High Classification Performance Research Articles

MRI-Driven Alzheimer’s Disease Diagnosis Using Deep Network Fusion and Optimal Selection of Feature

Alzheimer’s disease (AD) is a degenerative neurological condition characterized by cognitive decline, memory loss, and reduced everyday function, which eventually causes dementia. Symptoms develop years after the disease begins, making early detection difficult. While AD remains incurable, timely detection and prompt treatment can substantially slow its progression. This study presented a framework for automated AD detection using brain MRIs. Firstly, the deep network information (i.e., features) were extracted using various deep-learning networks. The information extracted from the best deep networks (EfficientNet-b0 and MobileNet-v2) were merged using the canonical correlation approach (CCA). The CCA-based fused features resulted in an enhanced classification performance of 94.7% with a large feature vector size (i.e., 2532). To remove the redundant features from the CCA-based fused feature vector, the binary-enhanced WOA was utilized for optimal feature selection, which yielded an average accuracy of 98.12 ± 0.52 (mean ± standard deviation) with only 953 features. The results were compared with other optimal feature selection techniques, showing that the binary-enhanced WOA results are statistically significant (p < 0.01). The ablation study was also performed to show the significance of each step of the proposed methodology. Furthermore, the comparison shows the superiority and high classification performance of the proposed automated AD detection approach, suggesting that the hybrid approach may help doctors with dementia detection and staging.

Deep learning-based surgical step recognition for laparoscopic right-sided colectomy.

Understanding the complex anatomy and surgical steps involved in laparoscopic right-sided colectomy (LAP-RC) is essential for standardizing the surgical procedure. Deep-learning (DL)-based computer vision can achieve this. This study aimed to develop a step recognition model for LAP-RC using a dataset of surgical videos with annotated step information and evaluate its recognition performance. This single-center retrospective study utilized a video dataset of laparoscopic ileocecal resection (LAP-ICR) and laparoscopic right-sided hemicolectomy (LAP-RHC) for right-sided colon cancer performed between January 2018 and March 2022. The videos were split into still images, which were divided into training, validation, and test sets using 66%, 17%, and 17% of the data, respectively. Videos were manually classified into eight main steps: 1) medial mobilization, 2) central vascular ligation, 3) dissection of the superior mesenteric vein, 4) retroperitoneal mobilization, 5) lateral mobilization, 6) cranial mobilization, 7) mesocolon resection, and 8) intracorporeal anastomosis. In a simpler version, consecutive surgical steps were combined, resulting in five steps. Precision, recall, F1 scores, and overall accuracy were assessed to evaluate the model's performance in the surgical step classification task. Seventy-eight patients were included; LAP-ICR and LAP-RHC were performed in 35 (44%) and 44 (56%) patients, respectively. The overall accuracy was 72.1% and 82.9% for the eight-step and combined five-step classification tasks, respectively. The automatic surgical step-recognition model for LAP-RCs, developed using a DL algorithm, exhibited a fairly high classification performance. A model that understands the complex steps of LAP-RC will aid the standardization of the surgical procedure.

Classification of Garlic Varieties with Fluorescent Spectroscopy Using Machine Learning

Machine learning techniques can produce fast, accurate and objective results in the analysis of agricultural products. These artificial intelligence-based systems are frequently encountered in studies on agriculture in the literature. This study reveals the usability of machine learning algorithms in classification of garlic cultivars using fluorescent spectroscopic data. For this, six types of garlic were used: Razgradski-11, Razgradski-12, Razgradski-115, Plovdivski-120, Yambolski-99 and Topolovgradski. In the first stage, the parsing analysis made from the fluorescent spectroscopic data of the garlics was carried out with seven different machine learning. The classification results of these seven types of machine learning algorithms were obtained. In the second stage, the classification results were obtained by adjusting the hyperparameters of each Machine Learning (ML) algorithm in order to control the improvability of the classification accuracy rates. Finally, performance metrics such as Specificity, precision, MCC, F1-Score of the classification processes obtained in the two stages were compared. In general, it was observed that the classification performances increased with the hyperparameter adjustment performed in the second stage. In this study, classification results with ML showed that fluorescent spectroscopy data of garlic strongly represented garlic species and provided high performance classification accuracy of 99.93% with Neural Network (NN), one of the machine learning methods using these data.

