Max-pooling Layers Research Articles

Ensemble fusion model for improved lung abnormality classification: Leveraging pre-trained models

Lung abnormalities pose significant health concerns, underscoring the need for swift and accurate diagnoses to facilitate timely medical intervention. This study introduces a novel methodology for the sub-classification of lung abnormalities within chest X-rays captured via smartphones. An accurate and timely diagnosis of lung abnormalities is essential for the successful implementation of appropriate therapy. In this paper, we propose a novel approach using a Convolutional neural network (CNN) with three maximum pooling layers and early fusion for sub-classifying lung abnormalities from chest Xrays. Based on the kind of abnormality, the CheXpert dataset is divided into 13 sub-classes, each of which is trained using a different sub-model. An early fusion procedure is then used to integrate the outputs of the sub-model.•3M-CNN (Method 1): We employed a Convolutional Neural Network (CNN) with three max pooling layers and an early fusion strategy to train dedicated sub-models for each of the 13 distinct sub-classes of lung abnormalities using the CheXpert dataset.•Ensemble Model (Method 2): Our ‘Ensemble model’ integrated the outputs of the trained sub-models, providing a powerful approach for the sub-classification of lung abnormalities.•Exceptional Accuracy: Our ‘3M-CNN’ and ‘fused model’ achieved an accuracy of 98.79%, surpassing established methodologies, which is beneficial in resource-constrained environments embracing smartphone-based imaging.

Analyzing and Improving Existing Neural Network-Based Approaches to Identify AI Generated Images

Images produced by AI software, particularly human faces, have the potential to spread misinformation, thus detection tools are essential to tell facts from fiction. However, current neural network-based tools misclassify images with extreme photography conditions, facial obstructions, or post-processing. This project aimed to address the limitations of existing detection models, especially regarding poorly-photographed or edited images. A MobileNet-v2-based neural network to identify synthetic images was analyzed for possible shortcomings. After obtaining its baseline accuracy, it was stress tested on validation data with brightness, contrast, hue, and saturation edited by set increments. The model performed poorly on processed images and faces with glasses or shadows. Possible limitations were insufficient data augmentation, the Global Average Pooling Layer, too few epochs trained, and inadequate regularization strength or dropout rate. The model was then modified to address these shortcomings by altering the brightness, contrast, hue, and saturation of training data, adding Gaussian noise to training images, implementing a Global Max Pooling Layer, augmenting the number of epochs trained, and changing the regularization strength and dropout rate. The modified model was evaluated on augmented and non-augmented images. Overall, the model improved when Gaussian noise was added with standard deviation 0.1, it was trained for 15 epochs, and there were no dropout layers. This was because the Gaussian noise layer approximated degraded image quality, and the model was too simple to learn image features when trained for fewer epochs or when neurons were disabled. This research could be expanded by testing multiple modifications at once.

Affine medical image registration with fusion feature mapping in local and global.

Objective. Medical image affine registration is a crucial basis before using deformable registration. On the one hand, the traditional affine registration methods based on step-by-step optimization are very time-consuming, so these methods are not compatible with most real-time medical applications. On the other hand, convolutional neural networks are limited in modeling long-range spatial relationships of the features due to inductive biases, such as weight sharing and locality. This is not conducive to affine registration tasks. Therefore, the evolution of real-time and high-accuracy affine medical image registration algorithms is necessary for registration applications.Approach. In this paper, we propose a deep learning-based coarse-to-fine global and local feature fusion architecture for fast affine registration, and we use an unsupervised approach for end-to-end training. We use multiscale convolutional kernels as our elemental convolutional blocks to enhance feature extraction. Then, to learn the long-range spatial relationships of the features, we propose a new affine registration framework with weighted global positional attention that fuses global feature mapping and local feature mapping. Moreover, the fusion regressor is designed to generate the affine parameters.Main results. The additive fusion method can be adaptive to global mapping and local mapping, which improves affine registration accuracy without the center of mass initialization. In addition, the max pooling layer and the multiscale convolutional kernel coding module increase the ability of the model in affine registration.Significance. We validate the effectiveness of our method on the OASIS dataset with 414 3D MRI brain maps. Comprehensive results demonstrate that our method achieves state-of-the-art affine registration accuracy and very efficient runtimes.

