Dimensional Convolutional Neural Network Research Articles

High Dimensional Convolutional Neural Network for EEG Connectivity-Based Diagnosis of ADHD.

Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disorder in children and adults and its early detection is effective in the successful treatment of children. Electroencephalography (EEG) has been widely used for classifying ADHD and normal children. In recent years, deep learning leads to more accurate classification. This study aims to adapt convolutional neural networks (CNNs) for classifying ADHD and normal children based on the connectivity measure of their EEG signals. In this experimental study, the dataset consisted of 61 ADHD and 60 normal children from which 13021 epochs were extracted as input for model training and evaluation. Synchronization likelihood (SL) and wavelet coherence (WC) were considered connectivity measures. The neighborhood between EEG channels was arranged in a two-dimensional matrix for better representation. Four-dimensional (4D) and six-dimensional (6D) connectivity tensors were composed as model inputs. Two architectures were developed, one 4D and 6D CNN for SL and WC-based diagnosis of ADHD, respectively. A 5-fold cross-validation was utilized to assess developed models. The average accuracy of 98.56% for 4D CNN and 98.85% for 6D CNN in epoch-based classification were obtained. In the case of subject-based classification, the accuracy was 99.17% for both models. Based on the evaluation metrics of the proposed models, ADHD children can be diagnosed and ADHD and normal children can be successfully distinguished.

UCRNet: Underwater color image restoration via a polarization-guided convolutional neural network

Underwater images always suffer from low contrast and color distortion due to the wavelength-dependent scattering and absorption effects caused by particles existing in turbid water, especially in high turbidity conditions. Based on the polarization properties of the backscattering light, polarimetric methods can estimate the intensity level of the backscattering and the transmittance of the media. Accordingly, they can separate the target signal from the undesired ones to achieve high-quality imaging. In addition, learning-based polarimetric methods are effective for gray-model image restoration, but the learning-based polarimetric technique for color image restoration has yet to be considered. In this paper, we propose a 3- dimensional convolutional neural network, which maintains the correlation of polarization information among different polarization channel images as well as embodies polarization constraints, for underwater color image restoration. The experimental results verify that the proposed solution improves the image quality (i.e., the image contrast, details, and color) and outperforms other existing methods, especially when the turbidity of scattering media is high. The proposed solution can be readily applied to practical applications and potentially realize the clear vision in other scattering media, including biomedical imaging and remote sensing.

Improving tuberculosis severity assessment in computed tomography images using novel DAvoU-Net segmentation and deep learning framework

Tuberculosis is a bacterial infection that usually affects the lungs; however, it can be treated if diagnosed earlier. Effective analysis of computed tomography images of lungs captured from tuberculosis patients is one of the effective approaches to determining the disease's severity. Previous medical computed tomography image analysis algorithms have suffered many setbacks during the segmentation process. In this study, a novel approach to image segmentation is proposed. DAvoU-Net model comprises Multi-Scale Residual Block and Receptive Dense Connection to segment tuberculosis-affected areas from original CT images. The 2-dimensional weight parameters of seven different deep pre-trained neural networks were first converted into 3-dimensional weights using a rotation operation before being used to train a new 3-dimensional convolutional neural network. The extracted image features from the 3-dimensional convolutional neural network serve as input into a bi-directional long short-time memory for extracting high-level discriminative features. The proposed DAvoU-Net model outperformed the original U-Net model in segmenting ImageCLEF 2019 tuberculosis dataset. It generated a dice score of 98.34 % compared to 84.21 % produced by the original U-Net. Overall, the combination of DAvoU-Net + ResNet-50 along with a 3-dimensional convolutional neural network and bi-directional long short-time memory performed better than the six other deep pre-trained neural networks. It produced an overall Area Under the ROC Curve (AUC) and a Classification Accuracy (ACC) of 0.8513 and 0.8119 respectively.

