Neural Network Architecture Research Articles

Advanced neural network systems for analysis of blood-based biomarkers in metastatic disease detection

Neural network systems have become the gold provocative standard in blood-based biomarker analysis for metastatic disease diagnosis, providing higher sensitivity and specificity for early cancer detection. The lexical method also showed that recent developments in artificial intelligence and machine learning technologies allow for the complex interpretation of various types of biomarkers, such as circulating tumor cells, cell-free DNA, microRNAs, and proteins. Many biochemical patterns of biomarkers could be easily handled through sophisticated algorithms and depending on these active approaches in diagnosing metastatic outcomes more accurately at earlier stages. Literature analysis involved a targeted selection of scientific peer-reviewed articles dedicated to neural network utilization in blood-based biomarker screening for cancer. Doing a systematic evaluation critically assessed methodologies associated with artificial intelligence, types of biomarkers, methods of detecting biomarkers, and clinical validation strategies. Filtering was done based on several factors including the identification of new architectures for neural networks, new biomarker analysis methods, and clinical applications primarily focusing on the early detection of metastatic diseases. Analysis of reviewed studies showed a high diagnostic accuracy when neural network systems are used for biomarker analysis. Volumes in Cancer Computational Biology Vol reported that the machine learning models have achieved different sensitivity rates of over 90% including the ability to detect early-stage metastatic disease. The use of several categories of biomarkers and application-improved neural networks for their evaluation showed that the combined use of markers yields better diagnostic results than using single biomarkers. Neural network systems have significant potential to transform metastatic disease diagnosis using circulating biomarker detection. Using sophisticated computational models biomarker patterns are analyzed in detail and the diagnosis is made earlier, thus contributing to better treatment outcomes. Some complexities that are potential barriers include enforcing analysis protocols as standards, correcting and calibrating, and embracing routine technological practice. High-end artificial neural networks are the new frontier in blood-based biomarker analysis of metastatic disease diagnosis. Applying AI within the classic biomarker detection approach increases diagnostic reliability, detection timeliness, and the overall outlook for patient treatment. The advancement in neural network architectures as well as biomarker analysis techniques provides for further improvement in cancer diagnostics in the future.

Adaptive neural network for quantum error mitigation

Quantum computing holds transformative promise, but its realization is hindered by the inherent susceptibility of quantum computers to errors. Quantum error mitigation has proved to be an enabling way to reduce computational error in present noisy intermediate scale quantum computers. This research introduces an innovative approach to quantum error mitigation by leveraging machine learning, specifically employing adaptive neural networks. With experiment and simulations done on 127-qubit IBM superconducting quantum computer, we were able to develop and train a neural network architecture to dynamically adjust output expectation values based on error characteristics. The model leverages a prior classifier module outcome on simulated quantum circuits with errors, and the antecedent neural network regression module adapts its parameters and response to each error characteristics. Results demonstrate the adaptive neural network’s efficacy in mitigating errors across diverse quantum circuits and noise models, showcasing its potential to surpass traditional error mitigation techniques with an accuracy of 99% using the fully adaptive neural network for quantum error mitigation. This work presents a significant application of classical machine learning methods towards enhancing the robustness and reliability of quantum computations, providing a pathway for the practical realization of quantum computing technologies.