Advanced Deep Learning Model for Predicting the Academic Performances of Students in Educational Institutions

Educational institutions are increasingly focused on supporting students who may be facing academic challenges, aiming to enhance their educational outcomes through targeted interventions. Within this framework, leveraging advanced deep learning techniques to develop recommendation systems becomes essential. These systems are designed to identify students at risk of underperforming by analyzing patterns in their historical academic data, thereby facilitating personalized support strategies. This research introduces an innovative deep learning model tailored for pinpointing students in need of academic assistance. Utilizing a Gated Recurrent Neural Network (GRU) architecture, the model is rich with features such as a dense layer, max-pooling layer, and the ADAM optimization method used to optimize performance. The effectiveness of this model was tested using a comprehensive dataset containing 15,165 records of student assessments collected across several academic institutions. A comparative analysis with existing educational recommendation models, like Recurrent Neural Network (RNN), AdaBoost, and Artificial Immune Recognition System v2, highlights the superior accuracy of the proposed GRU model, which achieved an impressive overall accuracy of 99.70%. This breakthrough underscores the model’s potential in aiding educational institutions to proactively support students, thereby mitigating the risks of underachievement and dropout.

Maximum output discrepancy computation for convolutional neural network compression

Network compression methods minimize the number of network parameters and computation costs while maintaining desired network performance. However, the safety assurance of many compression methods is based on a large amount of experimental data, whereas unforeseen incidents beyond the experiment data may result in unsafe consequences. In this work, we developed a discrepancy computation method for two convolutional neural networks by giving a concrete value to characterize the maximum output difference between the two networks after compression. Using Imagestar-based reachability analysis, we propose a novel method to merge the two networks to compute the difference. We illustrate reachability computation for each layer in the merged network, such as the convolution, max pooling, fully connected, and ReLU layers. We apply our method to a numerical example to prove its correctness. Furthermore, we implement our developed methods on the VGG16 model with the Quantization Aware Training (QAT) compression method; the results show that our approach can efficiently compute the accurate maximum output discrepancy between the original neural network and the compressed neural network.

Diverter transformer-based multi-encoder-multi-decoder network model for medical retinal blood vessel image segmentation

The retinal blood vessel is an essential part of the fundus structure. It is important to accurately analyze the structure and distribution of retinal vessels, which can help make accurate medical diagnoses. However, it is still challenging to extract detailed information due to the problems of fuzzy edges, low resolution, and lots of noise in retinal blood vessel medical images. To extract the image detail information effectively, we propose a new diverter transformer-based multi-encoder-multi-decoder network model in this paper. The network model consists of a feature encoder module and a feature decoder module. Among them, the feature encoding module consists of a diverter transformer with a diverter adaptive mechanism, three encoder units with a convolution layer and max-pooling layer, and the two decoder units in the feature decoding module consist of an inverse convolution layer and an up-sampling layer, respectively. The Local Context Module (LCNet Module) in the feature encoding module learns richer local context feature information layer by layer through changing the width of the network while downsampling; the Global Encoder Module1 (G-Encoder Module1) and the Global Encoder Module2 (G-Encoder Module2) extract the global feature representation of retinal blood vessel images by performing a max-pooling operation to transform the input data into a vector of fixed dimensions, thus helping the network model to better understand and extract the global feature representation of retinal blood vessel images. The two decoder units in the feature decoding module receive local and global feature information from three encoder units, LCNet Module, G-Encoder Module1 and G-Encoder Module2, respectively. Decoder Module1 generates segmentation prediction by layer-by-layer up-sampling operation, and Decoder Module2 recovers the feature information by downsampling and decoding operations and fuses the recovered feature information to output, obtaining the final segmentation of the retinal blood vessels. The proposed diverter transformer-based multi-encoder-multi-decoder network model is validated on the DRIVE and STARE datasets with other classical and state-of-the-art network models, and its segmentation accuracy is 97.25% and 97.93%, respectively. Compared with the classical U-Net model, the improvement is 2.24% and 1.42%, respectively. Compared with the state-of-the-art SPNet model, the accuracy is increased by 0.61% on DRIVE and 1.01% on STARE. It indicates that the network model proposed in this paper has a significant competitive advantage in improving the segmentation performance of retinal blood vessel images.

