Super Parameters Research Articles

Automatic and Accurate Diagnosis of Alzheimer's Disease from MRI Images Improved by Deep Convolutional Neural Network

Introduction: Identifying and diagnosing Alzheimer's disease in brain tissue is one of the serious challenges of identification in the field of medical image processing. Currently, MRI is the most common way to diagnose Alzheimer's among the imaging methods. Failure to correctly identify the involved tissue can lead to misdiagnosis as healthy brain tissue. Deep learning algorithm extracts useful information as a process of detecting relevant features. In this research, we decided to use convolutional neural network in processing medical images so that we can perform diagnosis with better accuracy compared to previous works. Method: Using the designed convolutional neural network, the features of MRI T1 images have been extracted. Alzheimer's images have been analyzed using Matlab2023a software and the intended outputs have been obtained. Results: Brain Alzheimer's T1 images have been analyzed after pre -processing and entering the designed deep neural network, and in the output of the proposed algorithm, the identification accuracy and identification speed of the algorithm with the improvement of super parameters were higher compared to other common methods, which accuracy was 96% and 100% sensitivity in identification. Conclusion: The purpose of the deep learning model is to make image data with large dimensions and a large number in an understandable form for machines. It is expected that in the future feature extraction will be done more accurately and more details will be available to machine vision systems to recognize objects in the image .

Soil Moisture Prediction Model Based on Improved GRU Recurrent Neural Network

Soil moisture plays a crucial role in land water and energy cycles, and has a certain impact on weather and climate change. In agricultural production, crop moisture status can be determined based on soil moisture, and timely and effective irrigation strategies can be formulated to ensure grain yield while saving water resources, maximizing the value of agricultural water resource utilization, and achieving sustainable development. Therefore, the accuracy of soil moisture prediction has important research value for many fields such as agriculture and climate. In this paper, the super parameters of GRU Recurrent neural network are optimized by intelligent seagull optimization algorithm using a small number of influencing factors, namely, atmospheric temperature, atmospheric humidity, rainfall and soil moisture data, and a soil moisture prediction model is established. The model was used to predict soil moisture for the next 12 hours, 24 hours, 36 hours, and 48 hours, respectively. The final experiment showed that the model in this paper had better predictive effect on soil moisture, with the best predictive evaluation index data being MAPE (12h) = 4.4120%, R2 (12h) = 0.94605, and RMSE (12h) = 1.9998; By comparing the prediction results of multiple time steps vertically, it was found that the prediction accuracy of the model in this paper decreased more smoothly, meeting the requirements of soil moisture prediction.

Daily Peak-Valley Electric-Load Forecasting Based on an SSA-LSTM-RF Algorithm

In recent years, with the development of societies and economies, the demand for social electricity has further increased. The efficiency and accuracy of electric-load forecasting is an important guarantee for the safety and reliability of power system operation. With the sparrow search algorithm (SSA), long short-term memory (LSTM), and random forest (RF), this research proposes an SSA-LSTM-RF daily peak-valley forecasting model. First, this research uses the Pearson correlation coefficient and the random forest model to select features. Second, the forecasting model takes the target value, climate characteristics, time series characteristics, and historical trend characteristics as input to the LSTM network to obtain the daily-load peak and valley values. Third, the super parameters of the LSTM network are optimized by the SSA algorithm and the global optimal solution is obtained. Finally, the forecasted peak and valley values are input into the random forest as features to obtain the output of the peak-valley time. The forest value of the SSA-LSTM-RF model is good, and the fitting ability is also good. Through experimental comparison, it can be seen that the electric-load forecasting algorithm based on the SSA-LSTM-RF model has higher forecasting accuracy and provides ideal performance for electric-load forecasting with different time steps.