Deep-Learning–Aided Diagnosis of Diabetic Retinopathy, Age-Related Macular Degeneration, and Glaucoma Based on Structural and Angiographic OCT

Timely diagnosis of eye diseases is paramount to obtaining the best treatment outcomes. OCT and OCT angiography (OCTA) have several advantages that lend themselves to early detection of ocular pathology; furthermore, the techniques produce large, feature-rich data volumes. However, the full clinical potential of both OCT and OCTA is stymied when complex data acquired using the techniques must be manually processed. Here, we propose an automated diagnostic framework based on structural OCT and OCTA data volumes that could substantially support the clinical application of these technologies. Cross sectional study. Five hundred twenty-six OCT and OCTA volumes were scanned from the eyes of 91 healthy participants, 161 patients with diabetic retinopathy (DR), 95 patients with age-related macular degeneration (AMD), and 108 patients with glaucoma. The diagnosis framework was constructed based on semisequential 3-dimensional (3D) convolutional neural networks. The trained framework classifies combined structural OCT and OCTA scans as normal, DR, AMD, or glaucoma. Fivefold cross-validation was performed, with 60% of the data reserved for training, 20% for validation, and 20% for testing. The training, validation, and test data sets were independent, with no shared patients. For scans diagnosed as DR, AMD, or glaucoma, 3D class activation maps were generated to highlight subregions that were considered important by the framework for automated diagnosis. The area under the curve (AUC) of the receiver operating characteristic curve and quadratic-weighted kappa were used to quantify the diagnostic performance of the framework. For the diagnosis of DR, the framework achieved an AUC of 0.95 ± 0.01. For the diagnosis of AMD, the framework achieved an AUC of 0.98 ± 0.01. For the diagnosis of glaucoma, the framework achieved an AUC of 0.91 ± 0.02. Deep learning frameworks can provide reliable, sensitive, interpretable, and fully automated diagnosis of eye diseases. Proprietary or commercial disclosure may be found after the references.

Abstract 9903: Artificial Intelligence Driven Wearable 2-minute Seismocardiography Test for Detection of Aortic Valve Stenosis Severity

Introduction: Standard diagnosis of aortic valve stenosis (AS) requires clinical imaging by echocardiography or cardiac MRI. The purpose of this study was to test the performance of easy to use, artificial intelligent (AI) driven seismocardiogram (SCG) chest acceleration recordings for point of care screening and identification of AS. Hypothesis: SCG measurements can be used by a low dimensional convolutional neural network (CNN) to accurately predict presence or absence of aortic valve stenosis (AS) and severity of dysfunction. Methods: This IRB approved study included 45 controls (age 46±17y, 20 F, no known cardiovascular disease) and 30 aortic valve disease (AVD) patients (age 30±20y, 7 F, 21 bicuspid). A wearable SCG device (integrated ECG/accelerometer microelectronics) was placed for 2min chest acceleration recordings on the same day as standard of care echocardiography or cardiac MRI, which was used as ground truth data to classify valve function (normal: AS-, mild AS: AS+, moderate/severe AS: AS++). A deep learning CNN (2х2 kernel, softmax/cross entropy classification) was configured to predict/detect degree of AS from spectrally decomposed SCG measurements. Randomized 50% split of subjects was used for CNN training/testing. Performance detecting AS was evaluated for classification accuracy. Results: AI driven SCG detection of valve function abnormality had high sensitivity/specificity (90%, 67%) with a net area under curve (AUC) 0.86. Correct classification of the degree of AS was achieved in 76% of subjects. Normal (AS-) vs abnormal function (AS+ or AS++) was accurately identified by AI driven SCG for 86% of subjects. Conclusions: This feasibility study demonstrates that the combination of easily measured 2 minute SCG with a deep processing pipeline could successfully detect the presence and severity of AS, supporting clinical promise for potential use as a low cost, front line screening tool to identify patients with aortic valve dysfunction.