Diagnosis of Adult ADHD Using EEG Signals Based on the Spectrogram and Convolutional Neural Networks

Attention deficit hyperactivity disorder (ADHD) is one of the most common mental disorders. This disease includes a combination of disorders in maintaining attention, hyperactivity, and impulsive behaviors. Diagnosis of ADHD is primarily clinical and based on history and examination. This study aims to provide a method for a more accurate diagnosis of adult ADHD using electroencephalography (EEG) signals. EEG signals recorded from 37 ADHD and 42 healthy adults were used as a control group in the age range of 20–68 years. We designed a convolutional neural network with three convolutional layers, three max-pooling layers, and one fully connected layer and trained it using spectrogram images obtained from EEG signals. The Cz channel was used for the diagnosis ADHD in four different states. To evaluate the performance of the proposed method, metrics such as accuracy, sensitivity, specificity, and precision were calculated. The results showed that using only one Cz channel has a good performance in diagnosing of ADHD. The highest accuracy of classification was related to the classification of two groups in the state when their eyes-open and eyes-closed spectrogram images were subtracted from each other. The results showed that proposed method based on deep learning can be a suitable method for diagnosing ADHD.

Recurrent Convolution Neural Network for Space Debris and Asteroids Detection and Classification on the Monitored Satellite Information Using Deep Space Network

Satellites are used for many monitoring, prediction and forecasting application operation in space , hence it has to be continuously monitored and controlled against the catastrophic damage due to collision of space debris and asteroids debris in the space. Timely and efficient Controlling of satellite will eliminate the heavy irreparable damages as number of space debris and asteroids debris is continuously increasing in the space. Hence monitoring of the satellites increases the sustainability and its safety navigation in the space operations. In order to achieve this objective, deep space network (Satellite ground station ) has to be designed as space situational awareness mechanism on basis of the detection and classification of the space objects towards providing optimal path for satellite trajectory with high processing speed and reduction of memory in data processing. In this work, a new deep learning architecture entitled as optimized recurrent convolution neural network is designed to satellite ground station towards detection and classification of space debris and asteroids debris. Initially architecture is interfaced with feature extraction technique named as principle component analysis to extract the space features and those extracted feature were detected and classified using deep learning model. In this work, convolution neural network composed of convolution layer is used to filter the discrete features, max pooling layer for downsizing the features and represent the feature map and finally fully connected layer to detect and classify the space debris and predict the optimal path for satellite trajectory on basis of the velocity of the debris in the particular orbit. Experimental results of the proposed model has been evaluated using accuracy and precision metric. Further design is implemented and verified using Xilinx Software. The performance analysis of the proposed model yields better accuracy on the classifying the monitored information on comparing with conventional approaches such as support vector machine and K- Nearest Neighbour Technique. Keywords: Satellite Collision Analysis, Deep learning, Space Debris , Asteroid Debris, Convolution Neural Network, Principle Component analysis , Deep Space Network

Preliminary Study of Dental Caries Detection by Deep Neural Network Applying Domain-Specific Transfer Learning

PurposeThe purpose of this study is to confirm whether it is possible to acquire a certain degree of diagnostic ability even with a small dataset using domain-specific transfer learning. In this study, we constructed a simulated caries detection model on panoramic tomography using transfer learning.MethodsA simulated caries model was trained and validated using 1094 trimmed intraoral images. A convolutional neural network (CNN) with three convolution and three max pooling layers was developed. We applied this caries detection model to 50 panoramic images and evaluated its diagnostic performance.ResultsThe diagnostic performance of the CNN model on the intraoral film was as follows: C0 84.6%; C1 90.6%; C2 88.6%. Finally, we tested 50 panoramic images with simulated caries insertion. The diagnostic performance of the CNN model on the panoramic image was as follows: C0 75.0%, C1 80.0%, C2 80.0%, and overall diagnostic accuracy was 78.0%. The diagnostic performance of the caries detection model constructed only with panoramic images was much lower than that of the intraoral film.ConclusionDomain-specific transfer learning methods may be useful for saving datasets and training time (179/250).

