False Negative Ratio Research Articles

Evaluation of tooth development stages with deep learning-based artificial intelligence algorithm

BackgroundThis study aims to evaluate the performance of a deep learning system for the evaluation of tooth development stages on images obtained from panoramic radiographs from child patients.MethodsThe study collected a total of 1500 images obtained from panoramic radiographs from child patients between the ages of 5 and 14 years. YOLOv5, a convolutional neural network (CNN)-based object detection model, was used to automatically detect the calcification states of teeth. Images obtained from panoramic radiographs from child patients were trained and tested in the YOLOv5 algorithm. True-positive (TP), false-positive (FP), and false-negative (FN) ratios were calculated. A confusion matrix was used to evaluate the performance of the model.ResultsAmong the 146 test group images with 1022 labels, there were 828 TPs, 308 FPs, and 1 FN. The sensitivity, precision, and F1-score values of the detection model of the tooth stage development model were 0.99, 0.72, and 0.84, respectively.ConclusionsIn conclusion, utilizing a deep learning-based approach for the detection of dental development on pediatric panoramic radiographs may facilitate a precise evaluation of the chronological correlation between tooth development stages and age. This can help clinicians make treatment decisions and aid dentists in finding more accurate treatment options.

An Improved Adaptive Neuro-fuzzy Inference Framework for Lung Cancer Detection and Prediction on Internet of Medical Things Platform

It has become increasingly difficult for medical practitioners to recognize illness in recent years due to the emergence of new diseases from their myriad causes on a daily basis. Due in large part to inadequate diagnostic and monitoring infrastructure, a substantial amount of illness and death are associated with lung cancer (LC). The aim of the paper is to find lung cancer early and help patients receive curative treatment. Quitting smoking or never starting is the best way to mitigate the potential for disease-related death. As a result, cutting-edge detection and monitoring technologies must be developed to enable rapid, accurate, and timely diagnosis. Fuzzy logic (FL) is one of the best approaches to modeling complex and uncertain systems; therefore, it helps us deal with these challenges. Fuzzy expert system for lung cancer [FES-LC] detection and prediction on Internet of medical things (IoMT) is employed to overcome the challenges. Hence, an enhanced adaptive neuro-fuzzy inference framework [ANF-IF] is proposed in the current research. The cloud-based application of an adaptive neuro-fuzzy inference system yields four risk categories: not at risk, slightly at risk, moderately at risk, and severely at risk. New methods and theoretical frameworks have made it possible to diagnose LC in its earliest stages with the help of magnetic nanoparticles (MNPs), which allow researchers to overcome the limitations of conventionally slow diagnostic efficiency. The proposed system exhibits a precision of 93.4%, accuracy of 95.1%, specificity of 90.6%, sensitivity of 92.8%, false positive rate of 0.22%, false negative ratio of 0.18%, and classification accuracy of 98.2%. The proposed method outperforms all methods and provides better lung cancer detection accuracy than others.

External quality assessments for SARS-CoV-2 genome detection in Austria : Acomparison of the first postpandemic round with results from the pandemic era.

External quality assessment (EQA) schemes provide objective feedback to participating laboratories about the performance of their analytical systems and information about overall regional analytical performance. The EQAs are particularly important during pandemics as they also assess the reliability of individual test results and show opportunities to improve test strategies. With the end of the COVID-19 pandemic, the testing frequency significantly decreased in Austria. Here, we analyzed whether this decrease had an effect on participation and/or performance in SARS-CoV‑2 virus detection EQAs, as compared to the pandemic era. Identical samples were sent to all participating laboratories, and the EQA provider evaluated the agreement of the reported results with defined targets. The EQA was operated under two schemes with identical samples and therefore we analyzed it as asingle EQA round. The performance of testing was reported as true positive ratios, comparing the post-pandemic data to previous rounds. Furthermore, subgroups of participants were analyzed stratified by laboratory type (medical or nonmedical) and the test system format (fully automated or requiring manual steps). While the frequency of false negative results per sample did not change during the 3 years of the pandemic (5.7%, 95% confidence interval [CI] 3.1-8.4%), an average per sample false negative ratio of 4.3% was observed in the first post-pandemic EQA (0%, 1.8%, and 11% for the 3 positive samples included in the test panel, n = 109 test results per sample). In this first post-pandemic EQA medical laboratories (average 0.4% false negative across 3samples, n = 90) and automated test systems (average 1.2% false negative, n = 261) had lower false negative ratios than nonmedical laboratories (22.8%, n = 19) and manual test systems (16.7%, n = 22). These lower average ratios were due to alow concentration sample, where nonmedical laboratories reported 36.8% and manual test systems 54.5% true positive results. Overall ratios of true positive results were below the mean of all results during the pandemic but were similar to the first round of the pandemic. Alower post-pandemic true positive ratio was associated with specific laboratory types and assay formats, particularly for samples with low concentration. The EQAs will continue to monitor the laboratory performance to ensure the same quality of epidemiological data after the pandemic, even if vigilance has decreased.

