A Bayesian-optimized design for an interpretable convolutional neural network to decode and analyze the P300 response in autism

Brain tumor is one of the most aggressive diseases nowadays, resulting in a very short life span if it is diagnosed at an advanced stage. The treatment planning phase is thus essential for enhancing the quality of life for patients. The use of Magnetic Resonance Imaging (MRI) in the diagnosis of brain tumors is extremely widespread, but the manual interpretation of large amounts of images requires considerable effort and is prone to human errors. Hence, an automated method is necessary to identify the most common brain tumors. Convolutional Neural Network (CNN) architectures are successful in image classification due to their high layer count, which enables them to conceive the features effectively on their own. The tuning of CNN hyperparameters is critical in every dataset since it has a significant impact on the efficiency of the training model. Given the high dimensionality and complexity of the data, manual hyperparameter tuning would take an inordinate amount of time, with the possibility of failing to identify the optimal hyperparameters. In this paper, we proposed a Bayesian Optimization-based efficient hyperparameter optimization technique for CNN. This method was evaluated by classifying 3064 T-1-weighted CE-MRI images into three types of brain tumors (Glioma, Meningioma, and Pituitary). Based on Transfer Learning, the performance of five well-recognized deep pre-trained models is compared with that of the optimized CNN. After using Bayesian Optimization, our CNN was able to attain 98.70% validation accuracy at best without data augmentation or cropping lesion techniques, while VGG16, VGG19, ResNet50, InceptionV3, and DenseNet201 achieved 97.08%, 96.43%, 89.29%, 92.86%, and 94.81% validation accuracy, respectively. Moreover, the proposed model outperforms state-of-the-art methods on the CE-MRI dataset, demonstrating the feasibility of automating hyperparameter optimization.

Recently, many e-commerce websites have encouraged their users to rate shopping items and write review texts. This review information has been very useful for understanding user preferences and item properties, as well as enhancing the capability to make personalized recommendations of these websites. In this paper, we propose to model user preferences and item properties using convolutional neural networks (CNNs) with dual local and global attention, motivated by the superiority of CNNs to extract complex features. By using aggregated review texts from a user and aggregated review text for an item, our model can learn the unique features (embedding) of each user and each item. These features are then used to predict ratings. We train these user and item networks jointly which enable the interaction between users and items in a similar way as matrix factorization. The local attention provides us insight on a user's preferences or an item's properties. The global attention helps CNNs focus on the semantic meaning of the whole review text. Thus, the combined local and global attentions enable an interpretable and better-learned representation of users and items. We validate the proposed models by testing on popular review datasets in Yelp and Amazon and compare the results with matrix factorization (MF), the hidden factor and topical (HFT) model, and the recently proposed convolutional matrix factorization (ConvMF+). Our proposed CNNs with dual attention model outperforms HFT and ConvMF+ in terms of mean square errors (MSE). In addition, we compare the user/item embeddings learned from these models for classification and recommendation. These results also confirm the superior quality of user/item embeddings learned from our model.

Electroencephalographic (EEG) signals are widely used to analyze brain functions. EEG analysis may benefit from interpretable convolutional neural networks (CNNs), realizing data-driven analysis pipelines capable of automatically learning the most meaningful EEG features by exploiting the entire information contained in the EEG. However, the available solutions enable a direct analysis of the learned EEG features mainly within only one domain (frequency domain), and it is not available yet a fully-interpretable CNN that eases the interpretation of features living in all the main domains characterizing the EEG, that is, frequency, spatial, and temporal domains. This hinders the application of neural networks to EEG analysis. Here, we propose a fully-interpretable CNN learning features in the frequency, spatial, and temporal domains that can be easily interpreted once the network is trained. The network processes the input by learning the optimal bank of bandpass filters (modeled with generalized Gaussian functions), and then the optimal combination of EEG channels. Finally, the network learns how to optimally recombine time samples. We test our approach to motor imagery decoding. The network significantly outperforms existing state-of-the-art interpretable CNNs, while maintaining the highest degree of feature interpretability spanning across multiple domains. Finally, the analysis conducted on network features matched the known correlates of motor imagery in the frequency, spatial and temporal domains. Our network represents a comprehensive analysis tool that could be used to automatically extract, in multiple domains, the most salient EEG features for the brain states under analysis.