Marigold: a machine learning-based web app for zebrafish pose tracking

BackgroundHigh-throughput behavioral analysis is important for drug discovery, toxicological studies, and the modeling of neurological disorders such as autism and epilepsy. Zebrafish embryos and larvae are ideal for such applications because they are spawned in large clutches, develop rapidly, feature a relatively simple nervous system, and have orthologs to many human disease genes. However, existing software for video-based behavioral analysis can be incompatible with recordings that contain dynamic backgrounds or foreign objects, lack support for multiwell formats, require expensive hardware, and/or demand considerable programming expertise. Here, we introduce Marigold, a free and open source web app for high-throughput behavioral analysis of embryonic and larval zebrafish.ResultsMarigold features an intuitive graphical user interface, tracks up to 10 user-defined keypoints, supports both single- and multiwell formats, and exports a range of kinematic parameters in addition to publication-quality data visualizations. By leveraging a highly efficient, custom-designed neural network architecture, Marigold achieves reasonable training and inference speeds even on modestly powered computers lacking a discrete graphics processing unit. Notably, as a web app, Marigold does not require any installation and runs within popular web browsers on ChromeOS, Linux, macOS, and Windows. To demonstrate Marigold’s utility, we used two sets of biological experiments. First, we examined novel aspects of the touch-evoked escape response in techno trousers (tnt) mutant embryos, which contain a previously described loss-of-function mutation in the gene encoding Eaat2b, a glial glutamate transporter. We identified differences and interactions between touch location (head vs. tail) and genotype. Second, we investigated the effects of feeding on larval visuomotor behavior at 5 and 7 days post-fertilization (dpf). We found differences in the number and vigor of swimming bouts between fed and unfed fish at both time points, as well as interactions between developmental stage and feeding regimen.ConclusionsIn both biological experiments presented here, the use of Marigold facilitated novel behavioral findings. Marigold’s ease of use, robust pose tracking, amenability to diverse experimental paradigms, and flexibility regarding hardware requirements make it a powerful tool for analyzing zebrafish behavior, especially in low-resource settings such as course-based undergraduate research experiences. Marigold is available at: https://downeslab.github.io/marigold/.

Integrating control-theoretic predictive deep learning for resource-efficient computing systems

Abstract Deep learning (DL) powers numerous applications on ubiquitous edge devices, but its high resource demands pose a challenge. Approximate computing is often proposed to alleviate this, yet such calculation usually suffers from a fixed level of accuracy loss. We propose a novel control-theoretic approach for predictive DL inference on resource constrained devices. Our system dynamically adjusts approximation levels based on a trade-off between resource utilisation and accuracy, considering future demands. Extensive experiments across diverse domains - human activity recognition, acoustic scene profiling, and computer vision - with various neural network architectures and approximation techniques, demonstrate that our approach achieves up to 50% energy savings while maintaining the desired inference accuracy and incurring minimal runtime overhead. Furthermore, we showcase our method in a real-world deployment on low-power edge devices and confirm its superiority over current state-of-the-art solutions.

Research on Urban Road Traffic Flow Prediction Based on Sa-Dynamic Graph Convolutional Neural Network

Neural network models based on GNNs often achieve good results in traffic flow prediction tasks of traffic networks. However, most existing GNN-based methods apply a fixed graph structure to capture spatial dependencies between nodes, and fixed graph structures may not be able to reflect the spatiotemporal changes in node dependencies. To address this, introducing a self-attention mechanism applied to an adaptive adjacency matrix, the neural network architecture is improved based on Graph WaveNet, and a new approach called self-attention dynamic graph wave network (SA-DGWN) is proposed, which can fit the spatiotemporal dependencies of the road network. In an experiment, traffic flow data were extracted based on RFID from certain roads in Nanjing, China. The results show that under the same configuration, compared to Graph WaveNet, MAE, MAPE, and RMSE from the proposed method reduced by 3.08%, 3.68%, and 2.6%, respectively. In addition, for the training data, we explored the impact of temporal feature and sampling periods on the training effect. The additional results indicate that adding hour-minute-second information to the input improved the model’s accuracy, reducing MAE, MAPE, and RMSE by 15.28%, 12.28%, and 14.01%, respectively. Adding day-of-the-week features also brought substantial performance improvements. For different sampling periods, the model performed better overall with a 10 min sampling period compared to 5 min and 15 min periods. For single-step prediction tasks, the longer the sampling period, the better the prediction effect.

