Machine Learning Approach For Classification Research Articles

A Systematic Survey of Automatic Image Description Generation Systems

Automatic image description generation is an active research area in Computer Vision and Natural Language Processing. Objects, their attributes, actions, and spatial relationships are identified in the image description generation system. Earlier, these systems used classical machine learning approaches. Later majority of these works follow deep learning strategies. The essential goal of these systems is to produce syntactically and semantically correct sentences. This review aims to synthesize the studies conducted from 2010 to 2023 to get a deeper view of various image description generation systems and their applications. The prominent contribution of this review is that it covers the different aspects of image captioning systems, such as the methods used, the various applied domains, evaluation measures, and the datasets used. A single synthesized study directs scholars regarding developing image captioning systems to date utilizing machine learning approaches. It also offers suggestions for researchers in this sector for the future. Image captioning is applied in many fields like natural images, medical images, remote sensing images, videos, etc. This review paper reviews the various taxonomies of image description generation systems. We also analyzed multiple methods used in the architecture of image captioning systems. The datasets and the evaluation metrics used in these systems are discussed. We studied the performance of the system under various circumstances.

Uncovering mangrove range limits using very high resolution satellite imagery to detect fine‐scale mangrove and saltmarsh habitats in dynamic coastal ecotones

AbstractMangroves are important ecosystems for coastal biodiversity, resilience and carbon dynamics that are being threatened globally by human pressures and the impacts of climate change. Yet, at several geographic range limits in tropical–temperate transition zones, mangrove ecosystems are expanding poleward in response to changing macroclimatic drivers. Mangroves near range limits often grow to smaller statures and form dynamic, patchy distributions with other coastal habitats, which are difficult to map using moderate‐resolution (30‐m) satellite imagery. As a result, many of these mangrove areas are missing in global distribution maps. To better map small, scrub mangroves, we tested Landsat (30‐m) and Sentinel (10‐m) against very high resolution (VHR) Planet (3‐m) and WorldView (1.8‐m) imagery and assessed the accuracy of machine learning classification approaches in discerning current (2022) mangrove and saltmarsh from other coastal habitats in a rapidly changing ecotone along the east coast of Florida, USA. Our aim is to (1) quantify the mappable differences in landscape composition and complexity, class dominance and spatial properties of mangrove and saltmarsh patches due to image resolution; and (2) to resolve mapping uncertainties in the region. We found that the ability of Landsat to map mangrove distributions at the leading range edge was hampered by the size and extent of mangrove stands being too small for detection (50% accuracy). WorldView was the most successful in discerning mangroves from other wetland habitats (84% accuracy), closely followed by Planet (82%) and Sentinel (81%). With WorldView, we detected 800 ha of mangroves within the Florida range‐limit study area, 35% more mangroves than were detected with Planet, 114% more than Sentinel and 537% more than Landsat. Higher‐resolution imagery helped reveal additional variability in landscape metrics quantifying diversity, spatial configuration and connectedness among mangrove and saltmarsh habitats at the landscape, class and patch scales. Overall, VHR satellite imagery improved our ability to map mangroves at range limits and can help supplement moderate‐resolution global distributions and outdated regional maps.

Thermo-kinetics study of microalgal biomass in oxidative torrefaction followed by machine learning regression and classification approaches

Solid biofuels derived from microalgae represent a low-cost, high-volume bioproduct opportunity with excellent CO2 biofixation capability, contributing significantly to the attainment of net zero. Oxidative torrefaction also offers a more economical pretreatment to upgrade its solid fuel properties. This study investigates the thermo-kinetics aspect of oxidative torrefaction of microalgae with varying O2 concentrations using thermogravimetric and differential thermal analysis (TGA-DTA). Isoconversional kinetic modeling is applied to mass-loss data and shows average activation energies of 172.57, 174.68, 199.42, and 209.03 kJ‧mol−1 at 0, 3, 12, and 21 % O2 concentrations, respectively. The effect of O2, especially at 12 vol%, significantly reduces the calculated activation energy at lower conversion. DTA also reveals lower heat flow values under oxidative than inert torrefaction. Thermo-kinetics data are then utilized to conduct machine learning approaches. Regression via artificial neural networks shows that the prediction of conversion and heat flow values are predominantly dictated by the temperature, followed by heating rate and O2 concentration. Finally, classification using the k-nearest neighbors algorithm highlights the effects of factors at specific ranges of conversion and heat flow responses. The O2 concentration is significant only at early conversion (X<0.1) and contributes to generally lower heat flow values.