Deep learning-driven ultrasound-assisted diagnosis: optimizing GallScopeNet for precise identification of biliary atresia.

Biliary atresia (BA) is a severe congenital biliary developmental abnormality threatening neonatal health. Traditional diagnostic methods rely heavily on experienced radiologists, making the process time-consuming and prone to variability. The application of deep learning for the automated diagnosis of BA remains underexplored. This study introduces GallScopeNet, a deep learning model designed to improve diagnostic efficiency and accuracy through innovative architecture and advanced feature extraction techniques. The model utilizes data from a carefully constructed dataset of gallbladder ultrasound images. A dataset comprising thousands of ultrasound images was employed, with the majority used for training and validation and a subset reserved for external testing. The model's performance was evaluated using five-fold cross-validation and external assessment, employing metrics such as accuracy and the area under the receiver operating characteristic curve (AUC), compared against clinical diagnostic standards. GallScopeNet demonstrated exceptional performance in distinguishing BA from non-BA cases. In the external test dataset, GallScopeNet achieved an accuracy of 81.21% and an AUC of 0.85, indicating strong diagnostic capabilities. The results highlighted the model's ability to maintain high classification performance, reducing misdiagnosis and missed diagnosis. GallScopeNet effectively differentiates between BA and non-BA images, demonstrating significant potential and reliability for early diagnosis. The system's high efficiency and accuracy suggest it could serve as a valuable diagnostic tool in clinical settings, providing substantial technical support for improving diagnostic workflows.

Detecting cardiac states with wearable photoplethysmograms and implications for out-of-hospital cardiac arrest detection.

Out-of-hospital cardiac arrest (OHCA) is a global health problem affecting approximately 4.4million individuals yearly. OHCA has a poor survival rate, specifically when unwitnessed (accounting for up to 75% of cases). Rapid recognition can significantly improve OHCA survival, and consumer wearables with continuous cardiopulmonary monitoring capabilities hold potential to "witness" cardiac arrest and activate emergency services. In this study, we used an arterial occlusion model to simulate cardiac arrest and investigated the ability of infrared photoplethysmogram (PPG) sensors, often utilized in consumer wearable devices, to differentiate normal cardiac pulsation, pulseless cardiac (i.e., resembling a cardiac arrest), and non-physiologic (i.e., off-body) states. Across the classification models trained and evaluated on three anatomical locations, higher classification performances were observed on the finger (macro average F1-score of 0.964 on the fingertip and 0.954 on the finger base) compared to the wrist (macro average F1-score of 0.837). The wrist-based classification model, which was trained and evaluated using all PPG measurements, including both high- and low-quality recordings, achieved a macro average precision and recall of 0.922 and 0.800, respectively. This wrist-based model, which represents the most common form factor in consumer wearables, could only capture about 43.8% of pulseless events. However, models trained and tested exclusively on high-quality recordings achieved higher classification outcomes (macro average F1-score of 0.975 on the fingertip, 0.973 on the finger base, and 0.934 on the wrist). The fingertip model had the highest performance to differentiate arterial occlusion pulselessness from normal cardiac pulsation and off-body measurements with macro average precision and recall of 0.978 and 0.972, respectively. This model was able to identify 93.7% of pulseless states (i.e., resembling a cardiac arrest event), with a 0.4% false positive rate. All classification models relied on a combination of time-, power spectral density (PSD)-, and frequency-domain features to differentiate normal cardiac pulsation, pulseless cardiac, and off-body PPG recordings. However, our best model represented an idealized detection condition, relying on ensuring high-quality PPG data for training and evaluation of machine learning algorithms. While 90.7% of our PPG recordings from the fingertip were considered of high quality, only 53.2% of the measurements from the wrist passed the quality criteria. Our findings have implications for adapting consumer wearables to provide OHCA detection, involving advancements in hardware and software to ensure high-quality measurements in real-world settings, as well as development of wearables with form factors that enable high-quality PPG data acquisition more consistently. Given these improvements, we demonstrate that OHCA detection can feasibly be made available to anyone using PPG-based consumer wearables.

