Average F1 Score Research Articles

Improving the risk profile of Indonesian enterprise taxpayers using multilabel classification

Optimizing tax revenues is difficult in Indonesia due to obstacles such as tax evasion and tax avoidance. It is closely related to an organization's compliance with tax regulations, known as the taxpayers risk profile. However, this mechanism does not accurately detect tax avoidance and tax evasion risks. To overcome this limitation, we use a multilabel classification machine learning method in this study, which classifies a single observation into one or more labels at once. The approach involves problem transformation (binary relevance and label powerset), algorithm adaptation (multilabel k-nearest neighbor (ML-kNN) and multilabel-adaptive resonance associative map (ML ARAM)), and ensemble (label space partitioning and random k-label sets with disjoint (RAkELd)). Based on the model performance comparisons, we discovered that the ML-ARAM method based on deep learning is the best, with an average F1-score of 95.5% and a hamming loss of 7.4%. We also examine the feature importance of the best model to reduce the dimensions of features so that we can identify the dominant factors that encourage a taxpayer entity to engage in tax avoidance or tax evasion. The findings of this study improve the accuracy of tax avoidance risk detection and tax evasion risk profiles using machine learning methods, ensuring maximum tax revenues in Indonesia.

Multichannel EMG-based gesture recognition utilizing advanced machine learning techniques: A random forest classifier for high-precision signal classification

This research examines how advanced machine learning algorithms can be used to classify multichannel electromyographic (EMG) signals with a high level of accuracy to assist in recognizing hand gestures. The goal is to create a robust and scalable system for gesture-based virtual control using EMG signals with potential applications in assistive technologies, rehabilitation, and human-computer interaction. Data were gathered using a MYO Thalmic bracelet containing eight EMG sensors on thirty-six subjects, and a Random Forest classifier was trained to identify seven distinct types of hand gestures (rest, fist clench, wrist flexion/extension, and radial/ulnar deviations). The machine learning pipeline included extensive preprocessing (i.e., EMG signal normalization and signal feature extraction; root mean square, waveform length, and zero crossing rate) and several hyperparameter tuning procedures to improve model performance. The Random Forest model (100 decision trees) achieved an overall classification accuracy of 98.68%, with a range of accuracies for each class (e.g., 95.2% wrist flexion and 91.8% ulnar deviation) when evaluated using cross-validation (i.e., average F1-score = 0.92, precision = 0.94, recall = .91). Overall, the study provides strong evidence for the effectiveness of ensemble learning methods at analyzing complex, multidimensional EMG signals. The high classification accuracy reported, in particular, demonstrates that the system could function for real-time recognition of hand gestures in a virtual environment. Ultimately, the initial work sets the stage for future exploration of a model that may be integrated with actuation models to control prosthetic limbs, virtual actors/avatars, and robotic devices. By demonstrating a scalable and efficient method of gesture recognition using EMG signals, these early findings enable future pathways and possibilities to design innovative, assistive solutions for digital systems that increase accessibility and interaction for users who are motor impaired or have a limited range of motion.

Enhancing Cardiovascular Disease Prediction Accuracy through an Ensemble Machine Learning Approach

This study explores the efficacy of an ensemble machine learning approach, specifically a Voting Classifier combining Decision Tree, k-Nearest Neighbors, and Gaussian Naive Bayes classifiers, in predicting cardiovascular diseases (CVDs). Utilizing a dataset consisting of 70,000 clinical records, the model was rigorously tested through 5-fold cross-validation, achieving remarkable results with average accuracies, precision, recall, and F1-scores all exceeding 99%. The findings validate the hypothesis that ensemble models, due to their capacity to leverage multiple learning algorithms, provide superior prediction accuracy and reliability compared to single predictor models. This research not only confirms the effectiveness of ensemble methods in medical diagnostics but also highlights their potential to enhance decision-making in clinical settings. Given the model's success in identifying various stages of cardiovascular conditions with high accuracy, it offers significant implications for early intervention and personalized patient management. Future research should aim to validate these results across more diverse populations and explore the integration of additional predictive factors that could refine the model's applicability. This study contributes to the computational health field by demonstrating how advanced machine learning techniques can be effectively applied in predicting health outcomes.