Surface soil moisture from combined active and passive microwave observations: Integrating ASCAT and SMAP observations based on machine learning approaches

The fusion of active and passive microwave measurements is expected to provide more robust surface soil moisture (SSM) mapping across various environmental conditions compared to the use of a single sensor. Thus, the integration of the newest L-band passive (i.e., Soil Moisture Active Passive, SMAP) and the active (i.e., the Advanced Scatterometer, ASCAT) observations provides an opportunity for SSM mapping with improved accuracy. However, this integration remains largely underexplored. In this context, the integration of SMAP brightness temperature (TB) and ASCAT backscattering coefficients for global-scale SSM estimation was investigated, by fully considering the potential error sources in conventional radiative transfer models (RTMs) as well as other SSM linked factors. Based on ground measurements from globally distributed dense networks with mitigated mismatch issues and spatial/temporal independent evaluation strategies, this study: (i) comprehensively evaluated four classical machine learning approaches, including Random Forest (RF), Long-Short Term Memory (LSTM), Support Vector Machine (SVM), and Cascaded Neural Network (CNN), and chose the best performing RF method to implement the final integration of SSM; (ii) compared the integration retrievals to those made using data from a single sensor (SMAP or ASCAT) with the same machine learning framework, as well as to the SMAP passive, ASCAT active, and ESA CCI active-passive combined SSM products. The results show the integration retrievals achieve satisfactory performance by obtaining an averaged unbiased root mean squared error (ubRMSE) of 0.042 m3/m3 and a temporal correlation of 0.756, which are superior to machine learning based SSM estimated from a single active or passive sensor, and also outperform the SMAP, ASCAT, and ESA CCI products. Moreover, the temporal resolution is evidently improved compared to the SMAP and ASCAT SSM products, with a temporal ratio exceeding 60% for most areas across the globe. Therefore, blending active and passive measurements affords a more reliable SSM mapping with increased sampling at the global scale, and could contribute to improved hydro-ecological applications.

Quantum deep learning-based anomaly detection for enhanced network security

Identifying and mitigating aberrant activities within the network traffic is important to prevent adverse consequences caused by cyber security incidents, which have been increasing significantly in recent times. Existing research mainly focuses on classical machine learning and deep learning-based approaches for detecting such attacks. However, exploiting the power of quantum deep learning to process complex correlation of features for anomaly detection is not well explored. Hence, in this paper, we investigate quantum machine learning and quantum deep learning-based anomaly detection methodologies to accurately detect network attacks. In particular, we propose three novel quantum auto-encoder-based anomaly detection frameworks. Our primary aim is to create hybrid models that leverage the strengths of both quantum and deep learning methodologies for efficient anomaly recognition. The three frameworks are formed by integrating the quantum autoencoder with a quantum one-class support vector machine, a quantum random forest, and a quantum k-nearest neighbor approach. The anomaly detection capability of the frameworks is evaluated using benchmark datasets comprising computer and Internet of Things network flows. Our evaluation demonstrates that all three frameworks have a high potential to detect the network traffic anomalies accurately, while the framework that integrates the quantum autoencoder with the quantum k-nearest neighbor yields the highest accuracy. This demonstrates the promising potential for the development of quantum frameworks for anomaly detection, underscoring their relevance for future advancements in network security.