Detection of Ulcerative Colitis Lesions from Weakly Annotated Colonoscopy Videos Using Bounding Boxes

Ulcerative colitis is a chronic disease characterized by bleeding and ulcers in the colon. Disease severity assessment via colonoscopy videos is time-consuming and only focuses on the most severe lesions. Automated detection methods enable fine-grained assessment but depend on the training set quality. To suit the local clinical setup, an internal training dataset containing only rough bounding box annotations around lesions was utilized. Following previous works, we propose to use linear models in suitable color spaces to detect lesions. We introduce an efficient sampling scheme for exploring the set of linear classifiers and removing trivial models i.e., those showing zero false negative or positive ratios. Bounding boxes lead to exaggerated false detection ratios due to mislabeled pixels, especially in the corners, resulting in decreased model accuracy. Therefore, we propose to evaluate the model sensitivity on the annotation level instead of the pixel level. Our sampling strategy can eliminate up to 25% of trivial models. Despite the limited quality of annotations, the detectors achieved better performance in comparison with the state-of-the-art methods. When tested on a small subset of endoscopic images, the best models exhibit low variability. However, the inter-patient model performance was variable suggesting that appearance normalization is critical in this context.

Skin Cancer Segmentation and Classification Using Vision Transformer for Automatic Analysis in Dermatoscopy-Based Noninvasive Digital System.

Skin cancer is a significant health concern worldwide, and early and accurate diagnosis plays a crucial role in improving patient outcomes. In recent years, deep learning models have shown remarkable success in various computer vision tasks, including image classification. In this research study, we introduce an approach for skin cancer classification using vision transformer, a state-of-the-art deep learning architecture that has demonstrated exceptional performance in diverse image analysis tasks. The study utilizes the HAM10000 dataset; a publicly available dataset comprising 10,015 skin lesion images classified into two categories: benign (6705 images) and malignant (3310 images). This dataset consists of high-resolution images captured using dermatoscopes and carefully annotated by expert dermatologists. Preprocessing techniques, such as normalization and augmentation, are applied to enhance the robustness and generalization of the model. The vision transformer architecture is adapted to the skin cancer classification task. The model leverages the self-attention mechanism to capture intricate spatial dependencies and long-range dependencies within the images, enabling it to effectively learn relevant features for accurate classification. Segment Anything Model (SAM) is employed to segment the cancerous areas from the images; achieving an IOU of 96.01% and Dice coefficient of 98.14% and then various pretrained models are used for classification using vision transformer architecture. Extensive experiments and evaluations are conducted to assess the performance of our approach. The results demonstrate the superiority of the vision transformer model over traditional deep learning architectures in skin cancer classification in general with some exceptions. Upon experimenting on six different models, ViT-Google, ViT-MAE, ViT-ResNet50, ViT-VAN, ViT-BEiT, and ViT-DiT, we found out that the ML approach achieves 96.15% accuracy using Google's ViT patch-32 model with a low false negative ratio on the test dataset, showcasing its potential as an effective tool for aiding dermatologists in the diagnosis of skin cancer.