Optimizing Convolutional Neural Network Architectures with Optimal Activation Functions for Pediatric Pneumonia Diagnosis Using Chest X-rays

Pneumonia remains a significant cause of morbidity and mortality among pediatric patients worldwide. Accurate and timely diagnosis is crucial for effective treatment and improved patient outcomes. Traditionally, pneumonia diagnosis has relied on a combination of clinical evaluation and radiologists’ interpretation of chest X-rays. However, this process is time-consuming and prone to inconsistencies in diagnosis. The integration of advanced technologies such as Convolutional Neural Networks (CNNs) into medical diagnostics offers a potential to enhance the accuracy and efficiency. In this study, we conduct a comprehensive evaluation of various activation functions within CNNs for pediatric pneumonia classification using a dataset of 5856 chest X-ray images. The novel Mish activation function was compared with Swish and ReLU, demonstrating superior performance in terms of accuracy, precision, recall, and F1-score in all cases. Notably, InceptionResNetV2 combined with Mish activation function achieved the highest overall performance with an accuracy of 97.61%. Although the dataset used may not fully represent the diversity of real-world clinical cases, this research provides valuable insights into the influence of activation functions on CNN performance in medical image analysis, laying a foundation for future automated pneumonia diagnostic systems.

Deep Reinforcement Learning‐Based Control for Real‐Time Hybrid Simulation of Civil Structures

ABSTRACTReal‐time Hybrid Simulation (RTHS) is a cyber‐physical technique that studies the dynamic behavior of a system by combining physical and numerical components that are coupled through a boundary condition enforcer. In structural engineering, the numerical components are subjected to environmental loads that become dynamic displacements of the physical substructure applied through an actuator. However, the dynamics of the coupling between components and the complexities of the system lead to synchronization challenges that affect the accuracy of the simulation. Thus, requiring tracking controllers to ensure the fidelity of the simulation. This paper studies deep reinforcement learning (DRL) as a novel alternative to designing the tracking controller. Three controllers are designed: a DRL agent combined with a conventional time delay compensation, a conventional feedback controller combined with a DRL agent, and a DRL agent with a complex neural network architecture. The proposed approaches are tested using a virtual RTHS benchmark problem, and the results are compared with an optimized controller that has a proportional‐integral‐derivative controller and phase‐lead compensation. The results show that DRL can address the synchronization challenges of RTHS with a model‐free approach and simple neural network architectures. The work shown in this study is a critical step toward model‐free control methodologies that can transform and further develop the RTHS method. The proposed methodology can be used to address important challenges related to RTHS, including nonlinearities and uncertainties of the physical substructure, complex boundary conditions, and the computational efficiency when physical structures with complex dynamics are present.

Can artificial intelligence understand our emotions? Deep learning applications with face recognition

Abstract Objective The aim of this study is to evaluate the ability to detect emotions from human facial expressions via facial recognition technologies and analyze the effectiveness of deep learning models in this process. Method This research was conducted between 01.04 and 01.07.2024. The data of the study were taken from the open access site https://www.kaggle.com/datasets/msambare/fer2013 (Kaggle, 2024). Python 3.8 is used in this study. The FER-2013 (Facial Expression Recognition 2013) dataset is a comprehensive collection of facial images labeled with various emotions. The dataset contains 35,887 grayscale facial images. Each image has a size of 48 × 48 pixels. The dataset consists of images belonging to 7 emotion categories: anger, disgust, fear, happy, sad, confused, and neutral. Results In our experiments on the FER2013 dataset, we evaluated the performance of three different models: MobileNetV3-L, EfficientNetV2-L, and our proposed EfficientMobileNet. The evaluation criteria were based on sensitivity, specificity, accuracy, and F1 scores to assess the effectiveness of each model comprehensively. The EfficientMobileNet model outperformed MobileNetV3-L and EfficientNetV2-L in all measured performance metrics. EfficientMobileNet was the most successful model for predicting emotions, with an accuracy of 77.6%. Conclusion The impressive results obtained by EfficientMobileNet on the Fer2013 dataset show potential for wider application, especially in image classification tasks involving low-quality or small-scale images. This performance supports the idea of the potential for further improvements in neural network architecture and model efficiency and accuracy. Future work should focus on optimizing the model for more challenging datasets, studying the impact of different architectural adjustments, and investigating the scalability of EfficientMobileNet across various domains and applications.