Artificial Intelligence in Ship Trajectory Prediction

Maritime traffic is increasing more and more, creating more complex navigation environments for ships. Ship trajectory prediction based on historical AIS data is a vital method of reducing navigation risks and enhancing the efficiency of maritime traffic control. At present, employing machine learning or deep learning techniques to construct predictive models based on AIS data has become a focal point in ship trajectory prediction research. This paper systematically evaluates various trajectory prediction methods, spanning classical machine learning approaches and emerging deep learning techniques, to uncover their respective merits and drawbacks. In this work, a variety of studies were investigated that applied different algorithms in ship trajectory prediction, including regression models (RMs), artificial neural networks (ANNs), Kalman filtering (KF), and random forests (RFs) in machine learning, along with deep learning such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory (LSTM), gate recurrent unit (GRU) networks, and sequence-to-sequence (Seq2seq) networks. The performance of predictive models based on different algorithms in trajectory prediction tasks was graded and analyzed. Among the existing studies, deep learning methods exhibit significant performance and considerable potential application value for maritime traffic systems, which can be assessed by future work on ship trajectory prediction research.

Bridging the Gap

Effective organization and retrieval of news content are heavily reliant on accurate news classification. While the mountainous research has been conducted in resourceful languages like English and Chinese, the researches on under-resourced languages like the Kurdish language are severely lacking. To address this challenge, we introduce a hybrid approach called RFO-CNN in this paper. The proposed method combines an improved version of red fox optimization algorithm (RFO) and convolutional neural network (CNN) for finetuning CNN’s parameters. Our model’s efficacy was tested on two widely used Kurdish news datasets, KNDH and KDC-4007, both of which contain news articles classified into various categories. We compared the performance of RFO-CNN to other cutting-edge deep learning models such as bidirectional long short-term memory networks and bidirectional encoder representations from transformers (BERT) transformers, as well as classical machine learning approaches such as multinomial naive bayes, support vector machine, and K-nearest neighbors. We trained and tested our datasets using four different scenarios: 60:40, 70:30, 80:20, and 90:10. Our experimental results demonstrate the superiority of the RFO-CNN model across all scenarios, outperforming the benchmark BERT model and other machine learning models in terms of accuracy and F1-score.

Damage and Nonlinearity Effects on Stress Wave Propagation in Planar Frame Structures: A Machine Learning Classification Approach Based on Stress Wave Amplitude Solution

Damage and Nonlinearity Effects on Stress Wave Propagation in Planar Frame Structures: A Machine Learning Classification Approach Based on Stress Wave Amplitude Solution

LSTM vs CNN in real ship trajectory classification

Abstract Ship-type identification in a maritime context can be critical to the authorities to control the activities being carried out. Although Automatic Identification Systems has been mandatory for certain vessels, if a vessel does not have them voluntarily or not, it can lead to a whole set of problems, which is why the use of tracking alternatives such as radar is fully complementary for a vessel monitoring systems. However, radars provide positions, but not what they are detecting. Having systems capable of adding categorical information to radar detections of vessels makes it possible to increase control of the activities being carried out, improve safety in maritime traffic, and optimize on-site inspection resources on the part of the authorities. This paper addresses the binary classification problem (fishing ships versus all other vessels) using unbalanced data from real vessel trajectories. It is performed from a deep learning approach comparing two of the main trends, Convolutional Neural Networks and Long Short-Term Memory. In this paper, it is proposed the weighted cross-entropy methodology and compared with classical data balancing strategies. Both networks show high performance when applying weighted cross-entropy compared with the classical machine learning approaches and classical balancing techniques. This work is shown to be a novel approach to the international problem of identifying fishing ships without context.