Predicting pedestrian-involved crash severity using inception-v3 deep learning model

This research leverages a novel deep learning model, Inception-v3, to predict pedestrian crash severity using data collected over five years (2016–2021) from Louisiana. The final dataset incorporates forty different variables related to pedestrian attributes, environmental conditions, and vehicular specifics. Crash severity was classified into three categories: fatal, injury, and no injury. The Boruta algorithm was applied to determine the importance of variables and investigate contributing factors to pedestrian crash severity, revealing several associated aspects, including pedestrian gender, pedestrian and driver impairment, posted speed limits, alcohol involvement, pedestrian age, visibility obstruction, roadway lighting conditions, and both pedestrian and driver conditions, including distraction and inattentiveness. To address data imbalance, the study employed Random Under Sampling (RUS) and the Synthetic Minority Oversampling Technique (SMOTE). The DeepInsight technique transformed numeric data into images. Subsequently, five crash severity prediction models were developed with Inception-v3, considering various scenarios, including original, under-sampled, over-sampled, a combination of under and over-sampled data, and the top twenty-five important variables. Results indicated that the model applying both over and under sampling outperforms models based on other data balancing techniques in terms of several performance metrics, including accuracy, sensitivity, precision, specificity, false negative ratio (FNR), false positive ratio (FPR), and F1-score. This model achieved prediction accuracies of 93.5%, 77.5%, and 85.9% for fatal, injury, and no injury categories, respectively. Additionally, comparative analysis based on several performance metrics and McNemar’s tests demonstrated that the predictive performance of the Inception-v3 deep learning model is statistically superior compared to traditional machine learning and statistical models. The insights from this research can be effectively harnessed by safety professionals, emergency service providers, traffic management centers, and vehicle manufacturers to enhance their safety measures and applications.

Proposed imaging assessment score for aneurysmal subarachnoid hemorrhage correlated with prognosis: Shinshu Aneurysmal subarachnoid hemorrhage score

BackgroundAneurysmal subarachnoid hemorrhage (aSAH) imaging has been shown to correlate with prognosis. However, no numerical index of bleeding severity has been established. This study aimed to propose a new simple scoring system for computed tomography imaging of aSAH and to confirm its effectiveness in retrospective and prospective studies. MethodsWe devised an image evaluation system as an objective index. This system was established by scoring six items, with a maximum total of 19 points. Using this score, named the Shinshu Aneurysmal Subarachnoid Hemorrhage Score (S-score), we performed a retrospective study of 210 patients with aSAH at a single institution to confirm its efficacy. Age and World Federation of Neurosurgical Societies grades were adopted as other verification items, and the modified Rankin Scale was used for prognostic evaluation. A multicenter prospective study was then conducted to examine the function of the score by examining 214 patients with aSAH. ResultsIn the retrospective study, the threshold of the S-score between good and poor prognoses was 9/19 points. The area under the curve by receiver operating characteristic analysis of the S-score was 0.819, suggesting efficacy, with an odds ratio (OR) of 1.291 (1.077–1.547). In the prospective study, the judgment capability of the S-score was evaluated with a sensitivity of 0.674, specificity of 0.881, positive predictive value of 0.789, negative predictive value of 0.804, false-positive ratio of 0.119, false-negative ratio of 0.325, positive likelihood ratio of 6.072, and negative likelihood ratio of 1.369. S-score showed a significant difference in prognosis. The OR was 1.183 (1.009–1.388). ConclusionsThe scoring system could contribute to patient prognosis assessment. S-score and its prognostic formulas may serve as an objective source of information in the development of clinical medicine.

Gradient bald vulture optimization enabled multi-objective Unet++ with DCNN for prostate cancer segmentation and detection