Application of Deep Learning Techniques and Bayesian Optimization with Tree Parzen Estimator in the Classification of Supply Chain Pricing Datasets of Health Medications

From the development and sale of a product through its delivery to the end customer, the supply chain encompasses a network of suppliers, transporters, warehouses, distribution centers, shipping lines, and logistics service providers all working together. Lead times, bottlenecks, cash flow, data management, risk exposure, traceability, conformity, quality assurance, flaws, and language barriers are some of the difficulties that supply chain management faces. In this paper, deep learning techniques such as Long Short-Term Memory (LSTM) and One Dimensional Convolutional Neural Network (1D-CNN) were adopted and applied to classify supply chain pricing datasets of health medications. Then, Bayesian optimization using the tree parzen estimator and All K Nearest Neighbor (AllkNN) was used to establish the suitable model hyper-parameters of both LSTM and 1D-CNN to enhance the classification model. Repeated five-fold cross-validation is applied to the developed models to predict the accuracy of the models. The study showed that the combination of 1D-CNN, AllkNN, and Bayesian optimization (1D-CNN+AllKNN+BO) outperforms other approaches employed in this study. The accuracy of the combination of 1D-CNN, AllkNN, and Bayesian optimization (1D-CNN+AllKNN+BO) from one-fold to 10-fold, produced the highest range between 61.2836% and 63.3267%, among other models.

A 3D convolutional neural network accurately predicts the permeability of gas diffusion layer materials directly from image data

Three dimensional convolutional neural networks (3D CNNs) were used to accurately predict the permeability of fuel cell gas diffusion layer (GDL) materials directly from 3D binary image data. A straightforward 3D CNN architecture was trained and evaluated on a dataset of 7500 numerically generated GDL materials using five-fold cross validation. The permeability of the generated GDL materials was calculated using pore network modelling and comported with literature reported values. The 3D CNN converged to acceptable prediction error quickly and with minimal hand tuning of hyperparameters. The trained 3D CNN had low prediction error when compared to contemporary models performing similar tasks, with an average mean absolute error (MAE) of 4.660E-13 m2 and an R2 of 0.9885, when evaluated on test fold materials. Lastly, a learning curve analysis was also performed.

Hybrid 1D-CNN and attention-based Bi-GRU neural networks for predicting moisture content of sand gravel using NIR spectroscopy

A non-destructive and rapid moisture content detection method of sand gravel material is required in loose material dams. The near-infrared (NIR) spectrum of sand materials is closely related to its moisture content. Recently, there is a growing need for fully using spectral information when establishing calibration models for sand gravel moisture content detection. To address these issues, a hybrid one dimensional-convolutional neural network (1D-CNN) and attention-based bidirectional gated recurrent unit (Bi-GRU) neural network was proposed to detect sand gravel moisture content with NIR spectrum. Two learners, namely, 1D-CNN and Bi-GRU, were constructed to extract local abstract information and sequence position information from the spectrum, respectively. In the 1D-CNN learner, multiple kernels CNN layers and one dimensional-separable convolution layers were conjunct to improve model accuracy and reduce network parameters. In the Bi-GRU learner, a multi-head self-attention mechanism was appended to evaluate the weights of the output features extracted by Bi-GRU layers. The proposed model achieved the best prediction results in LUCAS dataset (R2 greater than 0.75, RPD greater than 2.0) and our sand gravel spectral dataset (R2 = 0.96, RPD = 5.06) compared to other deep learning and conventional spectroscopy analysis methods. In addition, the top ten characteristic wavelength points of sand gravel were identified. These can be used to choose a discrete spectrum measuring instrument, which has a relatively low cost.

2-D Attention Based Convolutional Recurrent Neural Network for Speech Emotion Recognition

Recognizing speech emotions is a formidable challenge due to the complexity of emotions. The function of Speech Emotion Recognition(SER) is significantly impacted by the effects of emotional signals retrieved from speech. The majority of emotional traits, on the other hand, are sensitive to emotionally neutral elements like the speaker, speaking manner, and gender. In this work, we postulate that computing deltas for individual features maintain useful information which is mainly relevant to emotional traits while it minimizes the loss of emotionally irrelevant components, thus leading to fewer misclassifications. Additionally, Speech Emotion Recognition(SER) commonly experiences silent and emotionally unrelated frames. The proposed technique is quite good at picking up important feature representations for emotion relevant features. So here is a two dimensional convolutional recurrent neural network that is attention-based to learn distinguishing characteristics and predict the emotions. The Mel-spectrogram is used for feature extraction. The suggested technique is conducted on IEMOCAP dataset and it has better performance, with 68% accuracy value.