The precision and reliability of photovoltaic (PV) power forecasting play a crucial role in commercial PV plants. However, the stochastic and intermittent nature of solar radiation makes prediction difficult. Inspired by this, 4 different deep learning-based hybrid models are proposed to predict short-term PV power generation using long short term memory (LSTM) neural network and convolutional neural network (CNN) based on Bayesian Optimization (BO) in this paper. In addition, this paper explores feature selection using two benchmark models on different feature sets, and finally selects 5 features for prediction. The performances of direct forecasting results for both 1-hour ahead and 24-hour ahead of the above various models are compared on one year of hourly data from a real PV plant in Shandong, China. It is shown that using Bi-directional LSTM (BiLSTM) and CNN-BiLSTM models are more suitable for 1-hour ahead prediction, LSTM-CNN and CNN-BiLSTM models are more suitable for 24-hour ahead prediction. The case study shows that the model with Bayesian optimized optimal weights can reduce the error rate by up to 32.80% compared to the benchmark model and demonstrates the good prediction performance of the proposed approach on commercial PV plants.

Hyperspectral data consists of continuous narrow spectral bands. Due to this, it has less spatial and high spectral information. Convolutional neural networks (CNNs) emerge as a highly contextual information model for remote sensing applications. Unfortunately, CNNs have constraints in their underlying network architecture in regards to the global correlation of spatial and spectral features, making them less reliable for mining and representing the sequential properties of spectral signatures. In this article, limpid size attention network (LSANet) is proposed, which contains 3D and 2D convolution blocks for enhancement of spatial-spectral features of the hyperspectral image (HSI). In addition, limpid attention block (LAB) is designed to provide a global correlation of the spectral and spatial features through LS attention. Furthermore, the computational costs of LS-attention are less compared to the multi-head self-attention (MHSA) of the classical vision transformer (ViT). In the ViT encoder a conditional position encoding (CPE) module is utilized that dynamically generates tokens from the feature maps to capture a richer contextual representation. The LSANet obtained overall accuracy (OA) of 98.78%, 98.67%, 97.52% and 89.45%, respectively, on the Indian Pines (IP), Pavia University (PU), Salina Valley (SV) and Botswana datasets. Our model’s quantitative and qualitative results are considerably better than the classical CNN and transformer-based methods.

For hyperspectral image (HSI) classification, two branch networks generally use the convolution neural networks (CNNs) to extract the spatial features and the long short-term memory (LSTM) to learn the spectral features. However, CNN with a local kernel neglects the global properties of the whole HSI. LSTM doesn't consider the macroscopic and detailed information of spectra. In this paper, we propose a dual-view spectral and global spatial feature fusion network (DSGSF) to extract the spatial-spectral features for HSI classification, including a spatial subnetwork and a spectral subnetwork. In the spatial subnetwork, we propose a global spatial feature representation model based on the encoder-decoder structure with channel attention and spatial attention to learn the global spatial features. In the spectral subnetwork, we design a dual-view spectral feature aggregation model with view attention to learn the diversity of spectral features. By fusing the two subnetworks, we construct DSGSF to extract the spatial-spectral features of HSI with strong discriminating performance. Experimental results on three public datasets illustrate that the proposed method can achieve competitive results compared with the state-of-the-art methods. Code: https://github.com/RZWang-WH/DSGSF.

Graph convolutional network – Long short term memory neural network- multi layer perceptron- Gaussian progress regression model: A new deep learning model for predicting ozone concertation

Developmental disabilities such as autism spectrum disorder (ASD) affect a person’s ability to interact socially, and communicate effectively and also cause behavioral issues. Children with ASD cannot be cured but they might benefit from early intervention to enhance their cognitive abilities, favorite their growth , and affect their lives and families in a positive way. Multiple standard ASD screening tools are used such as the autism diagnostic observational schedule (ADOS) and the autism diagnostic interview (ADI), which are known to be lengthy and challenging without specialist training to administrate and score. The process of ASD assessment can be time-consuming and costly, and the growing number of autistic cases worldwide indicates an urgent need for a quick, simple, and dependable self-administered autism screening tool that may be used if a child displays some of the common signs of autism, and to ensure whether or not he should seek professional full ASD diagnosis. According to a number of studies, ASD individuals exhibit facial phenotypes that are distinct from those of normally developing children. Furthermore, convolutional neural networks (CNN) have mostly found utility in image classification applications due to their high classification accuracy. Using facial images, a dense convolutional network (Densenet) model, and cloud-based advantages, in this paper we proposed a practical, fast, and easy-to-use ASD online screening approach. Easily available through the internet via the link “https://asd-detector.herokuapp.com/”, our suggested web-based screening instrument may be a practical and trustworthy tool for practitioners in their ASD diagnostic procedures with a 98 percent testing dataset classification accuracy.