Revolutionizing Citrus Health: AI-Based Detection of Greening Disease and Nutrient Deficiencies in Sweet Orange

This paper explores the application of Artificial Intelligence (AI) techniques for detecting plant diseases, focusing on citrus greening disease and various foliar nutrient deficiencies in sweet orange. Agriculture faces numerous challenges from cultivation to harvesting, including significant yield losses due to disease infections and environmental hazards from excessive use of insecticides and fungicides. As the global population grows, the demand for food is surging, and traditional farming methods fall short in meeting this demand, often degrading soil health through intensive pesticide use. AI offers significant advantages over conventional techniques in disease detection. This study employs AI for visual detection of citrus greening disease and foliar nutrient deficiencies, achieving 87% accuracy on a test dataset of infected, nutrient-deficient, and healthy sweet orange leaves. Performance is evaluated using metrics such as Accuracy, Recall, Precision, and F1-Score. Specifically, Convolutional Neural Network (CNN) architectures, including Visual Geometry Group (VGG-16), are utilized for image-based detection and classification, demonstrating the potential of AI in enhancing plant health monitoring.

Influence of environmental renewal on students’ mental health development under the background of long-term online teaching

In the wake of the COVID-19 pandemic, the prevalence of online education in primary education has exhibited an upward trajectory. Relative to traditional learning environments, online instruction has evolved into a pivotal pedagogical modality for contemporary students. Thus, to comprehensively comprehend the repercussions of environmental changes on students’ psychological well-being in the backdrop of prolonged online education, this study employs an innovative methodology. Founded upon three elemental feature sequences—images, acoustics, and text extracted from online learning data—the model ingeniously amalgamates these facets. The fusion methodology aims to synergistically harness information from diverse perceptual channels to capture the students’ psychological states more comprehensively and accurately. To discern emotional features, the model leverages support vector machines (SVM), exhibiting commendable proficiency in handling emotional information. Moreover, to enhance the efficacy of psychological well-being prediction, this study incorporates an attention mechanism into the traditional Convolutional Neural Network (CNN) architecture. By innovatively introducing this attention mechanism in CNN, the study observes a significant improvement in accuracy in identifying six psychological features, demonstrating the effectiveness of attention mechanisms in deep learning models. Finally, beyond model performance validation, this study delves into a profound analysis of the impact of environmental changes on students’ psychological well-being. This analysis furnishes valuable insights for formulating pertinent instructional strategies in the protracted context of online education, aiding educational institutions in better addressing the challenges posed to students’ psychological well-being in novel learning environments.

Enhancing lane detection in autonomous vehicles with multi-armed bandit ensemble learning

This study introduces a novel ensemble learning technique namely Multi-Armed Bandit Ensemble (MAB-Ensemble), designed for lane detection in road images intended for autonomous vehicles. The foundation of the proposed MAB-Ensemble technique is inspired in terms of Multi-Armed bandit optimization to facilitate efficient model selection for lane segmentation. The benchmarking dataset namely TuSimple is used for training, validating and testing the proposed and existing lane detection techniques. Convolutional Neural Networks (CNNs) architecture which includes ENet, PINet, ResNet-50, ResNet-101, SqueezeNet, and VGG16Net are employed in lane detection problems to construct segmentation models and demonstrate proficiency in distinct road conditions. However, the proposed MAB-Ensemble technique overcomes the limitations of individual models by dynamically selecting the most suitable CNN model based on prevailing environmental factors. The proposed technique optimizes the segmentation accuracy and treats the attained accuracy as a reward signal in the context of reinforcement learning by interacting with the environment through CNN model selection. The MAB-Ensemble achieved an overall accuracy of 90.28% in different road conditions. The results overcome the performance of the individual CNN models and state-of-the-art ensemble techniques. Also, it demonstrates superior performance which includes daytime, night-time, and abnormal road conditions. The MAB-Ensemble technique offers a promising solution for robust lane detection by harnessing the collective strengths of diverse CNN models.