A hybrid CNN-BiLSTM approach for remaining useful life prediction of EVs lithium-Ion battery

For accelerating the technology development and facilitating the reliable operation of lithium-ion batteries, accurate prediction for battery remaining useful life (RUL) are both critical. In this paper, a 1D CNN-BiLSTM method is proposed to extract the RUL prediction of lithium-ion battery of Electric Vehicles (EVs). By using one dimensional convolutional neural network (1D CNN) and bidirectional long short-term memory (BiLSTM) neural network simultaneously, selecting the ELU activation function to apply to the convolutional layer, a hybrid neural network is proposed to improve the accuracy and stability of lithium-ion battery RUL prediction. The 1D CNN is used to fully mine the deep features of lithium-ion SOH data, while the BiLSTM is adopted to study the deep features in two directions, and the RUL prediction of lithium-ion battery is output through dense layer. To verify the effectiveness of the proposed method, the battery data of the National Aeronautics and Space Administration (NASA) are utilized to make some comparisons among the RNN model, LSTM model, BiLSTM model and hybrid neural network model. The results show that the hybrid one has higher generalization ability and prediction accuracy than the others.

Validation of a convolutional neural network for the automated creation of curved planar reconstruction images along the main pancreatic duct

PurposeTo evaluate the accuracy and time-efficiency of newly developed software in automatically creating curved planar reconstruction (CPR) images along the main pancreatic duct (MPD), which was developed based on a 3-dimensional convolutional neural network, and compare them with those of conventional manually generated CPR ones.Materials and methodsA total of 100 consecutive patients with MPD dilatation (≥ 3 mm) who underwent contrast-enhanced computed tomography between February 2021 and July 2021 were included in the study. Two radiologists independently performed blinded qualitative analysis of automated and manually created CPR images. They rated overall image quality based on a four-point scale and weighted κ analysis was employed to compare between manually created and automated CPR images. A quantitative analysis of the time required to create CPR images and the total length of the MPD measured from CPR images was performed.ResultsThe κ value was 0.796, and a good correlation was found between the manually created and automated CPR images. The average time to create automated and manually created CPR images was 61.7 s and 174.6 s, respectively (P < 0.001). The total MPD length of the automated and manually created CPR images was 110.5 and 115.6 mm, respectively (P = 0.059).ConclusionThe automated CPR software significantly reduced reconstruction time without compromising image quality.

FASTNN: A Deep Learning Approach for Traffic Flow Prediction Considering Spatiotemporal Features.

Traffic flow forecasting is a critical input to intelligent transportation systems. Accurate traffic flow forecasting can provide an effective reference for implementing traffic management strategies, developing travel route planning, and public transportation risk assessment. Recent deep learning approaches of spatiotemporal neural networks to predict traffic flow show promise, but could be difficult to separately model the spatiotemporal aggregation in traffic data and intrinsic correlation or redundancy of spatiotemporal features extracted by the filter of the convolutional network. This can introduce biases in the predictions that interfere with subsequent planning decisions in transportation. To solve the mentioned problem, the filter attention-based spatiotemporal neural network (FASTNN) was proposed in this paper. First, the model used 3-dimensional convolutional neural networks to extract universal spatiotemporal dependencies from three types of historical traffic flow, the residual units were employed to prevent network degradation. Then, the filter spatial attention module was constructed to quantify the spatiotemporal aggregation of the features, thus enabling dynamic adjustment of the spatial weights. To model the intrinsic correlation and redundancy of features, this paper also constructed a lightweight module, named matrix factorization based resample module, which automatically learned the intrinsic correlation of the same features to enhance the concentration of the model on information-rich features, and used matrix factorization to reduce the redundant information between different features. The FASTNN has experimented on two large-scale real datasets (TaxiBJ and BikeNYC), and the experimental results show that the FASTNN has better prediction performance than various baselines and variant models.