Optimizing Cardiomegaly Detection: A Random Forest Approach to Processed Chest X-ray Imagery

This study explores the application of a Random Forest Classifier for the automated detection of Cardiomegaly from chest X-ray images, utilizing a dataset processed and derived from the NIH Chest X-ray Dataset. Given the crucial need for accurate and timely diagnosis of Cardiomegaly to inform appropriate treatment decisions, this research aims to determine the efficacy of machine learning models in augmenting diagnostic processes. Employing image pre-processing techniques such as Sobel filtering for edge detection and Hu Moments for feature extraction, the study enhances the input features for the model. The performance of the classifier was evaluated using a 5-fold cross-validation approach, yielding results with average accuracy, precision, recall, and F1-scores ranging approximately between 52% and 54%. These findings suggest a moderate level of reliability and consistency, indicating the potential utility of ensemble machine learning methods in medical imaging analysis. However, the variability in performance across different data subsets highlights the challenges and necessitates further optimization. This research contributes to the ongoing discourse on integrating machine learning into clinical settings, demonstrating the potential benefits and current limitations. Future research is recommended to expand the dataset variety, integrate advanced deep learning methodologies, and rigorously test these models in clinical environments. The findings hold significant implications for the development of automated diagnostic tools in healthcare, potentially leading to enhanced diagnostic accuracy and efficiency.

The use of 4D data-independent acquisition-based proteomic analysis and machine learning to reveal potential biomarkers for stress levels.

Research suggests that individuals who experience prolonged exposure to stress may be at higher risk for developing psychological stress disorders. Currently, psychological stress is primarily evaluated by professional physicians using rating scales, which may be prone to subjective biases and limitations of the scales. Therefore, it is imperative to explore more objective, accurate, and efficient biomarkers for evaluating the level of psychological stress in an individual. In this study, we utilized 4D data-independent acquisition (4D-DIA) proteomics for quantitative protein analysis, and then employed support vector machine (SVM) combined with SHAP interpretation algorithm to identify potential biomarkers for psychological stress levels. Biomarkers validation was subsequently achieved through machine learning classification and a substantial amount of a priori knowledge derived from the knowledge graph. We performed cross-validation of the biomarkers using two batches of data, and the results showed that the combination of Glyceraldehyde-3-phosphate dehydrogenase and Fibronectin yielded an average area under the curve (AUC) of 92%, an average accuracy of 86%, an average F1 score of 79%, and an average sensitivity of 83%. Therefore, this combination may represent a potential approach for detecting stress levels to prevent psychological stress disorders.

Explainable Pre-Trained Language Models for Sentiment Analysis in Low-Resourced Languages

Sentiment analysis is a crucial tool for measuring public opinion and understanding human communication across digital social media platforms. However, due to linguistic complexities and limited data or computational resources, it is under-represented in many African languages. While state-of-the-art Afrocentric pre-trained language models (PLMs) have been developed for various natural language processing (NLP) tasks, their applications in eXplainable Artificial Intelligence (XAI) remain largely unexplored. In this study, we propose a novel approach that combines Afrocentric PLMs with XAI techniques for sentiment analysis. We demonstrate the effectiveness of incorporating attention mechanisms and visualization techniques in improving the transparency, trustworthiness, and decision-making capabilities of transformer-based models when making sentiment predictions. To validate our approach, we employ the SAfriSenti corpus, a multilingual sentiment dataset for South African under-resourced languages, and perform a series of sentiment analysis experiments. These experiments enable comprehensive evaluations, comparing the performance of Afrocentric models against mainstream PLMs. Our results show that the Afro-XLMR model outperforms all other models, achieving an average F1-score of 71.04% across five tested languages, and the lowest error rate among the evaluated models. Additionally, we enhance the interpretability and explainability of the Afro-XLMR model using Local Interpretable Model-Agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP). These XAI techniques ensure that sentiment predictions are not only accurate and interpretable but also understandable, fostering trust and reliability in AI-driven NLP technologies, particularly in the context of African languages.