POSS-CNN: An Automatically Generated Convolutional Neural Network with Precision and Operation Separable Structure Aiming at Target Recognition and Detection

Artificial intelligence is changing and influencing our world. As one of the main algorithms in the field of artificial intelligence, convolutional neural networks (CNNs) have developed rapidly in recent years. Especially after the emergence of NASNet, CNNs have gradually pushed the idea of AutoML to the public’s attention, and large numbers of new structures designed by automatic searches are appearing. These networks are usually based on reinforcement learning and evolutionary learning algorithms. However, sometimes, the blocks of these networks are complex, and there is no small model for simpler tasks. Therefore, this paper proposes POSS-CNN aiming at target recognition and detection, which employs a multi-branch CNN structure with PSNC and a method of automatic parallel selection for super parameters based on a multi-branch CNN structure. Moreover, POSS-CNN can be broken up. By choosing a single branch or the combination of two branches as the “benchmark”, as well as the overall POSS-CNN, we can achieve seven models with different precision and operations. The test accuracy of POSS-CNN for a recognition task tested on a CIFAR10 dataset can reach 86.4%, which is equivalent to AlexNet and VggNet, but the operation and parameters of the whole model in this paper are 45.9% and 45.8% of AlexNet, and 29.5% and 29.4% of VggNet. The mAP of POSS-CNN for a detection task tested on the LSVH dataset is 45.8, inferior to the 62.3 of YOLOv3. However, compared with YOLOv3, the operation and parameters of the model in this paper are reduced by 57.4% and 15.6%, respectively. After being accelerated by WRA, POSS-CNN for a detection task tested on an LSVH dataset can achieve 27 fps, and the energy efficiency is 0.42 J/f, which is 5 times and 96.6 times better than GPU 2080Ti in performance and energy efficiency, respectively.

The Optimization of Face Detection Technology Based on Neural Network and Deep Learning

Face detection is a biometric technology that automatically contains facial feature information. It integrates digital image processing, pattern recognition, and other technologies and collects images or video streams containing human faces by cameras or cameras for automatic detection and tracking. Starting from the idea of local features and deep learning, aiming at the problem that traditional convolutional neural network (CNN) only extracts features from the whole image and ignores practical local details, this article proposes a deep CNN model based on the fusion of global and local features. It explores the face detection algorithm with better performance under the interference of illumination, expression, and other internal or external factors. This method designs a suitable network structure according to the size of the training data set, and the core technology is the debugging of super parameters. The simulation results show that compared with SVM, the improved CNN has obvious advantages in the later stage of operation, and the error is reduced by 36.85%. Compared with the traditional face detection method, it can automatically extract image features and also automatically learn its model and get a higher recognition rate.

A Comparative Study of Motion Feature Recognition under Different Learning Methods

Abstract The accurate analysis and prediction of human motion posture can provide effective data support for sports training. With the continuous progress of computer science and technology and the gradual maturity of artificial intelligence technology, there is still some controversy about the recognition effect of motion features under different algorithms. Therefore, this paper compares the recognition effects of different learning methods, including two machine learning methods by using k-means classification, such as SVM and a deep learning method CNN. The motion data of three basic movements, such as running, jumping and turning, are collected and their feature information is extracted by sensor equipment. The recognition effect is compared with the overall accuracy rate as the evaluation index. Among them, when k-means classification is used, through feature extraction and selection, the final combination of mean and variance features is obtained and applied to SVM. When CNN is adopted, its performance can be optimised by adjusting its network structure and some super parameters in the convolution layer and pool layer.

Research on Fire Detection in Laboratories Based on CNN and Transfer Learning

To solve the problem that model may not be convergent owing to datasets with small sample sizes results from the secrecy of labs, a fire detection method for lab fire detection based on CNN (Convolutional Neural Network) and transfer leaning is proposed. Experiments of transfer learning were conducted with configuration of super parameters and a small lab fire dataset, whose pre-trained models are GoogLeNet, ResNet and MobileNet trained with ImageNet dataset. We analyze and assess the loss function curve and the performances of the above models. Result shows that MoblieNet model is chosen to be the optimum model for fire detection, providing a reference to lab fire detection with small sample size datasets. Different from artificial fire surveillance, the proposed method based on CNN and transfer learning realize real-time, autonomy and efficient fire detection.

Benign-malignant classification of pulmonary nodules by low-dose spiral computerized tomography and clinical data with machine learning in opportunistic screening.