Deep learning algorithms have shown significant performance improvements for facial expression recognition (FER). Most deep learning-based methods, however, focus more attention on spatial appearance features for classification, discarding much useful temporal information. In this work, we present a novel framework that jointly learns spatial features and temporal dynamics for FER. Given the image sequence of an expression, spatial features are extracted from each frame using a deep network, while the temporal dynamics are modeled by a convolutional network, which takes a pair of consecutive frames as input. Finally, the framework accumulates clues from fused features by a BiLSTM network. In addition, the framework is end-to-end learnable, and thus temporal information can be adapted to complement spatial features. Experimental results on three benchmark databases, CK+, Oulu-CASIA and MMI, show that the proposed framework outperforms state-of-the-art methods.

The aim of the present work was to propose and implement deep neural network (DNN)-based handheld diagnosis system for more accurate diagnosis and severity assessment of individuals with autism spectrum disorder (ASD). Initially, the learning of the proposed system for ASD diagnosis was performed by implementing DNN algorithms such as a convolutional neural network (CNN) and long short-term memory (LSTM), and multilayer perceptron (MLP) with DSM-V based acquired dataset. The performance of the DNN algorithms was analyzed based on parameters viz. accuracy, loss, mean squared error (MSE), precision, recall, and area under the curve (AUC) during the training and validation process. Later, the optimum DNN algorithm, among the tested algorithms, was implemented on handheld diagnosis system (HDS) and the performance of HDS was analyzed. The stability of proposed DNN-based HDS was validated with the dataset group of 20 ASD and 20 typically developed (TD) individuals. It was observed during comparative analysis that LSTM resulted better in ASD diagnosis as compared to other artificial intelligence (AI) algorithms such as CNN and MLP since LSTM showed stabilized results achieving maximum accuracy in less consumption of epochs with minimum MSE and loss. Further, the LSTM based proposed HDS for ASD achieved optimum results with 100% accuracy in reference to DSM-V, which was validated statistically using a group of ASD and TD individuals. The use of advanced AI algorithms could play an important role in the diagnosis of ASD in today's era. Since the proposed LSTM based HDS for ASD and determination of its severity provided accurate results with maximum accuracy with reference to DSM-V criteria, the proposed HDS could be the best alternative to the manual diagnosis system for diagnosis of ASD.

Hyperspectral image classification has been widely applied in many fields, but it also faces challenges because of small number of labeled samples. In this paper, we propose the Multiscale Multistage Spectral-Spatial Feature Fusion Framework (M^2 S^2 F^2 ) for hyperspectral image classification using small training samples. The Framework is the combination of two deep convolutional neural networks, which can extract more representative and discriminative features by combining the following operations. Firstly, two different scale 3-D cubes are the inputs for the spectral and spatial feature extraction respectively. Secondly, by fusing strong complementary information between different layers, we form multistage spectral and spatial features by fusion primary, intermediate and advanced features of the spectral and spatial features respectively. Spectral and spatial features are extracted by spectral and spatial skipped residual blocks, which can effectively alleviate the problems of gradient degradation. Thirdly, the fusion of complementary multistage spectral and spatial features can improve the classification accuracy. Experimental results on the IN, UP and KSC datasets show the effectiveness of the proposed method using small training samples.

In recent years, the integration of convolutional neural networks (CNNs) and Transformers has significantly advanced hyperspectral image (HSI) classification by jointly capturing local and global features. However, most existing methods primarily focus on the fusion of spectral–spatial features while neglecting the complementary information contained in frequency domain features. To address this issue, we propose a spatial frequency cross-attention fusion network (SFCFNet) that jointly models spectral, spatial, and frequency-domain features for HSI classification. The framework consists of three core modules: first, the multiscale spectral spatial feature learning (MSSFL) module extracts joint spectral spatial features using multi-scale 3D and 2D convolutions. Next, the triple-branch representation module (TBRM) employs three branches to capture global spatial features of large-scale structures, local spatial features of fine grained textures, and multi-scale frequency features based on Haar wavelet decomposition, providing complementary multi domain representations for subsequent deep fusion. Finally, the dual-domain feature cross-attention fusion module (DFCF) achieves effective fusion of spatial structures and frequency domain textures, enhancing the model's ability to separate complex backgrounds from fine-grained targets and thereby improving classification performance. Compared with other methods, SFCFNet achieves higher overall accuracy (OA) on the Salinas, Houston2013, WHU-Hi-LongKou, and Xuzhou datasets, reaching 99.05%, 98.07%, 98.76%, and 98.18%, respectively.