Assessment of the uncertainty and interpretability of deep learning models for mapping soil salinity using DeepQuantreg and game theory

This research introduces a new combined modelling approach for mapping soil salinity in the Minab plain in southern Iran. This study assessed the uncertainty (with 95% confidence limits) and interpretability of two deep learning (DL) models (deep boltzmann machine—DBM) and a one dimensional convolutional neural networks (1DCNN)—long short-term memory (LSTM) hybrid model (1DCNN-LSTM) for mapping soil salinity by applying DeepQuantreg and game theory (Shapely Additive exPlanations (SHAP) and permutation feature importance measure (PFIM)), respectively. Based on stepwise forward regression (SFR)—a technique for controlling factor selection, 18 of 47 potential controls were selected as effective factors. Inventory maps of soil salinity were generated based on 476 surface soil samples collected for measuring electrical conductivity (ECe). Based on Taylor diagrams, both DL models performed well (RMSE < 20%), but the 1DCNN-LSTM hybrid model performed slightly better than the DBM model. The uncertainty range associated with the ECe values predicted by both models estimated using DeepQuantilreg were similar (0–25 dS/m for the 1DCNN-LSTM hybrid model and 2–27 dS/m for DBM model). Based on the SFR and PFIM (permutation feature importance measure)—a measure in game theory, four controls (evaporation, sand content, precipitation and vertical distance to channel) were selected as the most important factors for soil salinity in the study area. The results of SHAP (Shapely Additive exPlanations)—the second measure used in game theory—suggested that five factors (evaporation, vertical distance to channel, sand content, cation exchange capacity (CEC) and digital elevation model (DEM)) have the strongest impact on model outputs. Overall, the methodology used in this study is recommend for applications in other regions for mapping environmental problems.

Finding the Signal from the Smoke

Cooking fires are dangerous. Every year, they are responsible for taking away more than 500 lives in the U.S. alone. Existing approaches using sensors usually require expensive retrofitting and are not feasible in real-life situations. This research presents Finding-Signals-from-Smoke (FiSS), a robust fire machine learning prediction model that aims to prevent cooking fires from starting using videos captured with a normal camera. FiSS is based on a 3-dimensional Convolutional Neural Network, which analyzes the video signals and models the complex relationships of the spatial-temporal features of smoke signals with fire ignition. It uses a segment-based video sampling and modeling framework that is able to generalize to a variety of kitchen/stove settings and achieve promising prediction performance. FiSS is trained and evaluated with video data from 30 full-scale kitchen fire experiments and can predict potential fire ignitions as early as 60 seconds before the moment of ignition. As a result, FiSS can be used in an early warning system to prevent fire ignitions and help to reduce casualties and injuries from cooking fires.

CNN-XGBoost fusion-based affective state recognition using EEG spectrogram image analysis

Recognizing emotional state of human using brain signal is an active research domain with several open challenges. In this research, we propose a signal spectrogram image based CNN-XGBoost fusion method for recognising three dimensions of emotion, namely arousal (calm or excitement), valence (positive or negative feeling) and dominance (without control or empowered). We used a benchmark dataset called DREAMER where the EEG signals were collected from multiple stimulus along with self-evaluation ratings. In our proposed method, we first calculate the Short-Time Fourier Transform (STFT) of the EEG signals and convert them into RGB images to obtain the spectrograms. Then we use a two dimensional Convolutional Neural Network (CNN) in order to train the model on the spectrogram images and retrieve the features from the trained layer of the CNN using a dense layer of the neural network. We apply Extreme Gradient Boosting (XGBoost) classifier on extracted CNN features to classify the signals into arousal, valence and dominance of human emotion. We compare our results with the feature fusion-based state-of-the-art approaches of emotion recognition. To do this, we applied various feature extraction techniques on the signals which include Fast Fourier Transformation, Discrete Cosine Transformation, Poincare, Power Spectral Density, Hjorth parameters and some statistical features. Additionally, we use Chi-square and Recursive Feature Elimination techniques to select the discriminative features. We form the feature vectors by applying feature level fusion, and apply Support Vector Machine (SVM) and Extreme Gradient Boosting (XGBoost) classifiers on the fused features to classify different emotion levels. The performance study shows that the proposed spectrogram image based CNN-XGBoost fusion method outperforms the feature fusion-based SVM and XGBoost methods. The proposed method obtained the accuracy of 99.712% for arousal, 99.770% for valence and 99.770% for dominance in human emotion detection.