Machine learning classification approaches to optimize ROP and TOB using drilling and geomechanical parameters in a carbonate reservoir

Drilling optimization has been broadly developed in terms of influential parameters. The assessment time and the effects of both geomechanical and drilling parameters were vital challenges of investigations. Drilling factors are applied force or rotation of drilling agents such as weight on bit (WOB), and geomechanical features represent mechanical indexes of rocks including unconfined compressive strength (UCS). Optimization efforts have been demonstrated on complex prediction methods whereas the simplicity of classification can offer some optimal ranges utilizing machine learning classifications in an accelerated process. In this study, a novel procedure using the supervised and semi-supervised learning approaches was conducted to classify and optimize the rate of penetration (ROP) and torque on bit (TOB). Firstly, in the case well, user-defined classes were assigned based on geomechanical units (GMU) and the ranges of high ROP and low TOB, thus classes divided drilling factors as GMUs of the case. Secondly, the feature selection was carried out by neural pattern recognition with three multi-objective optimization methods for classification. The inputs of classifications were WOB, hook load, pump pressure, flow rate, UCS, and internal friction angle. Classification approaches were decision trees, support vector machine (SVM), and ensemble learning. Finally, the bagged trees permutation and Laplacian SVM (LapSVM) algorithm separately revealed the significance of parameters and predicted the optimal ROP and TOB regions. Findings showed (1) in supervised classification of the case well, the cubic SVM and bagged trees had the highest area under the curve (AUC) and accuracy, on average 0.97 and 0.96, respectively. (2) The average accuracy of the supervised classifications in a test well was 91% except for the fine SVM, which makes them reliable for the fields with the least information. (3) The permutation outcomes for significant features, flow rate and UCS, exposed influential parameters for ROP and TOB optimization. (4) The semi-supervised method, LapSVM, not only acquired both ROP and TOB labels with an accuracy of 88% but also presented their optimal ranges in 95% of the assessed zones. (5) LapSVM deals with a limited training section perfectly opposed to the supervised version, which is vital for drilling investigation. (6) Implementing machine learning classification approaches with rock properties is a key factor in achieving effective drilling parameters in less time. More importantly, the recommended drilling factors concerning geomechanical properties can ameliorate both drilling performance and perception of upcoming collapse.

Exploring the Feasibility of Applying Deep Learning for the Early Prediction of Arthritis

Introduction: Arthritis is one of the most common chronic diseases. Early detection of arthritis and its progression can facilitate early intervention measures, lowering disease severity in patients. As electronic health records (EHR) become more accessible, this study assesses whether general health information and arthritis-related questionnaires can be used in arthritis diagnosis, without the involvement of costly imaging methods. Therefore, we created deep learning (DL) and machine learning (ML) models to explore the feasibility of combining EHR and modern computational tools to diagnose arthritis. Methods: A total of 782 arthritis patients and 4014 control patients were identified from the Osteoarthritis Initiative (OAI) – a ten-year-long observational study that included patient EHR in five time points. Six hundred variables were filtered by random forest classifier followed by manual filtering. Data were split properly to training, testing and validation set, and the training set was balanced. Sequential, nonsequential DL models, and five independent DL models for each time points were used. The accuracy, positive prevalence value (PPV), negative prevalence value (NPV), and area under curve (AUC), were assessed and compared with four classical ML models. SHAP (SHapley Additive exPlanations) summary analysis was also conducted. Results: Sequential and non-sequential deep learning models showed accuracies of ~ 0.97, and the four classical machine learning approaches showed accuracies of above 0.9. High positive and negative predicted values (&gt; 0.90) for all of the models suggested the potential clinical applicability of the model, while the SHAP analysis demonstrated its interpretability. Discussion: We tested various models and showed the ability to use machine learning methods for early diagnosis of arthritis with EHR. The models can be used as a screening tool to select susceptible patients for confirmatory tests such as X-ray and MRI. Identification of early disease states could facilitate protective measures that slow disease progression.

Predicting dental anxiety in young adults: classical statistical modelling approach versus machine learning approach