Prostate cancer (PCa) represents the general type of cancer and is considered the third leading reason of death worldwide. As a combined part of computer-aided detection (CAD) applications, magnetic resonance imaging (MRI) is extensively studied for the precise detection of PCa. However, various issues rely on MRI, which includes the complexity of interpretation and increased time. Thus, deep learning-based tumor detection and segmentation methods attempt to be imperative techniques for radiologists to execute their tasks more precisely. The objective is to present Gradient Bald vulture optimization (GBVO)-based Deep Convolution Neural Networks (DCNN) with U-Net++ for segmenting and detecting prostate cancer. Initially, image pre-processing is done using Non-Local Means (NLM) filter. After pre-processing, image segmentation is carried out using the proposed optimized multi-objective Unet++, where the objective function in Unet++ is modified using pixel-wise cross entropy and the Jaccard coefficient. In addition, the training of Unet++ is done using the newly designed Gradient Bald Eagle Optimization (GBEO), which is a combination of Stochastic Gradient Descent (SGD) and Bald Eagle optimization (BEO). Finally, cancer detection is done using DCNN. DCNN is trained using GBVO, which is the integration of the African Vultures Optimization Algorithm (AVOA) and GBEO. The proposed method outperformed state-of-the-art techniques. The developed method achieved the highest accuracy of 0.916, a false negative ratio (FNR) of 0.104, a false positive ratio (FPR) of 0.100, and a negative predictive value (NPV) of 0.903.

Earthquake Forecasting Using Optimized Levenberg–marquardt Back-propagation Neural Network

In this study, an effective earthquake forecasting model is introduced using a hybrid metaheuristic machine learning (ML) algorithm with CUDA-enabled parallel processing. To improve the performance and accuracy of the model, a novel hybrid ML model is developed that utilizes parallel processing. The model consists of a Chaotic Chimp based African Vulture Optimization Algorithm (CCAVO) for feature selection and a Hybrid Levenberg-Marquardt Back-Propagation Neural Network (HLMt-BPNN) for prediction. The proposed model follows a four-step process: preprocessing the raw data to identify seismic indications, extracting features from the preprocessed data, using optimized ML algorithms to forecast the earthquake and its expected time, epicenter, and magnitude, and implementing the model using the Python platform. The model's performance is evaluated using various criteria, including accuracy, precision, recall, F-measure, specificity, false negative ratio, false positive ratio, negative prediction value, Matthew’s correlation coefficient, root mean square error, mean absolute error, and mean absolute percentage error. The proposed model achieved an accuracy of 98%, which is higher than the accuracy of existing earthquake prediction methods.

YOLOv5 Drone Detection Using Multimodal Data Registered by the Vicon System.

This work is focused on the preliminary stage of the 3D drone tracking challenge, namely the precise detection of drones on images obtained from a synchronized multi-camera system. The YOLOv5 deep network with different input resolutions is trained and tested on the basis of real, multimodal data containing synchronized video sequences and precise motion capture data as a ground truth reference. The bounding boxes are determined based on the 3D position and orientation of an asymmetric cross attached to the top of the tracked object with known translation to the object's center. The arms of the cross are identified by the markers registered by motion capture acquisition. Besides the classical mean average precision (mAP), a measure more adequate in the evaluation of detection performance in 3D tracking is proposed, namely the average distance between the centroids of matched references and detected drones, including false positive and false negative ratios. Moreover, the videos generated in the AirSim simulation platform were taken into account in both the training and testing stages.

Effectiveness of one-shot dual-energy subtraction chest radiography with flat-panel detector in distinguishing between calcified and non-calcified nodules

The purpose of this study was to evaluate the added value of the soft tissue image obtained by the one-shot dual-energy subtraction (DES) method using a flat-panel detector compared with the standard image alone in distinguishing calcified from non-calcified nodules on chest radiographs. We evaluated 155 nodules (48 calcified and 107 non-calcified) in 139 patients. Five radiologists (readers 1 − 5) with 26, 14, 8, 6 and 3 years of experience, respectively, evaluated whether the nodules were calcified using chest radiography. CT was used as the gold standard of calcification and non-calcification. Accuracy and area under the receiver operating characteristic curve (AUC) were compared between analyses with and without soft tissue images. The misdiagnosis ratio (false positive plus false negative ratios) when nodules and bones overlapped was also examined. The accuracy of all radiologists increased after adding soft tissue images (readers 1 − 5: 89.7% vs. 92.3% [P = 0.206], 83.2% vs. 87.7% [P = 0.178], 79.4% vs. 92.3% [P < 0.001], 77.4% vs. 87.1% [P = 0.007], and 63.2% vs. 83.2% [P < 0.001], respectively). AUCs for all the readers improved, except for reader 2 (readers 1 − 5: 0.927 vs. 0.937 [P = 0.495], 0.853 vs. 0.834 [P = 0.624], 0.825 vs. 0.878 [P = 0.151], 0.808 vs. 0.896 [P < 0.001], and 0.694 vs. 0.846 [P < 0.001], respectively). The misdiagnosis ratio for nodules that overlapped with the bone decreased after adding soft tissue images in all readers (11.5% vs. 7.6% [P = 0.096], 17.6% vs. 12.2% [P = 0.144], 21.4% vs. 7.6% [P < 0.001], 22.1% vs. 14.5% [P = 0.050] and 35.9% vs. 16.0% [P < 0.001], respectively), particularly that of readers 3 − 5. In conclusion, the soft tissue images obtained using one-shot DES with a flat-panel detector have added value in distinguishing calcified from non-calcified nodules on chest radiographs, especially for less experienced radiologists.