A comparative study on machine learning approaches for rock mass classification using drilling data

Current rock engineering design in drill and blast tunnelling primarily relies on engineers' observational assessments. Measure While Drilling (MWD) data, a high-resolution sensor dataset collected during tunnel excavation, is underutilised, mainly serving for geological visualisation. This study aims to automate the translation of MWD data into actionable metrics for rock engineering. It seeks to link data to specific engineering actions, thus providing critical decision support for geological challenges ahead of the tunnel face. Leveraging a large and geologically diverse dataset of ∼500,000 drillholes from 15 tunnels, the research introduces models for accurate rock mass quality classification in a real-world tunnelling context. Both conventional machine learning and image-based deep learning are explored to classify MWD data into Q-classes and Q-values—examples of metrics describing the stability of the rock mass—using both tabular- and image data. The results indicate that the K-nearest neighbours algorithm in an ensemble with tree-based models using tabular data effectively classifies rock mass quality. It achieves a cross-validated balanced accuracy of 0.86 in classifying rock mass into the Q-classes A, B, C, D, E1, E2, and 0.95 for a binary classification with E versus the rest. Classification using a CNN with MWD-images for each blasting round resulted in a balanced accuracy of 0.82 for binary classification. Regressing the Q-value from tabular MWD-data achieved cross-validated R2 and MSE scores of 0.80 and 0.18 for a similar ensemble model as in classification. High performance in regression and classification boosts confidence in automated rock mass assessment. Applying advanced modelling on a unique dataset demonstrates MWD data's value in improving rock mass classification accuracy and advancing data-driven rock engineering design, reducing manual intervention.

Virtual Reality-enabled High-Performance Emotion Estimation with the Most Significant Channel Pairs

Human-computer interface (HCI) and electroencephalogram (EEG) signals are widely used in user experience (UX) interface designs to provide immersive interactions with the user. In the context of UX, EEG signals can be used within a metaverse system to assess user engagement, attention, emotional responses, or mental workload. By analyzing EEG signals, system designers can tailor the virtual environment, content, or interactions in real time to optimize UX, improve immersion, and personalize interactions. However, in this case, in addition to the signals' processing cost and classification accuracy, cybersickness in Virtual Reality (VR) systems needs to be resolved. At this point, channel selection methods can perform better for HCI and UX applications by reducing noisy and redundant information in generally unrelated EEG channels. For this purpose, a new method for EEG channel selection based on phase-locking value (PLV) analysis is proposed. We hypothesized that there are interactions between EEG channels in terms of PLV in repeated tasks in different trials of the emotion estimation experiment. Subsequently, frequency-based features were extracted. The features were classified by dividing them into bags using the Multiple-Instance Learning (MIL) variant. This study provides higher classification performance using fewer EEG channels for emotion prediction. The performance rate obtained in binary classification with the Random Forests (RF) algorithm is at a promising level of 99%. The proposed method achieved an accuracy of 99.38% for valence using all channels on the new dataset (VREMO) and 98.13% with channel selection. The benchmark dataset (DEAP) achieved accuracies of 98.16% using all channels and 98.13% with selected channels.

Deep Ring-Block-Wise Network for Hyperspectral Image Classification.

Deep learning has achieved many successes in the field of the hyperspectral image (HSI) classification. Most of existing deep learning-based methods have no consideration of feature distribution, which may yield lowly separable and discriminative features. From the perspective of spatial geometry, one excellent feature distribution form requires to satisfy both properties, i.e., block and ring. The block means that in a feature space, the distance of intraclass samples is close and the one of interclass samples is far. The ring represents that all class samples are overall distributed in a ring topology. Accordingly, in this article, we propose a novel deep ring-block-wise network (DRN) for the HSI classification, which takes full consideration of feature distribution. To obtain the good distribution used for high classification performance, in this DRN, a ring-block perception (RBP) layer is built by integrating the self-representation and ring loss into a perception model. By such way, the exported features are imposed to follow the requirements of both block and ring, so as to be more separably and discriminatively distributed compared with traditional deep networks. Besides, we also design an optimization strategy with alternating update to obtain the solution of this RBP layer model. Extensive results on the Salinas, Pavia Centre, Indian Pines, and Houston datasets have demonstrated that the proposed DRN method achieves the better classification performance in contrast to the state-of-the-art approaches.