Real-time single-pixel video imaging based on deep learning

The emergence of compressed sensing (CS) theory has enabled the development of single-pixel cameras that achieve high-resolution imaging using a single photodetector. However, traditional CS reconstruction algorithms require significant computational time and face an inherent trade-off between imaging resolution and frame rate, limiting current single-pixel cameras to static scene imaging. A key challenge lies in achieving real-time single-pixel imaging with both high frame rate and high resolution. This paper proposes a real-time single-pixel imaging technology based on deep learning. We design a deep convolutional neural network architecture incorporating residual networks to simulate the measurement and reconstruction process of CS-based single-pixel imaging. The network is trained on an image dataset and subsequently deployed for single-pixel imaging. The trained network can complete image reconstruction at a low sampling rate with minimal latency, enabling real-time single-pixel video capture at 128×128 resolution with 33 frames per second (fps) at a 4\% sampling rate. Furthermore, we implement a four-channel parallel signal processing method to achieve real-time single-pixel imaging video at 256×256 resolution at 33 fps.

Comprehensive Evaluation of Deepfake Detection Models: Accuracy, Generalization, and Resilience to Adversarial Attacks

The rise of deepfakes—synthetic media generated using artificial intelligence—threatens digital content authenticity, facilitating misinformation and manipulation. However, deepfakes can also depict real or entirely fictitious individuals, leveraging state-of-the-art techniques such as generative adversarial networks (GANs) and emerging diffusion-based models. Existing detection methods face challenges with generalization across datasets and vulnerability to adversarial attacks. This study focuses on subsets of frames extracted from the DeepFake Detection Challenge (DFDC) and FaceForensics++ videos to evaluate three convolutional neural network architectures—XCeption, ResNet, and VGG16—for deepfake detection. Performance metrics include accuracy, precision, F1-score, AUC-ROC, and Matthews Correlation Coefficient (MCC), combined with an assessment of resilience to adversarial perturbations via the Fast Gradient Sign Method (FGSM). Among the tested models, XCeption achieves the highest accuracy (89.2% on DFDC), strong generalization, and real-time suitability, while VGG16 excels in precision and ResNet provides faster inference. However, all models exhibit reduced performance under adversarial conditions, underscoring the need for enhanced resilience. These findings indicate that robust detection systems must consider advanced generative approaches, adversarial defenses, and cross-dataset adaptation to effectively counter evolving deepfake threats.

Er2O3-Doped Lead Borate Glasses: Advanced Optical and Radiation Shielding Performance

Abstract A new glass system with the composition 60B2O3 + 30PbF2 + (10−x)K2O + x Er2O3 (x = 0 to 3 mol%) were synthesized using the melt-quenching technique and comprehensively analyzed to evaluate their structural, optical, mechanical, and radiation shielding properties. Increasing Er2O3 concentration enhanced the density (from 4.260 to 4.89 g/cm3) and reduced the molar volume (from 29.28 to 28.98 cm3/mol), indicating a denser and more compact glass matrix. Optical studies revealed increased UV absorbance, a red shift in the cutoff wavelength, and a reduction in the optical energy gap from 3.487 to 3.335 eV (direct transitions). Urbach energy values increased from 0.722 to 1.083 eV, signifying heightened structural disorder. Enhanced refractive index and extinction coefficients further underscored the glasses’ potential for optical applications. Mechanical analyses demonstrated a significant increase in all elastic moduli, including Young’s, bulk, and shear moduli, with Er2O3 incorporation, indicating improved rigidity and mechanical stability. The radiation shielding performance of the glasses was assessed across photon energies of 0.015–15 MeV, incorporating both experimental data and machine learning (ML)-based predictions of mass attenuation coefficients (MAC). The ML model, developed using a neural network architecture, successfully predicted MAC values with high accuracy, demonstrating excellent agreement with XCOM-calculated results. Key shielding parameters, including half-value layer (HVL), effective atomic number (Zeff), and buildup factors (EABF and EBF), improved significantly with higher Er2O3 content. BPKE3 glass, with the highest Er2O3 concentration, exhibited the best shielding efficiency, outperforming conventional shielding materials in terms of lower HVL and buildup factors, coupled with higher MAC and Zeff values. This study highlights the dual role of Er2O3-doped lead borate glasses as efficient optical and radiation shielding materials. Machine learning effectively predicts shielding parameters, aiding material optimization for applications in nuclear, medical, and industrial fields.