ResNet-Transformer deep learning model-aided detection of dens evaginatus.

Dens evaginatus is a dental morphological developmental anomaly. Failing to detect it may lead to tubercles fracture and pulpal/periapical disease. Consequently, early detection and intervention of dens evaginatus are significant to preserve vital pulp. This study aimed to develop a deep learning model to assist dentists in early diagnosing dens evaginatus, thereby supporting early intervention and mitigating the risk of severe consequences. In this study, a deep learning model was developed utilizing panoramic radiograph images sourced from 1410 patients aged 3-16 years, with high-quality annotations to enable the automatic detection of dens evaginatus. Model performance and model's efficacy in aiding dentists were evaluated. The findings indicated that the current deep learning model demonstrated commendable sensitivity (0.8600) and specificity (0.9200), outperforming dentists in detecting dens evaginatus with an F1-score of 0.8866 compared to their average F1-score of 0.8780, indicating that the model could detect dens evaginatus with greater precision. Furthermore, with its support, young dentists heightened their focus on dens evaginatus in tooth germs and achieved improved diagnostic accuracy. Based on these results, the integration of deep learning for dens evaginatus detection holds significance and can augment dentists' proficiency in identifying such anomaly.

Improving the Accuracy of RPA Image Classification in Vegetation Mapping of Magsaysay, Occidental Mindoro using RGB Indices, Canopy Heights, and Feature Importance Weighting

Abstract. Mapping vegetation cover is an important step in generating baseline data for various purposes such as smart agriculture, disaster and hazard monitoring, as well as risk assessment and planning. For this purpose, the use of Remotely Piloted Aircraft (RPA) has become prevalent in recent years as an efficient and cost-effective way of obtaining very-high-resolution images. However, it is limited by its lack of spectral bands used for discriminating between land cover classes, especially vegetation. An object-based approach was used due to its suitability with high-resolution input datasets, as it can recognize complex shapes and patterns aside from spectral characteristics. The drone images were segmented using optimal parameters to produce image objects which were subsequently classified through supervised learning using the Random Forest (RF) algorithm. This study incorporated non-conventional spectral indices that use RGB bands only, such as Triangular Greenness Index (TGI), Excess Green (ExG), and Tree-Grass Differentiation Index (TGDI), as well as the canopy height data derived from RPA photogrammetry to improve classification accuracy. To further improve model performance, appropriate band weights for segmentation were determined by running a RF classifier to obtain band importance values. Accuracy assessments reveal that using additional indices and heights improved the accuracy resulting in a 20% increase in the average f1-score, with the vegetation classes improving by a 25% increase in their f1-scores (8–41% improvement per class). Using the integration of band importance values as weights to the object-based segmentation slightly decreased accuracy values for the vegetation classes by an average of 0.04 in the f-1 score. The methods developed to improve the accuracy of RPA image classification make it more suitable for mapping vegetation.

A deep learning approach for gastroscopic manifestation recognition based on Kyoto Gastritis Score.

