highlight some of the recent applications of generative models in drug design with a focus into research work from the biopharmaceutical industry. We present in some more detail REINVENT which is an open-source software developed within our group in AstraZeneca and the main platform for AI molecular design support for a number of medicinal chemistry projects in the company and we also demonstrate some of our work in library design. Finally, we present some of the main challenges in the application of AI in Drug Discovery and different approaches to respond to these challenges which define areas for current and future work.

A deep learning framework using CNN and stacked Bi-GRU for COVID-19 predictions in India

The novel coronavirus infection (COVID-19) first appeared in Wuhan, China, in December 2019. COVID-19 declared as a global pandemic by the WHO was the most rapidly spreading disease all across the world. India, the second most populated nation in the world, is still fighting it, when coronavirus reached the stage where community transmission takes place at an exponential rate. Therefore, it is crucial to examine the future trends of COVID-19 in India and anticipate how it will affect economic and social growth in a short run. In this paper, a new deep learning framework using CNN and stacked Bi-GRU has been developed for predicting and analyzing the COVID-19 cases in India. The proposed model can predict the next 30 days' new positive cases, new death cases, recovery rate and containment and health index values with high accuracy. The proposed method is compared against Gaussian process regression (GPR) model on COVID-19 datasets. The experimental result shows that the proposed framework is highly reliable for COVID-19 prediction over the GPR model.

CRNNTL: Convolutional Recurrent Neural Network and Transfer Learning for QSAR Modeling in Organic Drug and Material Discovery

Molecular latent representations, derived from autoencoders (AEs), have been widely used for drug or material discovery over the past couple of years. In particular, a variety of machine learning methods based on latent representations have shown excellent performance on quantitative structure–activity relationship (QSAR) modeling. However, the sequence feature of them has not been considered in most cases. In addition, data scarcity is still the main obstacle for deep learning strategies, especially for bioactivity datasets. In this study, we propose the convolutional recurrent neural network and transfer learning (CRNNTL) method inspired by the applications of polyphonic sound detection and electrocardiogram classification. Our model takes advantage of both convolutional and recurrent neural networks for feature extraction, as well as the data augmentation method. According to QSAR modeling on 27 datasets, CRNNTL can outperform or compete with state-of-art methods in both drug and material properties. In addition, the performances on one isomers-based dataset indicate that its excellent performance results from the improved ability in global feature extraction when the ability of the local one is maintained. Then, the transfer learning results show that CRNNTL can overcome data scarcity when choosing relative source datasets. Finally, the high versatility of our model is shown by using different latent representations as inputs from other types of AEs.

Deep-Data-Driven Neural Networks for COVID-19 Vaccine Efficacy

Vaccination strategies to lessen the impact of the spread of a disease are fundamental to public health authorities and policy makers. The socio-economic benefit of full return to normalcy is the core of such strategies. In this paper, a COVID-19 vaccination model with efficacy rate is developed and analyzed. The epidemiological parameters of the model are learned via a feed-forward neural network. A hybrid approach that combines residual neural network with variants of recurrent neural network is implemented and analyzed for reliable and accurate prediction of daily cases. The error metrics and a k-fold cross validation with random splitting reveal that a particular type of hybrid approach called residual neural network with gated recurrent unit is the best hybrid neural network architecture. The data-driven simulations confirm the fact that the vaccination rate with higher efficacy lowers the infectiousness and basic reproduction number. As a study case, COVID-19 data for the state of Tennessee in USA is used.

A Deep Learning Based Approach for Localization and Recognition of Pakistani Vehicle License Plates

License plate localization is the process of finding the license plate area and drawing a bounding box around it, while recognition is the process of identifying the text within the bounding box. The current state-of-the-art license plate localization and recognition approaches require license plates of standard size, style, fonts, and colors. Unfortunately, in Pakistan, license plates are non-standard and vary in terms of the characteristics mentioned above. This paper presents a deep-learning-based approach to localize and recognize Pakistani license plates with non-uniform and non-standardized sizes, fonts, and styles. We developed a new Pakistani license plate dataset (PLPD) to train and evaluate the proposed model. We conducted extensive experiments to compare the accuracy of the proposed approach with existing techniques. The results show that the proposed method outperformed the other methods to localize and recognize non-standard license plates.

Assessing the Effect of Climate Variables on the Incidence of Dengue Cases in the Metropolitan Region of Panama City

The present analysis uses the data of confirmed incidence of dengue cases in the metropolitan region of Panama from 1999 to 2017 and climatic variables (air temperature, precipitation, and relative humidity) during the same period to determine if there exists a correlation between these variables. In addition, we compare the predictive performance of two regression models (SARIMA, SARIMAX) and a recurrent neural network model (RNN-LSTM) on the dengue incidence series. For this data from 1999–2014 was used for training and the three subsequent years of incidence 2015–2017 were used for prediction. The results show a correlation coefficient between the climatic variables and the incidence of dengue were low but statistical significant. The RMSE and MAPE obtained for the SARIMAX and RNN-LSTM models were 25.76, 108.44 and 26.16, 59.68, which suggest that any of these models can be used to predict new outbreaks. Although, it can be said that there is a limited role of climatic variables in the outputs the models. The value of this work is that it helps understand the behaviour of cases in a tropical setting as is the Metropolitan Region of Panama City, and provides the basis needed for a much needed early alert system for the region.

Deep Neural Networks for Human's Fall-risk Prediction using Force-Plate Time Series Signal

Early and accurate identification of the balance deficits could reduce falls, in particular for older adults, a prone population. Our work investigates deep neural networks' capacity to identify human balance patterns towards predicting fall-risk. Human balance ability can be characterized based on commonly-used balance metrics, such as those derived from the force-plate time series. We hypothesized that low, moderate, and high risk of falling can be characterized based on balance metrics, derived from the force-plate time series, in conjunction with deep learning algorithms. Further, we predicted that our proposed One-One-One Deep Neural Networks algorithm provides a considerable increase in performance compared to other algorithms. Here, an open source force-plate dataset, which quantified human balance from a wide demographic of human participants (163 females and males aged 18-86) for varied standing conditions (eyes-open firm surface, eyes-closed firm surface, eyes-open foam surface, eyes-closed foam surface) was used. Classification was based on one of the several indicators of fall-risk tied to the fear of falling: the clinically-used Falls Efficacy Scale (FES) assessment. For human fall-risk prediction, the deep learning architecture implemented comprised of: Recurrent Neural Network (RNN), Long-Short Time Memory (LSTM), One Dimensional Convolutional Neural Network (1D-CNN), and a proposed One-One-One Deep Neural Network. Results showed that our One-One-One Deep Neural Networks algorithm outperformed the other aforementioned algorithms and state-of-the-art models on the same dataset. With an accuracy, precision, and sensitivity of 99.9%, 100%, 100%, respectively at the 12th epoch, we found that our proposed One-One-One Deep Neural Network model is the most efficient neural network in predicting human's fall-risk (based on the FES measure) using the force-plate time series signal. This is a novel methodology for an accurate prediction of human risk of fall.