Using Federated Learning for Collaborative Intrusion Detection Systems

Neural networks have become cutting edge machine learning models for detecting network attacks. Traditional implementations provide fast and accurate predictions, but require centralised storage of labelled historical data for training. This solution is not always suitable for real-world applications, where regulatory constraints and privacy concerns hamper the collection of sensitive data into a single server. Federated Learning has recently been proposed as a framework for training a centralised model without the need to share data between different providers. We use the CICIDS2017 dataset provided by the Canadian Institute of Cybersecurity to demonstrate the benefits of Neural Networks-based Federated Learning for the detection of the most relevant types of network attacks. We conclude that a federated-trained neural network outperforms locally-trained models (at isoarchitecture) in terms of F1-score and False Negative detection ratio. Further, such model has a minor loss of performance and convergence rapidity compared to a model trained over a hypothetical centralised dataset.

Secured mobile IOT ecosystem using Enhanced Multi-level Intelligent Trust Scheme

The Internet of Things (IoT) is a collection of heterogeneous, interconnected mobile ad-hoc networks (MANETs) devices. High-level security is always required to protect sensitive information travelling over MANET-IoT networks. IoT security ensures the privacy, confidentiality, and security of the users, infrastructure, and data. The Enhanced Multi-Level Intelligent Trust Scheme (EMITS) initially designs to authenticate and validate the data for the Mobile IoT (MIoT) ecosystem. The proposed model used the AODV (Adhoc On-Demand Distance Vector Routing) routing protocol with an intelligent trust scheme. This paper aims to present a MANET-IoT-based model combined with a hybrid cryptographic scheme that ensures secure communication between mobile IoT ecosystems against the black hole attack. The performance of the proposed EMITS-MIoT model is verified based on the packet delivery ratio (PDR), packet loss ratio (PLR), false positive ratio (FPR), false negative ratio (FNR), and attack detection rate (DR). The EMTS-IoT technique shows a 3–5% success in data transmission over existing techniques.

Preparation of a shared chip for cell capturing and imaging with nanoparticle scaffold, nanogap and circulating tumor cell detection.

Aim: To achieve accurate detection of circulating tumor cells (CTCs) in a large-volume sample. Materials & methods: Silica nanoparticles were crosslinked layer-by-layer on glass slides as the substrate of a chip using polyacrylic acid. Polyacrylic acid was immobilized as a spacerand capture ligands were immobilized on the spacer. Results: The chip can be integrally applied to capture, post-treatment and imaging detection for CTCs. The detected cell numbers were 33 and 40 for 9cell/ml samples and clinical blood samples (7.5ml), respectively. The detection ratio of positive samples was 100%. Conclusion: The significantly increased detected number for CTCs indicates that this methodology mayavoid or greatly reduce the false-negative ratio of positive clinical samples.

Learning from undercoded clinical records for automated International Classification of Diseases (ICD) coding.