LightDPH: Lightweight Dual-Projection-Head Hierarchical Contrastive Learning for Skin Lesion Classification

Effective skin cancer detection is crucial for early intervention and improved treatment outcomes. Previous studies have primarily focused on enhancing the performance of skin lesion classification models. However, there is a growing need to consider the practical requirements of real-world scenarios, such as portable applications that require lightweight models embedded in devices. Therefore, this study aims to propose a novel method that can address the major-type misclassification problem with a lightweight model. This study proposes an innovative Lightweight Dual Projection-Head Hierarchical contrastive learning (LightDPH) method. We introduce a dual projection-head mechanism to a contrastive learning framework. This mechanism is utilized to train a model with our proposed multi-level contrastive loss (MultiCon Loss), which can effectively learn hierarchical information from samples. Meanwhile, we present a distance-based weight (DBW) function to adjust losses based on hierarchical levels. This unique combination of MultiCon Loss and DBW function in LightDPH tackles the problem of major-type misclassification with lightweight models and enhances the model’s sensitivity in skin lesion classification. The experimental results demonstrate that LightDPH significantly reduces the number of parameters by 52.6% and computational complexity by 29.9% in GFLOPs while maintaining high classification performance comparable to state-of-the-art methods. This study also presented a novel evaluation metric, model efficiency score (MES), to evaluate the cost-effectiveness of models with scaling and classification performance. The proposed LightDPH effectively mitigates major-type misclassification and works in a resource-efficient manner, making it highly suitable for clinical applications in resource-constrained environments. To the best of our knowledge, this is the first work that develops an effective lightweight hierarchical classification model for skin lesion detection.

Using random forest and biomarkers for differentiating COVID-19 and Mycoplasma pneumoniaeinfections.

The COVID-19 pandemic has underscored the critical need for precise diagnostic methods to distinguish between similar respiratory infections, such as COVID-19 and Mycoplasma pneumoniae (MP). Identifying key biomarkers and utilizing machine learning techniques, such as random forest analysis, can significantly improve diagnostic accuracy. We conducted a retrospective analysis of clinical and laboratory data from 214 patients with acute respiratory infections, collected between October 2022 and October 2023 at the Second Hospital of Nanping. The study population was categorized into three groups: COVID-19 positive (n = 52), MP positive (n = 140), and co-infected (n = 22). Key biomarkers, including C-reactive protein (CRP), procalcitonin (PCT), interleukin- 6 (IL-6), and white blood cell (WBC) counts, were evaluated. Correlation analyses were conducted to assess relationships between biomarkers within each group. The random forest analysis was applied to evaluate the discriminative power of these biomarkers. The random forest model demonstrated high classification performance, with area under the ROC curve (AUC) scores of 0.86 (95% CI: 0.70-0.97) for COVID-19, 0.79 (95% CI: 0.64-0.92) for MP, 0.69 (95% CI: 0.50-0.87) for co-infections, and 0.90 (95% CI: 0.83-0.95) for the micro-average ROC. Additionally, the precision-recall curve for the random forest classifier showed a micro-average AUC of 0.80 (95% CI: 0.69-0.91). Confusion matrices highlighted the model's accuracy (0.77) and biomarker relationships. The SHAP feature importance analysis indicated that age (0.27), CRP (0.25), IL6 (0.14), and PCT (0.14) were the most significant predictors. The integration of computational methods, particularly random forest analysis, in evaluating clinical and biomarker data presents a promising approach for enhancing diagnostic processes for infectious diseases. Our findings support the use of specific biomarkers in differentiating between COVID-19 and MP, potentially leading to more targeted and effective diagnostic strategies. This study underscores the potential of machine learning techniques in improving disease classification in the era of precision medicine.

Virus Load Kinetics in Lassa Fever Patients Treated With Ribavirin: A Retrospective Cohort Study From Southern Nigeria

Abstract Background The standard of care for Lassa fever is the use of ribavirin with supportive therapy. There is little information on the course of viremia and its relationship with clinical outcomes in patients treated with ribavirin. Methods We conducted a retrospective analysis of virologic and clinical parameters of 152 reverse transcription polymerase chain reaction–confirmed Lassa fever cases admitted and treated with ribavirin therapy. We describe the Lassa virus RNA kinetics in blood in relation to the clinical course of the patients. Results The overall mortality was 9%. The median duration (interquartile range [IQR]) of illness before admission was 8 (5–12) days. Median (IQR) Ct values on admission (t0) were lower among patients who died (21 [20–27]) than in those who survived (34 [30–37]; P < .01). The receiver operating characteristics curve of the association between outcome and Ct value at t0 had a high classification performance, with an AUC of 0.92 (95% CI, 0.86–0.98). The median time to viral clearance (IQR) was 10 (5–15) days. The viral load decreased steadily with the duration of treatment, and all survivors achieved viral clearance within 25 days of hospitalization. Conclusions Our study demonstrates that the Ct value on admission has prognostic value and Lassa fever patients treated with ribavirin typically clear the virus within 3–4 weeks of hospitalization. This kinetics has implications for the design of clinical case management and future clinical trial protocols.