We developed a computer-aided diagnosis (CADx) method for classification between benign nodule, primary lung cancer, and metastatic lung cancer and evaluated the following: (i) the usefulness of the deep convolutional neural network (DCNN) for CADx of the ternary classification, compared with a conventional method (hand-crafted imaging feature plus machine learning), (ii) the effectiveness of transfer learning, and (iii) the effect of image size as the DCNN input. Among 1240 patients of previously-built database, computed tomography images and clinical information of 1236 patients were included. For the conventional method, CADx was performed by using rotation-invariant uniform-pattern local binary pattern on three orthogonal planes with a support vector machine. For the DCNN method, CADx was evaluated using the VGG-16 convolutional neural network with and without transfer learning, and hyperparameter optimization of the DCNN method was performed by random search. The best averaged validation accuracies of CADx were 55.9%, 68.0%, and 62.4% for the conventional method, the DCNN method with transfer learning, and the DCNN method without transfer learning, respectively. For image size of 56, 112, and 224, the best averaged validation accuracy for the DCNN with transfer learning were 60.7%, 64.7%, and 68.0%, respectively. DCNN was better than the conventional method for CADx, and the accuracy of DCNN improved when using transfer learning. Also, we found that larger image sizes as inputs to DCNN improved the accuracy of lung nodule classification.

Electrochemical machining is a promising non-traditional manufacturing process to make high-quality parts. The benefits of minimal thermally and mechanically induced stresses, free of burr, and a low surface roughness are appealing for industry and research institutes. However, the combined chemical reaction, electric field, fluid mechanics, and material properties involve a significant number of independent parameters which are difficult to analyze in order to draw comprehensive conclusions. To our current knowledge, process responses such as the material removal rate, optimal feed rate, and cutting profile cannot be represented accurately by analytical solutions. In recent years, deep learning has had tremendous success in analyzing sophisticated systems. The improved computation efficiency and reduced size of the training dataset required for deep learning have enabled various prediction models in the manufacturing industry. In this paper, a new approach is developed using the deep convolutional network with the Bayesian optimization algorithm to predict the diameters of the drilled hole from an electrochemical machining process. The Keras application programming interface (API) was used to build the deep convolutional network; the feed rate, pulse-on time, and voltage were used as input parameters to provide a fair comparison with a neural network from previous research. Random dropout layers were added to prevent overfitting of the network. Instead of tuning the network parameter by trial and error, the Bayesian parameter optimization algorithm was implemented to find the optimal set of parameters of the deep convolutional network that yields the minimum mean square error. The proposed algorithm was compared with a previously developed neural network with partially embedded physical knowledge. Improved training speed and accuracy were observed in comparison with the traditional neural network. The prediction model using the proposed deep learning algorithm demonstrated better prediction accuracy and provided a more systematic way to select the hyperparameter for the deep convolutional network.

Abstract. An accurate and rapid urban flood prediction model is essential to support decision-making for flood management. This study developed a deep-learning-technique-based data-driven model for flood predictions in both temporal and spatial dimensions, based on an integration of long short-term memory (LSTM) network, Bayesian optimization, and transfer learning techniques. A case study in northern China was applied to test the model performance, and the results clearly showed that the model can accurately predict the maximum water depths and flood time series for various hyetograph inputs, with substantial improvements in the computation time. The model predicted flood maps 19 585 times faster than the physically based hydrodynamic model and achieved a mean relative error of 9.5 %. For retrieving the spatial patterns of water depths, the degree of similarity of the flood maps was very high. In a best case scenario, the difference between the ground truth and model prediction was only 0.76 %, and the spatial distributions of inundated paths and areas were almost identical. With the adoption of transfer learning, the proposed model was well applied to a new case study and showed robust compatibility and generalization ability. Our model was further compared with two baseline prediction algorithms (artificial neural network and convolutional neural network) to validate the model superiority. The proposed model can potentially replace and/or complement the conventional hydrodynamic model for urban flood assessment and management, particularly in applications of real-time control, optimization, and emergency design and planning.