RSS-Based Wireless LAN Indoor Localization and Tracking Using Deep Architectures

Wireless Local Area Network (WLAN) positioning is a challenging task indoors due to environmental constraints and the unpredictable behavior of signal propagation, even at a fixed location. The aim of this work is to develop deep learning-based approaches for indoor localization and tracking by utilizing Received Signal Strength (RSS). The study proposes Multi-Layer Perceptron (MLP), One and Two Dimensional Convolutional Neural Networks (1D CNN and 2D CNN), and Long Short Term Memory (LSTM) deep networks architectures for WLAN indoor positioning based on the data obtained by actual RSS measurements from an existing WLAN infrastructure in a mobile user scenario. The results, using different types of deep architectures including MLP, CNNs, and LSTMs with existing WLAN algorithms, are presented. The Root Mean Square Error (RMSE) is used as the assessment criterion. The proposed LSTM Model 2 achieved a dynamic positioning RMSE error of 1.73m, which outperforms probabilistic WLAN algorithms such as Memoryless Positioning (RMSE: 10.35m) and Nonparametric Information (NI) filter with variable acceleration (RMSE: 5.2m) under the same experiment environment.

Automated Lung Cancer Segmentation Using a PET and CT Dual-Modality Deep Learning Neural Network

To develop an automated lung tumor segmentation method for radiation therapy planning based on deep learning and dual-modality positron emission tomography (PET) and computed tomography (CT) images. A 3-dimensional (3D) convolutional neural network using inputs from diagnostic PETs and simulation CTs was constructed with 2 parallel convolution paths for independent feature extraction at multiple resolution levels and a single deconvolution path. At each resolution level, the extracted features from the convolution arms were concatenated and fed through the skip connections into the deconvolution path that produced the tumor segmentation. Our network was trained/validated/tested by a 3:1:1 split on 290 pairs of PET and CT images from patients with lung cancer treated at our clinic, with manual physician contours as the ground truth. A stratified training strategy based on the magnitude of the gross tumor volume (GTV) was investigated to improve performance, especially for small tumors. Multiple radiation oncologists assessed the clinical acceptability of the network-produced segmentations. The mean Dice similarity coefficient, Hausdorff distance, and bidirectional local distance comparing manual versus automated contours were 0.79 ± 0.10, 5.8 ± 3.2 mm, and 2.8 ± 1.5 mm for the unstratified 3D dual-modality model. Stratification delivered the best results when the model for the large GTVs (>25 mL) was trained with all-size GTVs and the model for the small GTVs (<25 mL) was trained with small GTVs only. The best combined Dice similarity coefficient, Hausdorff distance, and bidirectional local distance from the 2 stratified models on their corresponding test data sets were 0.83 ± 0.07, 5.9 ± 2.5 mm, and 2.8 ± 1.4 mm, respectively. In the multiobserver review, 91.25% manual versus 88.75% automatic contours were accepted or accepted with modifications. By using an expansive clinical PET and CT image database and a dual-modality architecture, the proposed 3D network with a novel GTVbased stratification strategy generated clinically useful lung cancer contours that were highly acceptable on physician review.