Brain-inspired multimodal motion and fine-grained action recognition

IntroductionTraditional action recognition methods predominantly rely on a single modality, such as vision or motion, which presents significant limitations when dealing with fine-grained action recognition. These methods struggle particularly with video data containing complex combinations of actions and subtle motion variations.MethodsTypically, they depend on handcrafted feature extractors or simple convolutional neural network (CNN) architectures, which makes effective multimodal fusion challenging. This study introduces a novel architecture called FGM-CLIP (Fine-Grained Motion CLIP) to enhance fine-grained action recognition. FGM-CLIP leverages the powerful capabilities of Contrastive Language-Image Pretraining (CLIP), integrating a fine-grained motion encoder and a multimodal fusion layer to achieve precise end-to-end action recognition. By jointly optimizing visual and motion features, the model captures subtle action variations, resulting in higher classification accuracy in complex video data.Results and discussionExperimental results demonstrate that FGM-CLIP significantly outperforms existing methods on multiple fine-grained action recognition datasets. Its multimodal fusion strategy notably improves the model's robustness and accuracy, particularly for videos with intricate action patterns.

Combining Multi-Scale Fusion and Attentional Mechanisms for Assessing Writing Accuracy

Traditional methods of assessing handwritten characters are often too subjective, inefficient, and lagging in feedback, which makes it difficult for educators to achieve fully objective writing assessments and for writers to receive timely suggestions for improvement. In this paper, we propose a convolutional neural network (CNN) architecture that combines the attention mechanism with multi-scale feature fusion; specifically, the features are weighted by designing a bottleneck layer that combines the Squeeze-and-Excitation (SE) attention mechanism to highlight the important information and by applying a multi-scale feature fusion method to enable the network to capture both the global structure and the local details of Chinese characters. Finally, a high-quality dataset containing 26,800 images of handwritten Chinese characters is constructed based on the application scenario of the writing grade test, covering the common Chinese characters in the writing grade exam; The experimental results show that the proposed method achieves 98.6% accuracy on the writing grade exam dataset and 97.05% on the ICDAR-2013 public dataset, significantly improving recognition accuracy. The constructed dataset and improved model are suitable for application scenarios such as writing grade exams, which helps to improve marking efficiency and accuracy.

Deep Learning Framework Using Spatial Attention Mechanisms for Adaptable Angle Estimation Across Diverse Array Configurations

Rapid advancement of wireless communication systems and the increasing need for accurate, real-time signal processing have driven innovations in direction-of-arrival (DoA) estimation techniques. This paper introduces a novel convolutional neural network (CNN) architecture that combines spatial attention mechanisms with a transfer learning framework to enhance both accuracy and versatility in DoA estimation. The model integrates spatial attention layers to dynamically prioritize signal regions with the highest information value, allowing it to isolate relevant signals and suppress interference in noisy or crowded signal environments. In addition, we utilize a transfer learning framework that enables the model to generalize across various antenna array configurations (i.e., planar, linear, and circular arrays) with minimal additional training. Extensive simulation results benchmark the proposed model against existing state-of-the-art methods for DoA estimation, achieving improved absolute error across diverse conditions. This hybrid approach not only enhances DoA estimation precision, but also significantly reduces retraining requirements when adapting to new array configurations, positioning it as a robust, scalable tool for next-generation wireless communication systems.