The risk of gastric cancer can be predicted by gastroscopic manifestation recognition and the Kyoto Gastritis Score. This study aims to validate the applicability of AI approaches for recognizing gastroscopic manifestations according to the definition of Kyoto Gastritis Score, with the goal of improving early gastric cancer detection and reducing gastric cancer mortality. In this retrospective study, 29013 gastric endoscopy images were collected and carefully annotated into five categories according to the Kyoto Gastritis Score, i.e. atrophy (A), diffuse redness (DR), enlarged folds (H), intestinal metaplasia (IM), and nodularity (N). As a multi-label recognition task, we propose a deep learning approach composed of five GAM-EfficientNet models, each performing a multiple classification to quantify gastroscopic manifestations, i.e. no presentation or the severity score 0-2. This approach was compared with endoscopists of varying years of experience in terms of accuracy, specificity, precision, recall, and F1 score. The approach demonstrated good performance in identifying the five manifestations of the Kyoto Gastritis Score, with an average accuracy, specificity, precision, recall, and F1 score of 78.70%, 91.92%, 80.23%, 78.70%, and 0.78, respectively. The average performance of five experienced endoscopists was 72.63%, 90.00%, 77.68%, 72.63%, and 0.73, while that of five less experienced endoscopists was 66.60%, 87.44%, 70.88%, 66.60%, and 0.66, respectively. The sample t-test indicates that the approach's average accuracy, specificity, precision, recall, and F1 score for identifying the five manifestations were significantly higher than those of less experienced endoscopists, experienced endoscopists, and all endoscopists on average (p < 0.05). Our study demonstrates the potential of deep learning approaches on gastric manifestation recognition over junior, even senior endoscopists. Thus, the deep learning approach holds potential as an auxiliary tool, although prospective validation is still needed to assess its clinical applicability.

News dissemination: a semantic approach to barrier classification

AbstractThe dissemination of information worldwide is significantly facilitated by the news media, with many events having global relevance across various regions. However, certain news events receive limited coverage restricted to specific geographic areas, due to the barriers that hinder the spread of information. These barriers can be attributed to political, geographical, economic, cultural, or linguistic factors. In this research, we propose an approach for classifying these barriers by extracting semantic information from news articles using Wikipedia-concepts. Our methodology involves the collection of news articles, each annotated to indicate the specific barrier types, leveraging metadata from news publishers. Subsequently, we employ Wikipedia-concepts, in conjunction with the content of the news articles, as features to determine the barriers to news dissemination. Our approach is then compared with traditional text classification techniques, deep learning methods, and transformer-based models. We have performed experiments on news articles from ten categories of topics including health, sports, business, etc. The findings indicate that 1) Utilizing semantic knowledge yields distinct concepts across the ten categories, thereby enhancing the effectiveness and speed of the classification model. 2) The proposed approach, incorporating Wikipedia-concepts-based semantic knowledge, leads to improved performance in barrier classification when compared to using solely the body text of news articles. Specifically, there is an increase in the average F1-scores for four out of five barriers, with the economic barrier rising from 0.65 to 0.68, the linguistic barrier from 0.71 to 0.72, the political barrier from 0.68 to 0.70, and the geographical barrier from 0.63 to 0.68.

A multiple biomolecules-based rapid life detection protocol embedded in a rover scientific subsystem for soil sample analysis

The study of whether life exists, is extinct, or not depends on various sophisticated experimental studies, as many different signatures of life can be used. The experimental procedures that can be performed to identify life can be further restricted by time, resources, and mobility constraints. Therefore, any research analyzing the presence of extraterrestrial life must be precise and unambiguous. This research focuses on the objective of the extraterrestrial life detection domain and seeks to provide an efficient protocol that can produce life detection decisions based on empirical data obtained through chemical analysis under time and resource-constrained conditions. While the majority of existing frameworks in this field are designed to identify biomolecules, our goal is to accomplish the same with minimal operational expense and mission complexity. We argue that the thoughtful integration of multiple biomolecular detections with lesser complexity and a robust framework can improve overall mission performance by satisfying the necessary time and resource constraints. In this study, a rapid multiple biomolecules-based life detection protocol (MBLDP-R) from soil samples is developed from scratch and embedded in an operational scientific rover subsystem targeted for planetary analysis missions. The study uses artificial biomolecule samples and simulated extraterrestrial environments to illustrate the suggested protocol’s end-to-end process. First, we list a few significant biomolecules, including lipids, proteins, carbohydrates, nucleic acids, ammonia, and pigments. Then, a weighted qualitative test scoring is applied to sort out the best test method for the finally selected biomolecules which are used as operational analogue to showcase the protocol’s in-situ analysis and decision-making capabilities. Based on the suitable biomolecules, a scientific exploration subsystem is developed, and the implemented protocol is built to perform onboard sample analysis. Evaluation results show that: (1) the proposed MBLDP-R protocol could effectively predict the classes with an average f1-score of 98.65% (macro) and 90.00% (micro), (2) the area under the Receiver Operating Characteristics (AUC-ROC) curve shows the sample categories to be correctly predicted 92% of the time (97% for Extant, 88% for Extinct, and 92% in the case of NPL), and (3) the protocol is time-efficient with an average completion time of 17.60 min, demonstrating the protocol’s rapid nature in detecting biosignatures in soil samples. The research outcome yields useful additional data for related future studies, particularly in the design of scientific frameworks for mission-specific requirements with limited resources while also serving as a reference point for constraint evaluation methods for similar systems.