Many people were found with pulmonary nodules during physical examinations. It is of great practical significance to discriminate benign and malignant nodules by using data mining technology. The subjects' demographic data, baseline examination results, and annual follow-up low-dose spiral computerized tomography (LDCT) results were recorded. The findings from annual physical examinations of positive nodules, including highly suspicious nodules and clinically tentative benign nodules, was analyzed. The extreme gradient boosting (XGBoost) model was constructed and the Grid Search CV method was used to select the super parameters. External unit data were used as an external validation set to evaluate the generalization performance of the model. A total of 135,503 physical examinees were enrolled. Baseline testing found that 27,636 (20.40%) participants had clinically tentative benign nodules and 611 (0.45%) participants had highly suspicious nodules. The proportion of highly suspicious nodules in participants with negative baseline was about 0.12%-0.46%, which was lower than the baseline level except the follow-up of >5 years. In the 27,636 participants with clinically tentative benign nodules, only in the first year of LDCT re-examination was the proportion of highly suspicious nodules (1.40%) significantly greater than that of baseline screening (0.45%) (p < 0.001), and the proportion of highly suspicious nodules was not different between the baseline screening and other follow-up years (p > 0.05). Furthermore, 322 cases with benign nodules and 196 patients with malignant nodules confirmed by surgery and pathology were compared. A model and the top 15 most important clinical variables were determined by XGBoost algorithm. The area under the curve (AUC) of the model was 0.76 [95% CI: 0.67-0.84], and the accuracy was 0.75. The sensitivity and specificity of the model under this threshold were 0.78 and 0.73, respectively. In the validation of model using external data, the AUC was 0.87 and the accuracy was 0.80. The sensitivity and specificity were 0.83 and 0.77, respectively. It is important that pulmonary nodules could be more accurately identified at the first LDCT examination. A model with 15 variables which are routinely measured in the clinic could be helpful to distinguish benign and malignant nodules. It could help the radiological team issue a more accurate report; and it may guide the clinical team regarding LDCT follow-up.

DE_PSO_SVM: An Alternative Wine Classification Method Based on Machine Learning

Accurate classification of wine quality may help to improve making technology of wine. For achieving more effective quality classification, a classification method named DE_PSO_SVM (dataset enhancement (DE)_particle swarm optimization (PSO)_support vector machine (SVM)) is proposed. The correlation between feature attributes and classification labels of wine samples were analyzed to achieve dimension reduction. DE was achieved by calculating the different weight sums of adjacent odd and even rows, both of which belong to the same class of samples. PSO was used to search for the optimal parameters of a Gaussian kernel function, which were substituted in the SVM model to classify wine. K-nearest-neighbor (KNN), random forest (RF) and classification and regression tree (CART) were also used to test the wine classification. In 7-fold cross-validation on three wine datasets, the average Precision, Recall, and F1score of DE_PSO_SVM were best. The results show that enhancing datasets with small samples and searching for the optimal super parameters by PSO improved the performance of the wine classification model.

Optimized deep learning based single-phase broken fault type identification for active distribution networks

Single-phase broken faults occur frequently, affecting the reliability of distribution network. In order to effectively identify the fault type of single-phase broken fault, this paper proposes a new identification method, which is based on the combination of variational mode decomposition and stacked auto encoder with double optimization (AO-VMD-PSO-SAE). Firstly, the zero sequence voltage, which collected in line, is decomposed into a set of variational modal components. Nextly, the stack automatic encoder is used to conduct unsupervised training on the denoised data to establish a depth learning model, and the AO optimization algorithm and the PSO optimization algorithm are used to determine the super parameters in the model.​ Finally, simulation results supported and the validity of the method was verified. What the results show is that the proposed model named AO-VMD-PSO-SAE can accurately predict the types of single-phase broken fault under noise interference.

Prediction of pulp exposure risk of carious pulpitis based on deep learning.