ObjectivesTo predict and identify the key demographic and clinical exposure factors associated with dental anxiety among young adults, and to compare if the traditional statistical modelling approach provides similar results to the machine learning (ML) approach in predicting factors for dental anxiety.MethodsA cross-sectional study of Western Illinois University students. Three survey instruments (sociodemographic questionnaire, modified dental anxiety scale (MDAS), and dental concerns assessment tool (DCA)) were distributed via email to the students using survey monkey. The dependent variable was the mean MDAS scores, while the independent variables were the sociodemographic and dental concern assessment variables. Multivariable analysis was done by comparing the classical statistical model and the machine learning model. The classical statistical modelling technique was conducted using the multiple linear regression analysis and the final model was selected based on Akaike information Criteria (AIC) using the backward stepwise technique while the machine learining modelling was performed by comparing two ML models: LASSO regression and extreme gradient boosting machine (XGBOOST) under 5-fold cross-validation using the resampling technique. All statistical analyses were performed using R version 4.1.3.ResultsThe mean MDAS was 13.73 ± 5.51. After careful consideration of all possible fitted models and their interaction terms the classical statistical approach yielded a parsimonious model with 13 predictor variables with Akaike Information Criteria (AIC) of 2376.4. For the ML approach, the Lasso regression model was the best-performing model with a mean RMSE of 0.617, R2 of 0.615, and MAE of 0.483. Comparing the variable selection of ML versus the classical statistical model, both model types identified 12 similar variables (out of 13) as the most important predictors of dental anxiety in this study population.ConclusionThere is a high burden of dental anxiety within this study population. This study contributes to reducing the knowledge gap about the impact of clinical exposure variables on dental anxiety and the role of machine learningin the prediction of dental anxiety. The predictor variables identified can be used to inform public health interventions that are geared towards eliminating the individual clinical exposure triggers of dental anxiety are recommended.

Classification of Ionospheric Scintillations during high Solar Activity and Geomagnetic Storm over Visakhapatnam Region using Machine Learning Approach

The ionospheric plasma disturbances typically correlate with irregularities in electron density and ionospheric scintillations are produced in reaction to these variations generating radio signal fluctuations. Geolocation services and space based communication are endangered due to ionospheric scintillation which promptly produces fluctuations in information collected by Global Navigation Satellite Systems and this is at its strongest when the solar cycle is at its peak. Ionospheric space weather has a significant impact on Global Navigation Satellite Systems (GNSS) and one crucial aspect used in investigating ionospheric characteristics is total electron content (TEC). Due to fluctuations in time and space, the TEC obtained from GNSS signals is nonlinear and nonstationary. In this study, machine learning approaches for Classification of the ionospheric scintillations were used during the high solar activity and geomagnetic storm in the month of July 2023. This approach enables the classification of ionospheric phase scintillations using well-known classifiers: Decision Tree and Support Vector Machine.

Siamese Networks for Clinically Relevant Bacteria Classification Based on Raman Spectroscopy.

Identifying bacterial strains is essential in microbiology for various practical applications, such as disease diagnosis and quality monitoring of food and water. Classical machine learning algorithms have been utilized to identify bacteria based on their Raman spectra. However, convolutional neural networks (CNNs) offer higher classification accuracy, but they require extensive training sets and retraining of previous untrained class targets can be costly and time-consuming. Siamese networks have emerged as a promising solution. They are composed of two CNNs with the same structure and a final network that acts as a distance metric, converting the classification problem into a similarity problem. Classical machine learning approaches, shallow and deep CNNs, and two Siamese network variants were tailored and tested on Raman spectral datasets of bacteria. The methods were evaluated based on mean sensitivity, training time, prediction time, and the number of parameters. In this comparison, Siamese-model2 achieved the highest mean sensitivity of 83.61 ± 4.73 and demonstrated remarkable performance in handling unbalanced and limited data scenarios, achieving a prediction accuracy of 73%. Therefore, the choice of model depends on the specific trade-off between accuracy, (prediction/training) time, and resources for the particular application. Classical machine learning models and shallow CNN models may be more suitable if time and computational resources are a concern. Siamese networks are a good choice for small datasets and CNN for extensive data.