Cumulative learning based segmentation aided cell mixtures classification in digital holographic microscopy

The tumor microenvironment consists of several components like tumor cells, normal healthy cells, extracellular matrix, secreted factors, and other non-cellular components. The interaction among these components is crucial for tumor progression and metastasis. Identifying the presence of tumor cells among the normal healthy cells during the initial stages of cancer will aid in its early detection and treatment. One of the main features that differentiates a cancer cell from a normal healthy cell distinctively is its morphology. In this study, the effectiveness of machine learning algorithms is evaluated in classifying cells co-cultured in a Petri dish. To achieve this, digital holographic microscopy is used to capture the quantitative phase images of human dermal fibroblast and A375 human melanoma cancer cells co-cultured at 1:1, 1:2, and 2:1 ratios. Cell segmentation was performed using a cumulative learning approach. Subsequently, morphological and phase-based features were extracted to train the machine learning algorithms for binary classification of cells. Phase-based features in addition to the morphological features were found to provide improved performance of the classifier. The classifier demonstrated an average F1-score of 0.9 during testing.

Automatic Detection of Trees using Airborne LiDAR Data Based on Geometric Characteristics

Abstract. One of the essential factors in analyzing urban environments is the presence of trees. Thus, the development of automatic or semi-automatic tree detection strategies is important for monitoring and providing data for municipal authorities’ planning efforts. In this context, we propose an automatic method for detecting trees using LiDAR data collected by airborne platforms. The proposed strategy uses the omnivariance as a key attribute, which is estimated locally from eigenvalues. Additionally, it utilizes an adaptive process to determine the optimal radius, followed by successive filtering based on the majority filter and mathematical morphology operators. The effectiveness of the proposed approach was evaluated on six study areas from two distinct datasets (Presidente Prudente/Brazil and Palmerston/New Zealand). In general, the results indicate a completeness rate around 99% and a correctness rate around 91%, resulting in an average Fscore of 95%. These findings suggest that the proposed approach has potential to detect trees in urban regions using airborne LiDAR data. Compared to related works, the proposed strategy tends to have a better result in terms of completeness.

Potato disease identification using transfer learning approaches

Potato crop is one of the prominent consumed foods by human beings. When potato crops are infected by diseases it affects farmers negatively and to run in a loss. Therefore, early detection of the potato crop disease can play a vital role in minimizing the loss of the farmers. Nowadays, artificial intelligence technologies, more specifically deep learning techniques, provide solutions to many crops disease-related problems. However, training deep learning models requires a high computational power and huge amount of data as they are data hungry models. Also, designing a custom CNN models a difficult task and there are some variations to be considered. To avoid these difficulties, we adopted two pretrained models of DenseNet121 and VGG19 through transfer learning approaches. The achieved accuracy for DenseNet121 and VGG19 models are 82.6% and 98.56% respectively. DenseNet121 model obtained the average precision, sensitivity, and F1-score of 88.19%, 82.53, and 82.04%, respectively. Whereas VGG19 yields 98.39% of precision, 98.39% of sensitivity, and 97.26% of F1-score in ternary 3-class classification (early blight vs healthy vs late blight).