Real-time mask-wearing detection in video streams using deep convolutional neural networks for face recognition

This research aims to develop a real-time mask-wearing detection system using deep convolutional neural networks (CNNs). This is crucial in the coronavirus disease 2019 (COVID-19) pandemic to alert individuals who are not wearing masks early on, thereby reducing the spread of the virus. Since COVID-19 primarily spreads through respiratory droplets and mask-wearing is recommended, our proposed study utilizes computer vision techniques, specifically image processing, to detect masked and unmasked faces. We employ a customized CNN architecture consisting of five convolutional layers, followed by max-pooling layers and fully connected (FC) layers. The final output layer utilizes softmax activation for classification. The model is updated with optimized layer configurations and parameter values. We are developing an application that uses a digital camera as an input device. The application utilizes a dataset comprising 11,792 image samples, which are used for training and testing purposes with the 80:20 ratio. Real-time testing is conducted using human subjects captured by the camera. The experimental results demonstrate that the CNN method achieves a classification accuracy of 99% on the training data and 98.83% during real-time video testing. These findings suggest that the real-time mask detection system using CNN performs effectively.

MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model

Underwater light attenuation leads to decreased image contrast. This reduction in contrast subsequently decreases target visibility. Additionally, marine target detection is challenging due to multi-scale problems from varying target-to-device distances, complex target clustering, and noise from waterborne particulates.To address these issues, we propose MTD-YOLOv5.Initially, we enhance image contrast with grayscale equalization and mitigate color shift issues through color space transformation.We then introduce a novel feature extraction module, PCBR, combining max pooling and convolution layers for more effective target feature extraction from the background.Furthermore, we present the Multi-Scale Perceptual Hybrid Pooling (MHP) module.This module integrates horizontal and vertical receptive fields to establish long-range dependencies, thereby capturing hidden target information in deep network feature maps. In the Labeled Fishes in the Wild test datasets, MTD-YOLOv5 achieves a precision of 88.1% and a mean Average Precision (mAP[0.5:.95]) of 49.6%.These results represent improvements of 2.6% in precision and 0.4% in mAP over the original YOLOv5.

Abstract B038: A preliminary study on predicting pathological complete response to neoadjuvant chemotherapy in triple-negative breast cancer

Abstract Purpose In this preliminary study, we are aiming to develop a deep learning model to predict pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) for triple-negative breast cancer (TNBC) patient based on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). Background Neoadjuvant chemotherapy (NAC) has become a grown treatment method for patients with triple-negative breast cancer (TNBC). However, not every patient can achieve pathological complete response (pCR) to NAC. NAC also has side effects such as nausea, hair loss, appetite loss, mouth sores, and nerve damage. Additionally, delaying surgery to remove the tumor can allow it to spread, making treatment more difficult. Therefore, it is necessary to predict pCR to NAC reliably in a non-invasive way. Deep learning methods, especially convolutional neural network (CNN), have been shown great success in image-based treatment response prediction. Methods In this preliminary study, totally 78 TNBC patients on DCE-MRI collected from public dataset (Duke University) were used, among which are 23 patients with pCR and 55 patients without pCR. For each patient, the median slice from the tumor area of the fat-saturated gradient echo T1-weighted pre-contrast sequence MRI was extracted. A convolutional neural network model was built to predict whether the patients receive pCR or not after NAC. In the constructed prediction model, three convolutional layers were used to capture the features from DCE-MRI. Each convolutional layer was followed by a max-pooling layer to reduce the dimensions of the feature maps. Two fully connected layers were added into the model so that the model has a binary outcome which provides the prediction of pCR. Additionally, “ReLU” was used as the activation function in all the layers except for the output layer which used “softmax” instead. By using augmentation technique, we solved the problems of small sample size and unbalance data. Results In this study, 75% dataset were used for training and the remaining 25% were used for testing. The model we developed has a 75% accuracy, with a sensitivity of 0.67 and a specificity of 0.79. These results indicate that the model performs well in predicting patients who do not achieve pathological complete response (pCR). Additionally, the area under the receiver operating characteristic (AUC) curve is calculated as 0.74, indicating the outstanding performance. Citation Format: Xi Chen, Zhiguo Zhou. A preliminary study on predicting pathological complete response to neoadjuvant chemotherapy in triple-negative breast cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B038.