This study aims to predict the risk of deep caries exposure in radiographic images based on the convolutional neural network model, compare the prediction results of the network model with those of senior dentists, evaluate the performance of the model for teaching and training stomatological students and young dentists, and assist dentists to clarify treatment plans and conduct good doctor-patient communication before surgery. A total of 206 cases of pulpitis caused by deep caries were selected from the Department of Stomatological Hospital of Tianjin Medical University from 2019 to 2022. According to the inclusion and exclusion criteria, 104 cases of pulpitis were exposed during the decaying preparation period and 102 cases of pulpitis were not exposed. The 206 radiographic images collected were randomly divided into three groups according to the proportion: 126 radiographic images in the training set, 40 radiographic images in the validation set, and 40 radiographic images in the test set. Three convolutional neural networks, visual geometry group network (VGG), residual network (ResNet), and dense convolutional network (DenseNet) were selected to analyze the rules of the radiographic images in the training set. The radiographic images of the validation set were used to adjust the super parameters of the network. Finally, 40 radiographic images of the test set were used to evaluate the performance of the three network models. A senior dentist specializing in dental pulp was selected to predict whether the deep caries of 40 radiographic images in the test set were exposed. The gold standard is whether the pulp is exposed after decaying the prepared hole during the clinical operation. The prediction effect of the three network models (VGG, ResNet, and DenseNet) and the senior dentist on the pulp exposure of 40 radiographic images in the test set were compared using receiver operating characteristic (ROC) curve, area under the ROC curve (AUC), accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and F1 score to select the best network model. The best network model was DenseNet model, with AUC of 0.97. The AUC values of the ResNet model, VGG model, and the senior dentist were 0.89, 0.78, and 0.87, respectively. Accuracy was not statistically different between the senior dentist (0.850) and the DenseNet model (0.850)(P>0.05). Kappa consistency test showed moderate reliability (Kappa=0.6>0.4, P<0.05). Among the three convolutional neural network models, the DenseNet model has the best predictive effect on whether deep caries are exposed in imaging. The predictive effect of this model is equivalent to the level of senior dentists specializing in dental pulp.

Performance Analysis of Parallel Computing in Image Classification

The use of parallel computing can speed up the training of deep learning models. The traditional neural network model ResNet is chosen for testing in this paper on the effectiveness of data parallelism in image classification, and test data is provided in a 6-GPU environment. In this paper, it is suggested that various factors should be considered when building affordable hardware configurations to expedite model training in practical application scenarios. The communication costs are not insignificant because today's large computing clusters are primarily offered via cloud computing. Another crucial point to remember is that the number of CUDA cores is the primary hardware foundation for GPU acceleration technology. Therefore, acceleration may not be affected by more incredible video memory or fewer CUDA cores for some particular graphics cards. In addition, the issue of beyond-model performance in parallel computing is another issue that must be disregarded. Due to the parallel strategy's limitations, it is essential to tweak the super parameters while speeding up model training. The model's performance is more likely to be ensured by the lower learning rate and Batch size. This paper's experimental conclusion can support building an appropriate hardware configuration scheme.

Nature-Inspired DBN based Optimization Techniques for Image De-noising

Whale optimization algorithm (WOA) and particle swarm optimization (PSO) are heuristic techniques used to solve various engineering optimization problems. In this paper, these algorithms have been used in combination with a relatively less explored deep-learning model, viz., deep belief network (DBN) for Gaussian de-noising. DBNs are stacked restricted Boltzmann machines (RBMs) whose typical architectural characteristics make deep learning feasible by reducing the training complexity. The de-noising results of images corrupted by additive white Gaussian noise (AWGN) using three proposed networks; MWOA-DBN, WOA-DBN, and PSO-DBN are provided. Super parameters (step ratio and dropout rate) are optimized using MWOA, WOA, and PSO with root mean square error as the fitness function to circumvent over-fitting. The nature of convergence of the fitness function is tested for variation in step ratio, and dropout rate. The performance of the de-noising method is tested on bench-mark metrics like peak signal-to-noise ratio (PSNR), structural similarity (SSIM), and root-mean-square error (RMSE). It is observed that the performance of the proposed methods outperforms the state-of-the-art image de-noising techniques.

Stacking strategy-assisted random forest algorithm and its application

Short-term power load forecasting provides important guidance for the improvement of power marketing and control levels of power enterprises. In this paper, a novel method, named RF-TStacking, is proposed to forecast the short-term load. This study starts from the influence factors of the power load, the random forest is applied to estimate the importance of the influence factors of short-term load. Based on Stacking strategy, the integration of LightGBM and random forest is realized to achieve short-term power load forecasting. To improve the generalization ability of the load model, random put back sampling is used to sample each primary learner, and the average value is taken as the result of each primary learner. The Bayesian optimization is used to adjust the super parameters of the model to improve the accuracy of the selection of influencing factors. The load data of a region in northwest China are used for the testing, and it is found that the model can provide stable prediction results.