Understanding Learning from EEG Data: Combining Machine Learning and Feature Engineering Based on Hidden Markov Models and Mixed Models.

Theta oscillations, ranging from 4-8 Hz, play a significant role in spatial learning and memory functions during navigation tasks. Frontal theta oscillations are thought to play an important role in spatial navigation and memory. Electroencephalography (EEG) datasets are very complex, making any changes in the neural signal related to behaviour difficult to interpret. However, multiple analytical methods are available to examine complex data structures, especially machine learning-based techniques. These methods have shown high classification performance, and their combination with feature engineering enhances their capability. This paper proposes using hidden Markov and linear mixed effects models to extract features from EEG data. Based on the engineered features obtained from frontal theta EEG data during a spatial navigation task in two key trials (first, last) and between two conditions (learner and non-learner), we analysed the performance of six machine learning methods on classifying learner and non-learner participants. We also analysed how different standardisation methods used to pre-process the EEG data contribute to classification performance. We compared the classification performance of each trial with data gathered from the same subjects, including solely coordinate-based features, such as idle time and average speed. We found that more machine learning methods perform better classification using coordinate-based data. However, only deep neural networks achieved an area under the ROC curve higher than 80% using the theta EEG data alone. Our findings suggest that standardising the theta EEG data and using deep neural networks enhances the classification of learner and non-learner subjects in a spatial learning task.

Lobish: Symbolic Language for Interpreting Electroencephalogram Signals in Language Detection Using Channel-Based Transformation and Pattern.

Electroencephalogram (EEG) signals contain information about the brain's state as they reflect the brain's functioning. However, the manual interpretation of EEG signals is tedious and time-consuming. Therefore, automatic EEG translation models need to be proposed using machine learning methods. In this study, we proposed an innovative method to achieve high classification performance with explainable results. We introduce channel-based transformation, a channel pattern (ChannelPat), the t algorithm, and Lobish (a symbolic language). By using channel-based transformation, EEG signals were encoded using the index of the channels. The proposed ChannelPat feature extractor encoded the transition between two channels and served as a histogram-based feature extractor. An iterative neighborhood component analysis (INCA) feature selector was employed to select the most informative features, and the selected features were fed into a new ensemble k-nearest neighbor (tkNN) classifier. To evaluate the classification capability of the proposed channel-based EEG language detection model, a new EEG language dataset comprising Arabic and Turkish was collected. Additionally, Lobish was introduced to obtain explainable outcomes from the proposed EEG language detection model. The proposed channel-based feature engineering model was applied to the collected EEG language dataset, achieving a classification accuracy of 98.59%. Lobish extracted meaningful information from the cortex of the brain for language detection.

Efficiency of Artificial Intelligence Methods for Hearing Loss Type Classification: an Evaluation

The evaluation of hearing loss is primarily conducted by pure tone audiometry testing, which is often regarded as golden standard for assessing auditory function. If the presence of hearing loss is determined, it is possible to differentiate between three types of hearing loss: sensorineural, conductive, and mixed. This study presents a comprehensive comparison of a variety of AI classification models, performed on 4007 pure tone audiometry samples that have been labeled by professional audiologists in order to develop an automatic classifier of hearing loss type. The tested models include Logistic Regression, Support Vector Machines, Stochastic Gradient Descent, Decision Trees, Random Forest, Feedforward Neural Network (FNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). The presented work also investigates the influence of training dataset augmentation with the use of a Conditional Generative Adversarial Network on the performance of machine learning algorithms and examines the impact of various standardization procedures on the effectiveness of deep learning architectures. Overall, the highest classification performance, was achieved by LSTM with an out-of-training accuracy of 97.56%.