Perimeter monitoring of urban buried pipeline threated by construction activities based on distributed fiber optic sensing and real-time object detection.

Urban construction activities seriously jeopardize the security of buried pipeline. Distributed optical fiber vibration monitoring is one of the most promising ways to prevent third-party threats, of which the biggest challenge is to quickly and accurately detect rare abnormal events from extremely large amounts of time-space raw data. By analogy with image recognition, the task here is similar to object detection if considering the time-space optical signals as the grayscale images and the abnormal events as the objects. Given this, what we believe to be a novel monitoring method is proposed, which consists of two Faster R-CNN models, a max pooling layer and a monitoring strategy. In the field tests, the 86-hour optical vibration signals for 5.25 km distance are recognized within 6.6 minutes with the recognition rate of 98.85% for construction activities, and only two false alarms are issued. The proposed method can reduce the recognition time by 99.59% compared to the CNN-based method.

Efficient deep learning architecture for the classification of diseased plant leaves

<span>The classification of plant leaf diseases via machine learning and deep learning algorithms has a great deal of potential for enhancing agricultural operations by allowing the early and accurate diagnosis of diseases. These systems can potentially develop into useful instruments for environmentally responsible farming and increased food safety as technological advancements continue. In this work, an efficient deep learning architecture has been developed to classify the diseased plant leaves. A ten-layer architecture is designed, which includes 5-convolutional layers using different numbers of filters (32, 64, 128, 256, and 512) and for dimension reduction, five max-pooling layers are used. The PlantVillage dataset which consists of more than 50,000 plant leaf samples is used to analyze the proposed architecture's performance. The performances are evaluated across different training and testing configurations and different dropout configurations. When compared to well-known transfer learning methods using visual geometric group (VGG16), AlexNet, and GoogleNet architectures, the proposed architecture obtains a higher level of performance with 98.18% classification accuracy.</span>

STransU2Net: Transformer based hybrid model for building segmentation in detailed satellite imagery.

As essential components of human society, buildings serve a multitude of functions and significance. Convolutional Neural Network (CNN) has made remarkable progress in the task of building extraction from detailed satellite imagery, owing to the potent capability to capture local information. However, CNN performs suboptimal in extracting larger buildings. Conversely, Transformer has excelled in capturing global information through self-attention mechanisms but are less effective in capturing local information compared to CNN, resulting in suboptimal performance in extracting smaller buildings. Therefore, we have designed the hybrid model STransU2Net, which combines meticulously designed Transformer and CNN to extract buildings of various sizes. In detail, we designed a Bottleneck Pooling Block (BPB) to replace the conventional Max Pooling layer during the downsampling phase, aiming to enhance the extraction of edge information. Furthermore, we devised the Channel And Spatial Attention Block (CSAB) to enhance the target location information during the encoding and decoding stages. Additionally, we added a Swin Transformer Block (STB) at the skip connection location to enhance the model's global modeling ability. Finally, we empirically assessed the performance of STransU2Net on both the Aerial imagery and Satellite II datasets, The IoU achieved state-of-the-art results with 91.04% and 59.09%, respectively, outperforming other models.