To develop an unbiased objective for learning automatic coding algorithms from clinical records annotated with only partial relevant International Classification of Diseases codes, as annotation noise in undercoded clinical records used as training data can mislead the learning process of deep neural networks. We use Medical Information Mart for Intensive Care III as our dataset. We employ positive-unlabeled learning to achieve unbiased loss estimation, which is free of misleading training signal. We then utilize reweighting mechanism to compensate for the imbalance between positive and negative samples. To further close the performance gap caused by poor quality annotation, we integrate the supervision provided by the automatic annotation tool Medical Concept Annotation Toolkit which can ease the heavy burden of manual validation. Our benchmarking results show that positive-unlabeled learning with reweighting outperforms competitive baseline methods over a range of missing label ratios. Integrating supervision provided by annotation tool further boosted the performance. Considering the annotation noise and severe imbalance, unbiased loss estimation and reweighting mechanism are both important for learning from undercoded clinical records. Unbiased loss requires the estimation of false negative ratios and estimation through trained models is practical and competitive. The combination of positive-unlabeled learning with reweighting and supervision provided by the annotation tool is a promising solution to learn from undercoded clinical records.

False Negative Awareness in Indicator-Based Caching Systems

Distributed caching systems such as content distribution networks often advertise their content via lightweight approximate indicators (e.g., Bloom filters) to efficiently inform clients where each datum is likely cached. While false-positive indications are necessary and well understood, most existing works assume no false-negative indications. Our work illustrates practical scenarios where false-negatives are unavoidable and ignoring them significantly impacts system performance. Specifically, we focus on false-negatives induced by indicator staleness, which arises whenever the system advertises the indicator only periodically, rather than immediately reporting every change in the cache. Such scenarios naturally occur, e.g., in bandwidth-constraint environments or when latency impedes each client’s ability to obtain an updated indicator. Our work introduces novel false-negative aware access policies that continuously estimate the false-negative ratio and sometimes access caches despite negative indications. We present optimal policies for homogeneous settings and provide approximation guarantees for our algorithms in heterogeneous environments. We further perform an extensive simulation study with multiple real system traces. We show that our false-negative aware algorithms incur a significantly lower service cost than existing approaches or match the cost of these approaches while requiring an order of magnitude fewer resources (e.g., caching capacity or bandwidth).

Development and Validation of a Risk Assessment Model for Pulmonary Nodules Using Plasma Proteins and Clinical Factors

Deficiencies in risk assessment for patients with pulmonary nodules (PNs) contribute to unnecessary invasive testing and delays in diagnosis. What is the accuracy of a novel PN risk model that includes plasma proteins and clinical factors? How does the accuracy compare with that of an established risk model? Based on technology using magnetic nanosensors, assays were developed with seven plasma proteins. In a training cohort (n= 429), machine learning approaches were used to identify an optimal algorithm that subsequently was evaluated in a validation cohort (n= 489), and its performance was compared with the Mayo Clinic model. In the training set, we identified a support vector machine algorithm that included the seven plasma proteins and six clinical factors that demonstrated an area under the receiver operating characteristic curve of 0.87 and met other selection criteria. The resulting risk reclassification model (RRM) was used to recategorize patients with a pretest risk of between 10%and 84%, and its performance was assessed across five risk strata (low,≤ 10%; moderate, 10%-34%; intermediate, 35%-70%; high, 71%-84%; very high, > 85%). Stratification by the RRM decreased the proportion of intermediate-risk patients from 26.7%to 10.8%(P< .001) and increased the low-risk and high-risk strata from 16.8%to 21.9%(P< .001) and from 3.7%to 12.1%(P< .001), respectively. Among patients classified as low risk by the RRM and Mayo Clinic model, the corresponding true-negative to false-negative ratios were 16.8 and 19.5, respectively. Among patients classified as very high risk by the RRM and Mayo Clinic model, the corresponding true-positive to false-positive ratios were 28.5 and 17.0, respectively. Compared with the Mayo Clinic model, the RRM provided higher specificity at the low-risk threshold and higher sensitivity at the very high-risk threshold. The RRM accurately reclassified some patients into low-risk and very high-risk categories, suggesting the potential to improve PN risk assessment.