Med-MGF: multi-level graph-based framework for handling medical data imbalance and representation

BackgroundModeling patient data, particularly electronic health records (EHR), is one of the major focuses of machine learning studies in healthcare, as these records provide clinicians with valuable information that can potentially assist them in disease diagnosis and decision-making.MethodsIn this study, we present a multi-level graph-based framework called MedMGF, which models both patient medical profiles extracted from EHR data and their relationship network of health profiles in a single architecture. The medical profiles consist of several layers of data embedding derived from interval records obtained during hospitalization, and the patient-patient network is created by measuring the similarities between these profiles. We also propose a modification to the Focal Loss (FL) function to improve classification performance in imbalanced datasets without the need to imputate the data. MedMGF’s performance was evaluated against several Graphical Convolutional Network (GCN) baseline models implemented with Binary Cross Entropy (BCE), FL, class balancing parameter α\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\alpha$$\\end{document}, and Synthetic Minority Oversampling Technique (SMOTE).ResultsOur proposed framework achieved high classification performance (AUC: 0.8098, ACC: 0.7503, SEN: 0.8750, SPE: 0.7445, NPV: 0.9923, PPV: 0.1367) on an extreme imbalanced pediatric sepsis dataset (n=3,014, imbalance ratio of 0.047). It yielded a classification improvement of 3.81% for AUC, 15% for SEN compared to the baseline GCN+α\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\alpha$$\\end{document}FL (AUC: 0.7717, ACC: 0.8144, SEN: 0.7250, SPE: 0.8185, PPV: 0.1559, NPV: 0.9847), and an improvement of 5.88% in AUC and 22.5% compared to GCN+FL+SMOTE (AUC: 0.7510, ACC: 0.8431, SEN: 0.6500, SPE: 0.8520, PPV: 0.1688, NPV: 0.9814). It also showed a classification improvement of 3.86% for AUC, 15% for SEN compared to the baseline GCN+α\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\alpha$$\\end{document}BCE (AUC: 0.7712, ACC: 0.8133, SEN: 0.7250, SPE: 0.8173, PPV: 0.1551, NPV: 0.9847), and an improvement of 14.33% in AUC and 27.5% in comparison to GCN+BCE+SMOTE (AUC: 0.6665, ACC: 0.7271, SEN: 0.6000, SPE: 0.7329, PPV: 0.0941, NPV: 0.9754).ConclusionWhen compared to all baseline models, MedMGF achieved the highest SEN and AUC results, demonstrating the potential for several healthcare applications.

Stable attractors for neural networks classification via ordinary differential equations (SA-nODE)

Abstract A novel approach for supervised classification is presented which sits at the intersection of machine learning and dynamical systems theory. At variance with other methodologies that employ ordinary differential equations for classification purposes, the untrained model is a priori constructed to accommodate for a set of pre-assigned stationary stable attractors. Classifying amounts to steer the dynamics towards one of the planted attractors, depending on the specificity of the processed item supplied as an input. Asymptotically the system will hence converge on a specific point of the explored multi-dimensional space, flagging the category of the object to be eventually classified. Working in this context, the inherent ability to perform classification, as acquired ex post by the trained model, is ultimately reflected in the shaped basin of attractions associated to each of the target stable attractors. The performance of the proposed method is here challenged against simple toy models crafted for the purpose, as well as by resorting to well established reference standards. More precisely, we achieved an accuracy of 98.06 % on the MNIST test set and 88.21% on the Fashion MNIST test set. Although this method does not reach the performance of state-of-the-art deep learning algorithms, it illustrates that continuous dynamical systems with closed analytical interaction terms can serve as high-performance classifiers.

FBENet: Feature-Level Boosting Ensemble Network for Hashimoto's Thyroiditis Ultrasound Image Classification.

Distinguishing Hashimoto's thyroiditis (HT) lesions from ordinary thyroid tissues is difficult with ultrasound images. Challenges in achieving high performance of HT ultrasound image classification include the low resolution, blurred features and large area of irrelevant noise. To address these problems, we propose a Feature-level Boosting Ensemble Network (FBENet) for HT ultrasound image classification. Specifically, to capture the features of suspicious HT lesions efficiently, an Ensemble Feature Boosting Module (EFBM) is introduced into the feature-level ensemble to boost the blurred features. Then, the spatial attention mechanism is adopted in backbone models to improve the feature focusing performance and representation ability. Furthermore, feature-level ensemble technique is employed in the training process to achieve more comprehensive feature representation ability. Experimentally, FBENet was trained on 6,503 HT ultrasound images, and tested on 1,626 HT ultrasound images with 82.92% accuracy and 89.24% AUC on average.