Optimized CNN‐Bi‐LSTM–Based BCI System for Imagined Speech Recognition Using FOA‐DWT

Speech imagery is emerging as a significant neuro‐paradigm for designing an electroencephalography (EEG)‐based brain–computer interface (BCI) system for the purpose of rehabilitation, medical neurology, and to aid people with disabilities in interacting with their surroundings. Neural correlates of speech imagery EEG signals are variable and weak as compared to the vocal state; hence, it is challenging to interpret them using machine learning (ML)–based classifiers. The applicability of modern deep learning methods such as convolutional neural networks (CNNs) and bidirectional long short‐term memory (Bi‐LSTM) networks has seen substantial advances in complex EEG signal analysis as compared to ML‐based methods. The objective of this article is to design a firefly‐optimized discrete wavelet transform (DWT) and CNN‐Bi‐LSTM–based imagined speech recognition (ISR) system to interpret imagined speech EEG signals. This study utilizes two publicly available datasets. EEG signal is enhanced using firefly optimization algorithm (FOA)–based optimized soft thresholding of high‐frequency detail components obtained by DWT decomposition. The enhanced EEG signal is augmented using sliding window data augmentation to increase the training data. Frequency‐domain features like power spectral density (PSD), frequency band power (FBP), band ratios, peak frequency, mean frequency, median frequency, spectral entropy, and relative power are extracted from augmented EEG segments. The extracted feature vector is fed to the designed CNN‐Bi‐LSTM classifier such that the EEG data are classified into two‐class, three‐class, and four‐class categories. To achieve optimal performance, the CNN‐Bi‐LSTM model was optimized using the Keras tuner library. The designed CNN consists of one‐dimensional (1‐D) convolutional layers and max pooling layers for familiarizing local associations along with mining hierarchical connections, and the Bi‐LSTM network acquires long‐term dependencies from the features learned by the former CNN. Bi‐LSTM network improves the performance and acquires potentially more affluent representations by looking at the sequence in both forward and reverse ways to capture representations that might be left unexploited by the sequential‐order kind alone. The performance of the designed FOA‐DWT‐CNN‐Bi‐LSTM–based ISR system is assessed using four evaluation measures: accuracy, F1 score, recall, and precision. It is found that the proposed system achieves the highest classification accuracy of 99.43 ± 2.5%, 94.41 ± 3.31%, and 89.57 ± 4.3% for two‐class, three‐class, and four‐class categories, respectively.

A pooling convolution model for multi-classification of ECG and PCG signals

Electrocardiogram (ECG) and phonocardiogram (PCG) signals are physiological signals generated throughout the cardiac cycle. The application of deep learning techniques to recognize ECG and PCG signals can greatly enhance the efficiency of cardiovascular disease detection. Therefore, we propose a series of straightforward and effective pooling convolutional models for the multi-classification of ECG and PCG signals. Initially, these signals undergo preprocessing. Subsequently, we design various structural blocks, including a stacked block (MCM) comprising convolutional layer and max-pooling layers, along with its variations, as well as a residual block (REC). By adjusting the number of structural blocks, these models can handle ECG and PCG data with different sampling rates. In the final tests, the models utilizing the MCM structural block achieved accuracies of 98.70 and 92.58% on the ECG and PCG fusion datasets, respectively. These accuracies surpass those of all networks utilizing its variations. Moreover, compared to the models employing the REC structural block, the accuracies are improved by 0.02 and 4.30%, respectively. Furthermore, this research has been validated through tests conducted on multiple ECG and PCG datasets, along with comparisons to other published literature. To further validate the generalizability of the model, an additional experiment involving the classification of a synchronized ECG-PCG dataset was conducted. This dataset is divided into seven different levels of fatigue based on the amount of exercise performed by each healthy subject during the testing process. The results indicate that the model using the MCM block also achieved the highest accuracy.

Detection of ADHD Disorder in Children Using Layer-Wise Relevance Propagation and Convolutional Neural Network: An EEG Analysis