Design of an Optimal Convolutional Neural Network Architecture for MRI Brain Tumor Classification by Exploiting Particle Swarm Optimization

The classification of brain tumors using MRI scans is critical for accurate diagnosis and effective treatment planning, though it poses significant challenges due to the complex and varied characteristics of tumors, including irregular shapes, diverse sizes, and subtle textural differences. Traditional convolutional neural network (CNN) models, whether handcrafted or pretrained, frequently fall short in capturing these intricate details comprehensively. To address this complexity, an automated approach employing Particle Swarm Optimization (PSO) has been applied to create a CNN architecture specifically adapted for MRI-based brain tumor classification. PSO systematically searches for an optimal configuration of architectural parameters—such as the types and numbers of layers, filter quantities and sizes, and neuron numbers in fully connected layers—with the objective of enhancing classification accuracy. This performance-driven method avoids the inefficiencies of manual design and iterative trial and error. Experimental results indicate that the PSO-optimized CNN achieves a classification accuracy of 99.19%, demonstrating significant potential for improving diagnostic precision in complex medical imaging applications and underscoring the value of automated architecture search in advancing critical healthcare technology.

Music Genre Classification: A Comprehensive Study on Feature Fusion with CNN and MLP Architectures

The classification of music genres has been a key focus in the field of music information retrieval (MIR), many researchers are also looking for a number of ways to solve this problem. Early approach Feature-based methods such as Support Vector Machines (SVMs) and K-nearest Neighbors (k-NN) are mainly used for handcrafted features such as Mel frequency cepstrum coefficients (MFCC). As convolutional neural networks (CNNs) are increasingly used to capture complex time-frequency patterns in audio data, the performance and accuracy of music genre classification has been dramatically improved.Recent research advances have been made in integrating attention mechanisms and mixing mechanisms, which further improves the accuracy of the genre classification model. But that's it. Methods often rely on a single type of input data, such as spectrograms or raw audio data, and are therefore limited. They are able to fully capture the multifaceted nature of music.In this research, we propose a novel multi-input neural network architecture that uniquely integrates a CNN for Mel spectrogram processing with a Multilayer Perceptron (MLP) for handling additional feature data extracted from CSV files. This two-branch approach can effectively help us understand music data more comprehensively while processing spectral and feature-based information. The GTZAN dataset was a great help for our experiments, and the details of this part will be explained later. In addition, our model effectively combines the advantages of both CNN and MLP methods, thereby improving the classification accuracy of multiple music genres. And finally, our results show that the system achieves [0.77] accuracy, support [10], recall [0.67], and F1 score [0.65].

Knowledge-point classification using simple LSTM-based and siamese-based networks for virtual patient simulation

BackgroundIn medical education, enhancing thinking skills is vital. The Virtual Diagnosis and Treatment Platform (VP) refines medical students’ diagnostic abilities through interactive patient interviews (simulated patient interactions). By analyzing the questions asked during these interviews, the VP evaluates students’ aptitude in medical history inquiries, offering insights into their thinking capabilities. This study aimed to extract insights from case summaries and patient interviews to improve evaluation and feedback in medical education.MethodsThis study employs a systematic approach to knowledge-point classification by utilizing both simple long short-term memory (LSTM)-based and Siamese-based networks, coupled with cross-validation techniques. The dataset under scrutiny originates from the “Clinical Diagnosis and Treatment Skills Competitions” spanning the first to third years in Taiwan. The methodology involves generating knowledge points from sequential questions posed during case summaries and patient interviews. These knowledge points are then subjected to classification using the designated neural network architectures.ResultsThe experimental findings reveal promising outcomes, particularly when the Siamese-based network is used for knowledge-point classification. Through repeated (stratified) 10-fold cross validation, the accuracies achieved consistently exceeded 93%, with a standard deviation less than 0.007. These results underscore the efficacy of the proposed methodologies in enhancing virtual clinical diagnosis systems.ConclusionsThis study underscores the viability of leveraging advanced neural network architectures, particularly the Siamese-based network, for knowledge-point classification within virtual clinical diagnosis systems. By effectively discerning and classifying knowledge points derived from case summaries and patient interviews, these systems offer invaluable insights into students’ thinking capabilities in medical education. The robust accuracies attained through cross-validation affirm the feasibility and efficacy of the proposed methodologies, thus paving the way for enhanced virtual clinical training platforms.