LearningTuple: A packet classification scheme with high classification and high update

Packet classification is widely used in network infrastructures and is the key technique that supports security and other functions. The real-time nature of network services naturally demands high classification speed, while the emerging SDN makes rule changes more flexible, thus placing higher demands on the performance of rule update in classification schemes. In this paper, Learning Tuple(LT) is proposed to achieve high classification performance for packets while maintaining the high update characteristics of tuple space-based schemes. Specifically, to solve the issue of excessive tuples and rule overlap due to merging tuples, LT iteratively divides the space by using rule overlap and hash collisions as negative feedback and applies a reinforcement learning algorithm, SARSA, at each level to ensure its reasonableness. Efficient space partitioning guides the construction of tuples, and an excellent rule mapping method called PLR is designed, which improves classification performance. Experimental results demonstrate that compared with classic and advanced classification schemes TSS, TupleMerge, MultilayerTuple, PartitionSort, HybridTSS, and TupleTree, LT achieves average classification performance improvements of 9.23x, 1.74x, 1.45x, 2.85x, 1.37x and 1.25x, as well as average update performance improvements of 1.83x, 6.75x, 1.22x, 6.16x, 1.21x, 10.66x, respectively.

SFT-SGAT: A semi-supervised fine-tuning self-supervised graph attention network for emotion recognition and consciousness detection

Emotional recognition is highly important in the field of brain-computer interfaces (BCIs). However, due to the individual variability in electroencephalogram (EEG) signals and the challenges in obtaining accurate emotional labels, traditional methods have shown poor performance in cross-subject emotion recognition. In this study, we propose a cross-subject EEG emotion recognition method based on a semi-supervised fine-tuning self-supervised graph attention network (SFT-SGAT). First, we model multi-channel EEG signals by constructing a graph structure that dynamically captures the spatiotemporal topological features of EEG signals. Second, we employ a self-supervised graph attention neural network to facilitate model training, mitigating the impact of signal noise on the model. Finally, a semi-supervised approach is used to fine-tune the model, enhancing its generalization ability in cross-subject classification. By combining supervised and unsupervised learning techniques, the SFT-SGAT maximizes the utility of limited labeled data in EEG emotion recognition tasks, thereby enhancing the model’s performance. Experiments based on leave-one-subject-out cross-validation demonstrate that SFT-SGAT achieves state-of-the-art cross-subject emotion recognition performance on the SEED and SEED-IV datasets, with accuracies of 92.04% and 82.76%, respectively. Furthermore, experiments conducted on a self-collected dataset comprising ten healthy subjects and eight patients with disorders of consciousness (DOCs) revealed that the SFT-SGAT attains high classification performance in healthy subjects (maximum accuracy of 95.84%) and was successfully applied to DOC patients, with four patients achieving emotion recognition accuracies exceeding 60%. The experiments demonstrate the effectiveness of the proposed SFT-SGAT model in cross-subject EEG emotion recognition and its potential for assessing levels of consciousness in patients with DOC.

Machine-learning models for shoulder rehabilitation exercises classification using a wearable system.

The objective of this study is to train and test machine-learning (ML) models to automatically classify shoulder rehabilitation exercises. The cohort included both healthy and patients with rotator-cuff (RC) tears. All participants performed six shoulder rehabilitation exercises, following guidelines developed by the American Society of Shoulder and Elbow Therapists. Each exercise was repeated six times, while wearing a wearable system equipped with three magneto-inertial sensors. Six supervised machine-learning models (k-Nearest Neighbours, Support Vector Machine, Decision Tree, Random Forest (RF), Logistic Regression and Adaptive Boosting) were trained for the classification. The algorithms' ability to accurately classify exercise activities was evaluated using the nested cross-validation method, with different combinations of outer and inner folds. A total of 19 healthy subjects and 17 patients with complete RC tears were enroled in the study. The highest classification performances were achieved by the RF classifier, with an accuracy of 89.91% and an F1-score of 89.89%. The results of this study highlight the feasibility and effectiveness of using wearable sensors and ML algorithms to accurately classify shoulder rehabilitation exercises. These findings suggest promising prospects for implementing the proposed wearable system in remote home-based monitoring scenarios. The ease of setup and modularity of the system reduce user burden enabling patient-driven sensor positioning. Level III.