Ship motion attitude prediction model based on IWOA-TCN-Attention

Aiming at the problem of low prediction accuracy of ship motion with the characteristics of non-stationary, nonlinear and stochastic, this paper proposes a combined prediction model of ship motion based on improved whale optimization algorithm, temporal convolutional networks and attention mechanism (IWOA-TCN-Attention). One research direction of ship motion prediction is to use time series for regression prediction. The TCN model is used in the research of time series prediction because of its long-term memory of time series. IWOA-TCN-Attention model firstly establishes a deep learning network based on TCN model, uses this network to obtain the original features of ship motion attitude time series data, and then gives different weight values to different original features through attention mechanism, so as to highlight important features and reduce the impact of secondary features on prediction results. At the same time, in order to further optimize the prediction accuracy of TCN-Attention model, this paper proposes an IWOA optimization algorithm by improving the WOA optimization algorithm. This algorithm is used to optimize the super parameters of TCN-Attention model, and finally obtain the optimal prediction effect. The simulation experiment shows that IWOA-TCN-Attention model can effectively realize the prediction of ship motion, reduce the error of motion prediction and improve the prediction accuracy.

A Category-Contrastive Guided-Graph Convolutional Network Approach for the Semantic Segmentation of Point Clouds

The semantic segmentation of light detection and ranging (LiDAR) point clouds plays an important role in 3D scene intelligent perception and semantic modeling. The unstructured, sparse and uneven characteristics of point clouds pose great challenges to the representation of the local geometric shapes, which degrades semantic segmentation performance. To address the challenges of describing local geometric shapes due to unstructured and sparse 3D point clouds, this paper proposes a category-contrastive-guided graph convolutional network (CGGC-Net) for the semantic segmentation of LiDAR point clouds. First, a detailed geometric structure of the raw point clouds is encoded to represent the inherent geometric pattern within the local neighborhood. At the same time, the geometric structures information is transmitted across multiple layers, so that the geometric structure encoding information containing different receptive fields and richer neighborhood spatial structure can be aggregated. Following this, the graph convolution neural network uses the edge convolution layer to adaptively describe the semantic correlation between the query point and its neighboring points, and combines the attention mechanism to gather the surrounding feature information to the query point. As a result, the graph convolution neural network and attention mechanism are iteratively stacked for the aggregation and fusion of spatial context semantic information, to generate highly discriminative semantic feature representation. Finally, the super parameters of the model are learned through a multitask optimization strategy guided by category-aware contrastive loss and cross-entropy loss. Experiments are conducted on the public SemanticKITTI dataset and the Stanford Large-Scale 3D Indoor Spaces dataset to demonstrate the effectiveness and reliability of the proposed CGGC-Net from both quantitative and qualitative perspectives. The results indicate its capability of automatically classifying LiDAR point clouds, with a mean intersection-over-union (mIoU) of 58.4%. Moreover, multiple comparative experiments also demonstrate the superior performance of the proposed method, exceeding state-of-the-art methods.

The Influence of Greenhouse Effect on Earth’s Atmosphere Based on Artificial Intelligence

Greenhouse gases absorb and release a large amount of ground thermal radiation, and modify part of the energy between the earth and the atmosphere, so that it has a greenhouse effect. It is the main cause of global warming and climate anomaly caused by modern human activities. In order to solve the shortcomings of the existing research on the impact of greenhouse effect on the Earth’s atmosphere, this paper briefly introduces the numerical methods of radiation and heat conduction coupling, the components of the Earth’s atmosphere, and the sources and sinks of greenhouse gases. The calculation and analysis of spectral absorption coefficient and the configuration of super parameters of depth learning model are discussed. A numerical simulation of carbon dioxide concentration based on depth learning is constructed and the C02 concentration profile is calculated using the C02 statistical inversion flow of principal component analysis. The calculated temperatures of the standard atmosphere and the standard atmosphere with the carbon dioxide concentration doubled were compared. The data showed that the temperature in the upper stratosphere rose to 58.