Purpose: Attention-Deficit-Hyperactivity-Disorder (ADHD) is a neurodevelopmental disorder that begins in early childhood and often persists into adulthood, causing personality issues and social behavior problems. Thus, detecting ADHD in its early stages and developing an effective therapy is of tremendous interest. This study presents a deep learning-based model for ADHD diagnosis in children. Materials and Methods: The 'First-National-EEG-Data-Analysis-Competition-with-Clinical-Application' dataset is used for this purpose. Following preprocessing, data is segmented into 3-second epochs, and frequency features are extracted from these epochs. The Fourier transform is applied to each channel separately, and the resulting two-dimensional matrix (channel×frequency) for each epoch is used as the Convolutional Neural Network's (CNN) input. The CNN is made up of two convolutional layers, two max pooling layers and two fully connected layers as well as the output layer (a total of 9 layers) for classification. To improve the method's performance, the output of the classification of each input variable is analyzed. In other words, the role of each channel/frequency in the final classification is being investigated using the Layer-wise Relevance Propagation (LRP) algorithm. Results: According to the results of the LRP algorithm, only efficient channels are employed as Convolutional Neural Network (CNN) inputs in the following stage. This method yields a final accuracy of 94.52% for validation data. In this study, the feature space is visualized, useful channels are selected, and deep structure capabilities are exploited to diagnose ADHD disorder. Conclusion: The findings suggest that the proposed technique can be used to effectively diagnose ADHD in children.

A fine-tuning deep residual convolutional neural network for emotion recognition based on frequency-channel matrices representation of one-dimensional electroencephalography

Emotion recognition (ER) plays a crucial role in enabling machines to perceive human emotional and psychological states, thus enhancing human-machine interaction. Recently, there has been a growing interest in ER based on electroencephalogram (EEG) signals. However, due to the noisy, nonlinear, and nonstationary properties of electroencephalography signals, developing an automatic and high-accuracy ER system is still a challenging task. In this study, a pretrained deep residual convolutional neural network model, including 17 convolutional layers and one fully connected layer with transfer learning technique in combination frequency-channel matrices (FCM) of two-dimensional data based on Welch power spectral density estimate from the one-dimensional EEG data has been proposed for improving the ER by automatically learning the underlying intrinsic features of multi-channel EEG data. The experiment result shows a mean accuracy of 93.61 ± 0.84%, a mean precision of 94.70 ± 0.60%, a mean sensitivity of 95.13 ± 1.02%, a mean specificity of 91.04 ± 1.02%, and a mean F1-score of 94.91 ± 0.68%, respectively using 5-fold cross-validation on the DEAP dataset. Meanwhile, to better explore and understand how the proposed model works, we noted that the ranking of clustering effect of FCM for the same category by employing the t-distributed stochastic neighbor embedding strategy is: softmax layer activation is the best, the middle convolutional layer activation is the second, and the early max pooling layer activation is the worst. These findings confirm the promising potential of combining deep learning approaches with transfer learning techniques and FCM for effective ER tasks.

Developing a Quality Control System in a Continuous Hot Air Heating Process in Surimi Seafood Processing Using Image Analysis and Artificial Intelligence

In this study, the feasibility of classifying surimi gels during a continuous heating process using an artificial intelligence (AI) algorithm on labeled images was investigated. Surimi paste with varying corn starch concentrations (0%, 5%, and 10%) and moisture content levels (78% and 80%) from Alaska pollock were analyzed for the subtle physical changes. Rheological characterization and K-means clustering analysis performed on entire images captured from different batches of heated surimi gel indicated a four-stage transformation from its initial state to gel formation with the temperature ranges spanning 25 to <40 °C, 40 to <50 °C, 50 to <55 °C, and 55 to 80 °C. Subsequently, a Convolutional Neural Network (CNN) model based on the temperature classification was designed to interpret and classify these images. A total of 1000 to 1200 images were used for the training, testing, and validation purposes in the ratio 7:1:2. The CNN architecture incorporated essential elements including an input layer, convolutional layers, rectified linear unit (ReLU) activation functions, normalization layers, and max-pooling layers. The CNN model achieved validation accuracy >92.67% for individual mixture composition, 94.53% for classifying surimi samples based on moisture content, and gelation level, and 89.73% for complex classifications involving moisture content, starch concentration, and gelation stages. Additionally, it exhibited high average precision, recall, and F1 scores (>0.92), indicating precision and sensitivity in identifying relevant instances. The success of CNN in non-destructively classifying surimi gels with different moisture and starch contents is demonstrated in this work.