Selective Partitioned Regression for Accurate Kidney Health Monitoring

The number of people diagnosed with advanced stages of kidney disease have been rising every year. Early detection and constant monitoring are the only minimally invasive means to prevent severe kidney damage or kidney failure. We propose a cost-effective machine learning-based testing system that can facilitate inexpensive yet accurate kidney health checks. Our proposed framework, which was developed into an iPhone application, uses a camera-based bio-sensor and state-of-the-art classical machine learning and deep learning techniques for predicting the concentration of creatinine in the sample, based on colorimetric change in the test strip. The predicted creatinine concentration is then used to classify the severity of the kidney disease as healthy, intermediate, or critical. In this article, we focus on the effectiveness of machine learning models to translate the colorimetric reaction to kidney health prediction. In this setting, we thoroughly evaluated the effectiveness of our novel proposed models against state-of-the-art classical machine learning and deep learning approaches. Additionally, we executed a number of ablation studies to measure the performance of our model when trained using different meta-parameter choices. Our evaluation results indicate that our selective partitioned regression (SPR) model, using histogram of colors-based features and a histogram gradient boosted trees underlying estimator, exhibits much better overall prediction performance compared to state-of-the-art methods. Our initial study indicates that SPR can be an effective tool for detecting the severity of kidney disease using inexpensive lateral flow assay test strips and a smart phone-based application. Additional work is needed to verify the performance of the model in various settings.

Exploring deep learning techniques for wild animal behaviour classification using animal‐borne accelerometers

Abstract Machine learning‐based behaviour classification using acceleration data is a powerful tool in bio‐logging research. Deep learning architectures such as convolutional neural networks (CNN), long short‐term memory (LSTM) and self‐attention mechanism as well as related training techniques have been extensively studied in human activity recognition. However, they have rarely been used in wild animal studies. The main challenges of acceleration‐based wild animal behaviour classification include data shortages, class imbalance problems, various types of noise in data due to differences in individual behaviour and where the loggers were attached and complexity in data due to complex animal‐specific behaviours, which may have limited the application of deep learning techniques in this area. To overcome these challenges, we explored the effectiveness of techniques for efficient model training: data augmentation, manifold mixup and pre‐training of deep learning models with unlabelled data, using datasets from two species of wild seabirds and state‐of‐the‐art deep learning model architectures. Data augmentation improved the overall model performance when one of the various techniques (none, scaling, jittering, permutation, time‐warping and rotation) was randomly applied to each data during mini‐batch training. Manifold mixup also improved model performance, but not as much as random data augmentation. Pre‐training with unlabelled data did not improve model performance. The state‐of‐the‐art deep learning models, including a model consisting of four CNN layers, an LSTM layer and a multi‐head attention layer, as well as its modified version with shortcut connection, showed better performance among other comparative models. Using only raw acceleration data as inputs, these models outperformed classic machine learning approaches that used 119 handcrafted features. Our experiments showed that deep learning techniques are promising for acceleration‐based behaviour classification of wild animals and highlighted some challenges (e.g. effective use of unlabelled data). There is scope for greater exploration of deep learning techniques in wild animal studies (e.g. advanced data augmentation, multimodal sensor data use, transfer learning and self‐supervised learning). We hope that this study will stimulate the development of deep learning techniques for wild animal behaviour classification using time‐series sensor data.

MetRoBERTa: Leveraging Traditional Customer Relationship Management Data to Develop a Transit-Topic-Aware Language Model

Transit riders’ feedback provided in ridership surveys, customer relationship management (CRM) channels, and, in more recent times, through social media, is key for transit agencies to better gauge the efficacy of their services and initiatives. Getting a holistic understanding of riders’ experience through the feedback shared in those instruments is often challenging, mostly because of the open-ended, unstructured nature of text feedback. In this paper, we propose leveraging traditional transit CRM feedback to develop and deploy a transit-topic-aware large language model (LLM) capable of classifying open-ended text feedback to relevant transit-specific topics. First, we utilize semi-supervised learning to engineer a training dataset of 11 broad transit topics detected in a corpus of 6 years of customer feedback provided to the Washington Metropolitan Area Transit Authority. We then use this dataset to train and thoroughly evaluate a language model based on the RoBERTa architecture. We compare our LLM—MetRoBERTa—with classical machine learning approaches utilizing keyword-based and lexicon representations. Our model outperforms those methods across all evaluation metrics, providing an average topic classification accuracy of 90%. Finally, we provide a value proposition of this work demonstrating how the language model, alongside additional text processing tools, can be applied to add structure to open-ended text sources of feedback such as Twitter. The framework and results we present provide a pathway for an automated, generalizable approach for ingesting, visualizing, and reporting transit riders’ feedback at scale, enabling agencies to better understand and improve customer experience.