SeFi-CD: A Semantic First Change Detection Paradigm That Can Detect Any Change You Want

The existing change detection (CD) methods can be summarized as the visual-first change detection (ViFi-CD) paradigm, which first extracts change features from visual differences and then assigns them specific semantic information. However, CD is essentially dependent on change regions of interest (CRoIs), meaning that the CD results are directly determined by the semantics changes in interest, making its primary image factor semantic of interest rather than visual. The ViFi-CD paradigm can only assign specific semantics of interest to specific change features extracted from visual differences, leading to the inevitable omission of potential CRoIs and the inability to adapt to different CRoI CD tasks. In other words, changes in other CRoIs cannot be detected by the ViFi-CD method without retraining the model or significantly modifying the method. This paper introduces a new CD paradigm, the semantic-first CD (SeFi-CD) paradigm. The core idea of SeFi-CD is to first perceive the dynamic semantics of interest and then visually search for change features related to the semantics. Based on the SeFi-CD paradigm, we designed Anything You Want Change Detection (AUWCD). Experiments on public datasets demonstrate that the AUWCD outperforms the current state-of-the-art CD methods, achieving an average F1 score 5.01% higher than that of these advanced supervised baselines on the SECOND dataset, with a maximum increase of 13.17%. The proposed SeFi-CD offers a novel CD perspective and approach.

Profiling public perception of emerging technologies: gene editing, brain chips and exoskeletons. A data-analytics framework

Three AI developments, classified as forms of human enhancements, center around progress at the intersection of AI, nanotechnology and biotechnology. Our research advances the understanding of AI and human enhancement by data-driven analytics and offers practical tools for future research and societal applications. It is based on a survey that was launched by PRC in February 2021 to more than 5,000 respondents from the U.S. It consists of about 100 questions that are grouped using a prefix code by the 3 above-mentioned human enhancement, science role, concerns and excitements, perceived algorithm fairness and demographics. To investigate this survey and extract insights regarding the general attitude, a data analytics framework is proposed that consists of clustering using DBSCAN and K-means, ANOVA for clusters, PCA, t-SNE and UMAP for graphical visualization, prediction and advanced customers’ profiles analyses. Both clustering methods indicate distinct profiles for AI customers. Most of them are moderate, but two smaller groups define the tech-ethics advocates and tech-forward visionaries. For prediction, in the multi-class classification task, the ROC-AUC score is 0.852 and average F1 Score is 0.987. Following the results, both technology creators and legislators must work collaboratively to ensure that technological advancements are ethically grounded, widely accepted and aligned with societal values.

LEAVES: An Expandable Light-curve Data Set for Automatic Classification of Variable Stars

With the increasing amount of astronomical observation data, it is an inevitable trend to use artificial intelligence methods for automatic analysis and identification of light curves for full samples. However, data sets covering all known classes of variable stars that meet all research needs are not yet available. There is still a lack of standard training data sets specifically designed for any type of light-curve classification, but existing light-curve training sets or data sets cannot be directly merged into a large collection. Based on the open data sets of the All-Sky Automated Survey for SuperNovae, Gaia, and Zwicky Transient Facility, we construct a compatible light-curve data set named LEAVES for automated recognition of variable stars, which can be used for training and testing new classification algorithms. The data set contains a total of 977,953 variable and 134,592 nonvariable light curves, in which the supported variables are divided into six superclasses and nine subclasses. We validate the compatibility of the data set through experiments and employ it to train a hierarchical random forest classifier, which achieves a weighted average F1-score of 0.95 for seven-class classification and 0.93 for 10-class classification. Experimental results prove that the classifier is more compatible than the classifier established based on a single band and a single survey, and has wider applicability while ensuring classification accuracy, which means it can be directly applied to different data types with only a relatively small loss in performance compared to a dedicated model.