Palpation localization of radial artery based on 3-dimensional convolutional neural networks

Palpation localization is essential for detecting physiological parameters of the radial artery for pulse diagnosis of Traditional Chinese Medicine (TCM). Detecting signal or applying pressure at the wrong location can seriously affect the measurement of pulse waves and result in misdiagnosis. In this paper, we propose an effective and high accuracy regression model using 3-dimensional convolution neural networks (CNN) processing near-infrared picture sequences to locate radial artery upon radius at the wrist. Comparing with early studies using 2-dimensional models, 3Dcnn introduces temporal features with the third dimension to leverage pulsation rhythms, and had achieved superior performance accuracy as 0.87 within 50 pixels at testing resolution of 1024 × 544. Model visualization shows that the additional dimension of the temporal convolution highlights dynamic changes within image sequences. This study presents the great potential of our constructed model to be applied in real wrist palpation location scenarios to bring the key convenience for pulse diagnosis.

Detecting Asymptomatic Infections of Rice Bacterial Leaf Blight Using Hyperspectral Imaging and 3-Dimensional Convolutional Neural Network With Spectral Dilated Convolution.

Rice is one of the most important food crops for human beings. Its total production ranks third in the grain crop output. Bacterial Leaf Blight (BLB), as one of the three major diseases of rice, occurs every year, posing a huge threat to rice production and safety. There is an asymptomatic period between the infection and the onset periods, and BLB will spread rapidly and widely under suitable conditions. Therefore, accurate detection of early asymptomatic BLB is very necessary. The purpose of this study was to test the feasibility of detecting early asymptomatic infection of the rice BLB disease based on hyperspectral imaging and Spectral Dilated Convolution 3-Dimensional Convolutional Neural Network (SDC-3DCNN). First, hyperspectral images were obtained from rice leaves infected with the BLB disease at the tillering stage. The spectrum was smoothed by the Savitzky–Golay (SG) method, and the wavelength between 450 and 950 nm was intercepted for analysis. Then Principal Component Analysis (PCA) and Random Forest (RF) were used to extract the feature information from the original spectra as inputs. The overall performance of the SDC-3DCNN model with different numbers of input features and different spectral dilated ratios was evaluated. Lastly, the saliency map visualization was used to explain the sensitivity of individual wavelengths. The results showed that the performance of the SDC-3DCNN model reached an accuracy of 95.4427% when the number of inputs is 50 characteristic wavelengths (extracted by RF) and the dilated ratio is set at 5. The saliency-sensitive wavelengths were identified in the range from 530 to 570 nm, which overlaps with the important wavelengths extracted by RF. According to our findings, combining hyperspectral imaging and deep learning can be a reliable approach for identifying early asymptomatic infection of the rice BLB disease, providing sufficient support for early warning and rice disease prevention.

Optimalisasi Saluran Komunikasi Berbasis Gelombang Mikro Sebagai Alternatif Sistem Pemantauan Curah Hujan

As a vast archipelagic country with diverse topographic conditions and has an annual average rainfall of more than 2000 mm, Indonesia is prone to hydrometeorological disasters. Based on Indonesia's disaster data, throughout 2021 there were 3,658 incidents of floods and landslides distributed throughout Indonesia. This makes real-time rainfall monitoring with high density indispensable. Indonesia currently has a rainfall monitoring system about 1000 automatic rain gauges, so an increase in the spatial resolution of network is necessary. The increasing density of monitoring equipment using rain gauges and weather radar poses the problem of high procurement and operational costs. Therefore, several alternative rainfall monitoring systems are needed. In this article, we review several studies that focus on the utilization of terrestrial and satellite communication link operating in high frequency bands as an alternative for measuring rainfall. Optimization of the satellite communication system network is more suitable than terrestrial networks to be applied in Indonesia with archipelagic areas because it has a large number of point distributions with wider coverage. The use of artificial intelligence with deep learning techniques such as one dimensional convolutional neural network (1D-CNN) is also very promising to estimate rainfall intensity because it has a high accuracy of 93%..