LSTM-CNN model of drowsiness detection from multiple consciousness states acquired by EEG

This study aimed to design a deep neural network for electroencephalography (EEG)-based drowsiness detection in multiple consciousness states, i.e., “awake,” “sleep,” and “drowsiness.” Few studies have seriously considered the optimal input vector size or labeling method in classifying multiple consciousness states, which may affect classification performance. To determine the optimal input vector length, i.e., window length, three neural network models (long short-term memory [LSTM], convolutional neural network [CNN], and combined LSTM and CNN) and four feature-based models were tested with six different levels of window length. The EEG dataset was acquired from 19 participants with randomly assigned auditory stimuli and button responses. The EEG data were labeled into three classes (awake, sleep, and drowsiness) based on the defined button response pattern corresponding to the stimuli. The results demonstrated that when the input vector size exceeded 8 sec, the performance of the neural network models dropped rapidly; however, when the window size was less than 8 sec, the performance change according to the window size was small. In contrast, the performance of feature-based models increased continuously as the window size increased. The LSTM model yielded the best accuracy (86%) for a 1 sec window length, and the LSTM-CNN model yielded the best kappa index (0.77) for a 4 sec window length. In addition, the proposed model was applied to the binary classification of normal consciousness (awake) and low consciousness (drowsiness and sleep) states to determine whether this model works appropriately in actual applications such as drowsiness detection in a driving environment. For binary classification, the LSTM-CNN model resulted in 0.95 F1 scores in 4000-ms. When a short input data (500 msec) is used, the LSTM-CNN model resulted in an average accuracy of 85.6% and a kappa index of 0.77 for the three-class classification problem and 0.94 F1 scores for the binary classification problem. In conclusion, we demonstrated that the proposed model could effectively detect drowsiness. Furthermore, a significant correlation was found between reaction time and drowsiness. However, using the reaction time as an index for labeling drowsiness was challenging because of the high false-negative ratio.

A multi-level weighted concept drift detection method

The concept drift detection method is an online learner. Its main task is to determine the position of drifts in the data stream, so as to reset the classifier after detecting the drift to improve the learning performance, which is very important in practical applications such as user interest prediction or financial transaction fraud detection. Aiming at the inability of existing drift detection methods to balance the detection delay, false positives, false negatives, and space–time efficiency, a new level transition threshold parameter is proposed, and a multi-level weighted mechanism including "Stable Level-Warning Level-Drift Level" is innovatively introduced in the concept drift detection. The instances in the window are weighted in levels, and the double sliding window is also applied. Based on this, a multi-level weighted drift detection method (MWDDM) is proposed. In particular, two variants which are MWDDM_H and MWDDM_M are proposed based on Hoeffding inequality and Mcdiarmid inequality, respectively. Experiments on artificial datasets show that MWDDM_H and MWDDM_M can detect abrupt and gradual concept drift faster than other comparison algorithms while maintaining a low false positive ratio and false negative ratio. Experiments on real-world datasets show that MWDDM has the highest classification accuracy in most cases while maintaining good space time efficiency.

Eavesdropping Detection in BB84 Quantum Key Distribution Protocols

The nature of quantum mechanics provides us with an opportunity to statistically detect eavesdropping in quantum key distribution (QKD) protocols, which is unimaginable in classical digital communications. By utilizing Hoeffding’s inequality, this study analyzes the upper bounds of the false-positive ratio (FPR) and false-negative ratio (FNR) of eavesdropping detection in the Bennett–Brassard-84 (BB84) QKD protocol, where eavesdropping is detected if the measured quantum bit error rate (QBER) is equal to or higher than a threshold. The analysis clarifies the trade-off between the accuracy of eavesdropping detection and the economy of quantum resources in the BB84 protocol. Owing to the central limit theorem, the QBER measured by 300 quantum bits (qubits) is sufficient to guarantee lower than 0.009% of the FPR and FNR of eavesdropping detection. To deal with rapidly varying quantum channel conditions, this study further introduces grouped BB84 protocol and combinatory eavesdropping detection algorithms. A polarization basis is changeable for a group of qubits, and eavesdropping is judged by a combination of criteria between QBER and group-QBER in the proposed protocol and algorithms. In our extensive simulation study, the grouped BB84 protocol with 300 qubits comparison guarantees at least 99.92% accuracy in eavesdropping detection under rapidly varying quantum channel conditions.