Application of hyperspectral band selection method based on deep reinforcement learning to low-value recyclable waste classification

Hyperspectral images (HSIs) have proven effective for classification of Low-value Recyclable Waste (LVRW). However, the high correlation between bands in HSIs introduces redundant information. In this paper, to overcome the challenge of low applicability of existing band selection methods of LVRW HSIs, we propose a Dueling Double Deep Q Network based on Supervised Band Selection method (D3QN-SBS). Specifically, we formulate band selection as a reinforcement learning problem, treating it as a combinatorial optimization task that explores band combinations within a discrete space, using overall accuracy as the reward to tune the policy and optimise the state-action value function. The results of comparative experiments show that D3QN-SBS outperforms other methods when selecting 2–10 bands, where achieves an accuracy of about 99.24 %, 99.10 %, 99.05 %, and 99.16 % in k-NN, SVM-RBF, RF, and MLP based on 10 bands, the precision, recall, and F1-score are nearly 100 % for OTHERS and are more than 99 % for PS. In K-fold cross-validation, the majority of the folds under four classifiers achieves above 98 % for four evaluation metrics and the average F1-scores are 99.25 %, 99.01 %, 99.06 %, and 99.17 %. This approach can be deployed in LVRW sorting equipment, contributing to the advancement of hyperspectral imaging technologies in plastic waste sorting field.

Automated Recognition of Submerged Body-like Objects in Sonar Images Using Convolutional Neural Networks

The Police Robot for Inspection and Mapping of Underwater Evidence (PRIME) is an uncrewed surface vehicle (USV) currently being developed for underwater search and recovery teams to assist in crime scene investigation. The USV maps underwater scenes using sidescan sonar (SSS). Test exercises use a clothed mannequin lying on the seafloor as a target object to evaluate system performance. A robust, automated method for detecting human body-shaped objects is required to maximise operational functionality. The use of a convolutional neural network (CNN) for automatic target recognition (ATR) is proposed. SSS image data acquired from four different locations during previous missions were used to build a dataset consisting of two classes, i.e., a binary classification problem. The target object class consisted of 166 196 × 196 pixel image snippets of the underwater mannequin, whereas the non-target class consisted of 13,054 examples. Due to the large class imbalance in the dataset, CNN models were trained with six different imbalance ratios. Two different pre-trained models (ResNet-50 and Xception) were compared, and trained via transfer learning. This paper presents results from the CNNs and details the training methods used. Larger datasets are shown to improve CNN performance despite class imbalance, achieving average F1 scores of 97% in image classification. Average F1 scores for target vs background classification with unseen data are only 47% but the end result is enhanced by combining multiple weak classification results in an ensemble average. The combined output, represented as a georeferenced heatmap, accurately indicates the target object location with a high detection confidence and one false positive of low confidence. The CNN approach shows improved object detection performance when compared to the currently used ATR method.

Segment-Based Unsupervised Deep Learning for Human Activity Recognition using Accelerometer Data and SBOA based Channel Attention Networks

The deep learning community has increasingly focused on the critical challenges of human activity segmentation and detection based on sensors, which have numerous real-world applications. In most prior efforts, activity segmentation and recognition have been treated as separate processes, relying on pre-segmented sensor streams. This research proposes an unsupervised deep learning approach for Human Activity Recognition (HAR) that is segment-based, with an emphasis on activity continuity. The approach integrates segment-based SimCLR with Segment Feature Decorrelation (SDFD) and a new framework that leverages pairs of segment data for contrastive learning of visual representations. Furthermore, the Secretary Bird Optimization Algorithm (SBOA) and Channel Attention with Spatial Attention Network (CASANet) are utilized to enhance the performance of sensor-based human activity detection. CASANet effectively extracts key features and spatial dependencies in sensor data, while SBOA optimizes the model for greater accuracy and generalization. Evaluations on two publicly available datasets—Mhealth and PAMAP2—demonstrated an average F1 score of 98%, highlighting the approach’s efficacy in improving activity recognition performance.