Evaluating Different Deep Learning Approaches for Tree Health Classification Using High-Resolution Multispectral UAV Data in the Black Forest, Harz Region, and Göttinger Forest

We present an evaluation of different deep learning and machine learning approaches for tree health classification in the Black Forest, the Harz Mountains, and the Göttinger Forest on a unique, highly accurate tree-level dataset. The multispectral UAV data were collected from eight forest plots with diverse tree species, mostly conifers. As ground truth data (GTD), nearly 1500 tree polygons with related attribute information on the health status of the trees were used. This data were collected during extensive fieldwork using a mobile application and subsequent individual tree segmentation. Extensive preprocessing included normalization, NDVI calculations, data augmentation to deal with the underrepresented classes, and splitting the data into training, validation, and test sets. We conducted several experiments using a classical machine learning approach (random forests), as well as different convolutional neural networks (CNNs)—ResNet50, ResNet101, VGG16, and Inception-v3—on different datasets and classes to evaluate the potential of these algorithms for tree health classification. Our first experiment was a binary classifier of healthy and damaged trees, which did not consider the degree of damage or tree species. The best results of a 0.99 test accuracy and an F1 score of 0.99 were obtained with ResNet50 on four band composites using the red, green, blue, and infrared bands (RGBI images), while VGG16 had the worst performance, with an F1 score of only 0.78. In a second experiment, we also distinguished between coniferous and deciduous trees. The F1 scores ranged from 0.62 to 0.99, with the highest results obtained using ResNet101 on derived vegetation indices using the red edge band of the camera (NDVIre images). Finally, in a third experiment, we aimed at evaluating the degree of damage: healthy, slightly damaged, and medium or heavily damaged trees. Again, ResNet101 had the best performance, this time on RGBI images with a test accuracy of 0.98 and an average F1 score of 0.97. These results highlight the potential of CNNs to handle high-resolution multispectral UAV data for the early detection of damaged trees when good training data are available.

O2 supplementation disambiguation in clinical narratives to support retrospective COVID-19 studies

BackgroundOxygen saturation, a key indicator of COVID-19 severity, poses challenges, especially in cases of silent hypoxemia. Electronic health records (EHRs) often contain supplemental oxygen information within clinical narratives. Streamlining patient identification based on oxygen levels is crucial for COVID-19 research, underscoring the need for automated classifiers in discharge summaries to ease the manual review burden on physicians.MethodWe analysed text lines extracted from anonymised COVID-19 patient discharge summaries in German to perform a binary classification task, differentiating patients who received oxygen supplementation and those who did not. Various machine learning (ML) algorithms, including classical ML to deep learning (DL) models, were compared. Classifier decisions were explained using Local Interpretable Model-agnostic Explanations (LIME), which visualize the model decisions.ResultClassical ML to DL models achieved comparable performance in classification, with an F-measure varying between 0.942 and 0.955, whereas the classical ML approaches were faster. Visualisation of embedding representation of input data reveals notable variations in the encoding patterns between classic and DL encoders. Furthermore, LIME explanations provide insights into the most relevant features at token level that contribute to these observed differences.ConclusionDespite a general tendency towards deep learning, these use cases show that classical approaches yield comparable results at lower computational cost. Model prediction explanations using LIME in textual and visual layouts provided a qualitative explanation for the model performance.

Prediction of Drug-like Compounds Solubility in Supercritical Carbon Dioxide: A Comparative Study between Classical Density Functional Theory and Machine Learning Approaches

Prediction of Drug-like Compounds Solubility in Supercritical Carbon Dioxide: A Comparative Study between Classical Density Functional Theory and Machine Learning Approaches