Prediction and risk assessment of sepsis-associated encephalopathy in ICU based on interpretable machine learning

Sepsis-associated encephalopathy (SAE) is a major complication of sepsis and is associated with high mortality and poor long-term prognosis. The purpose of this study is to develop interpretable machine learning models to predict the occurrence of SAE after ICU admission and implement the individual prediction and analysis. Patients with sepsis admitted to ICU were included. SAE was diagnosed as glasgow coma score (GCS) less than 15. Statistical analysis at baseline was performed between SAE and non-SAE. Six machine learning classifiers were employed to predict the occurrence of SAE, and the adjustment of model super parameters was performed by using Bayesian optimization method. Finally, the optimal algorithm was selected according to the prediction efficiency. In addition, professional physicians were invited to evaluate our model prediction results for further quantitative assessment of the model interpretability. The preliminary analysis of variance showed significant differences in the incidence of SAE among patients with pathogen infection. There were significant differences in physical indicators like respiratory rate, temperature, SpO2 and mean arterial pressure (P < 0.001). In addition, the laboratory results were also significantly different. The optimal classification model (XGBoost) indicated that the best risk factors (cut-off points) were creatinine (1.1 mg/dl), mean respiratory rate (18), pH (7.38), age (72), chlorine (101 mmol/L), sodium (138.5 k/ul), SAPSII score (23), platelet count (160), and phosphorus (2.4 and 5.0 mg/dL). The ranked features derived from the best model (AUC is 0.8837) were mechanical ventilation, duration of mechanical ventilation, phosphorus, SOFA score, and vasopressin usage. The SAE risk prediction model based on XGBoost created here can make very accurate predictions using simple indicators and support the visual explanation. The interpretable model was effectively evaluated by professional physicians and can help them predict the occurrence of SAE more intuitively.

Prediction of Carbon Emission Right Price Based on XGBoost Algorithm

Reasonable carbon market price prediction can facilitate the carbon market participants, such as physical producers, to achieve the goal of efficient emission reduction through the market mechanism. In this paper, we use XGBoost, an integrated algorithm in machine learning, to forecast the domestic carbon price from 2013 to 2021. Pearson coefficient is utilized to calculate the correlation of data features, and perform PCA dimensionality reduction on the features with high correlation coefficients. Before using PCA to reduce the dimensions, in order to make the feature data more suitable for later model training, the feature data is standardized. After the standardization, PCA selects super parameters with maximum likelihood estimation and outputs the dimension reduction results of features. Finally, the integrated algorithm XGBoost is used to form a prediction model for the carbon price. RMSE output from cross validation are used to evaluate the accuracy and error of the prediction results. The results show that this paper confirms that the integrated algorithm XGBoost model has a good prediction ability for carbon price, and provides a novel idea for the field of carbon emission price prediction. It is expected to provide some rational basis for the carbon market participants to make investment decisions, so as to avoid the carbon market risk caused by the change of carbon price.

Underwater Image Classification Algorithm Based on Convolutional Neural Network and Optimized Extreme Learning Machine

In order to deal with the target recognition in the complex underwater environment, we carried out experimental research. This includes filtering noise in the feature extraction stage of underwater images rich in noise, or with complex backgrounds, and improving the accuracy of target classification in the recognition process. This paper discusses our contribution to improving the accuracy of underwater target classification. This paper proposes an underwater target classification algorithm based on the improved flow direction algorithm (FDA) and search agent strategy, which can simultaneously optimize the weight parameters, bias parameters, and super parameters of the extreme learning machine (ELM). As a new underwater target classifier, it replaces the full connection layer in the traditional classification network to build a classification network. In the first stage of the network, the DenseNet201 network pre-trained by ImageNet is used to extract features and reduce dimensions of underwater images. In the second stage, the optimized ELM classifier is trained and predicted. In order to weaken the uncertainty caused by the random input weight and offset of the introduced ELM, the fuzzy logic, chaos initialization, and multi population strategy-based flow direction algorithm (FCMFDA) is used to adjust the input weight and offset of the ELM and optimize the super parameters with the search agent strategy at the same time. We tested and verified the FCMFDA-ELM classifier on Fish4Knowledge and underwater robot professional competition 2018 (URPC 2018) datasets, and achieved 99.4% and 97.5% accuracy, respectively. The experimental analysis shows that the FCMFDA-ELM underwater image classifier proposed in this paper has a greater improvement in classification accuracy, stronger stability, and faster convergence. Finally, it can be embedded in the recognition process of underwater targets to improve the recognition performance and efficiency.