Macro-averaged Precision Research Articles

Red piranha optimization (RPO): a natural inspired meta-heuristic algorithm for solving complex optimization problems

An optimization algorithm is a step-by-step procedure which aims to achieve an optimum value (maximum or minimum) of an objective function. Several natural inspired meta-heuristic algorithms have been inspired to solve complex optimization problems by utilizing the potential advantages of swarm intelligence. In this paper, a new nature-inspired optimization algorithm which mimics the social hunting behavior of Red Piranha is developed, which is called Red Piranha Optimization (RPO). Although the piranha fish is famous for its extreme ferocity and thirst for blood, it sets the best examples of cooperation and organized teamwork, especially in the case of hunting or saving their eggs. The proposed RPO is established through three sequential phases, namely; (i) searching for a prey, (ii) encircling the prey, and (iii) attacking the prey. A mathematical model is provided for each phase of the proposed algorithm. RPO has salient properties such as; (i) it is very simple and easy to implement, (ii) it has a perfect ability to bypass local optima, and (iii) it can be employed for solving complex optimization problems covering different disciplines. To ensure the efficiency of the proposed RPO, it has been applied in feature selection, which is one of the important steps in solving the classification problem. Hence, recent bio-inspired optimization algorithms as well as the proposed RPO have been employed for selecting the most important features for diagnosing Covid-19. Experimental results have proven the effectiveness of the proposed RPO as it outperforms the recent bio-inspired optimization techniques according to accuracy, execution time, micro average precision, micro average recall, macro average precision, macro average recall, and f-measure calculations.

A Deceptive Reviews Detection Method Based on Multidimensional Feature Construction and Ensemble Feature Selection

Deceptive reviews on social media and e-commerce websites are inflammatory and will significantly affect the judgment and purchase behavior of other users. At present, many researchers build models based on single text features to detect deceptive reviews. However, deceptive reviewers will deliberately imitate the text style of true reviews when writing reviews. At this time, these methods based on text features are not necessarily effective. What’s more, detection performance is limited because the category distribution is likely to be unbalanced in practice. In this work, to address these shortcomings, a deceptive review detection method based on multidimensional feature construction and ensemble feature selection is proposed. Our proposal constructs 3-D features including text feature, reviewer behavior feature, and deceptive score feature. In addition, to alleviate the impact of unbalanced category distribution, a data resampling algorithm is applied which incorporates random under-sampling (RUS) and Borderline-SMOTE algorithm. Furthermore, we integrate the results of different feature selection based on the Chi-square test, Information gain, and XGBoost feature importance. Our method addresses the limitation of single dimension features and can provide useful detection. Experimental results area under the curve (AUC, Macro Average Precision, and weighted F1-score) show that the proposed method performs well in the task of deceptive review detection on two Amazon datasets. Compared with other advanced methods, our method achieves additional performance gains in the case of poor text quality and datasets with unbalanced category distribution.

AI-assisted mole detection for online dermatology triage in telemedicine settings

Skin moles are one of the most critical conditions for early diagnosis of severe conditions such as melanoma. Early identification of moles has become crucial nowadays, primarily due to malignant melanoma, a dangerous type of skin cancer. Most of the recent advances concerning this domain of dermatology deal mostly with either classifying moles as benign or malignant or with methods that help delineate moles from skin images. However, there are minimal sources of exploration in just determining whether moles are present in a given query image. In this paper, we have employed the latest state-of-the-art neural networks to identify the presence of moles in dermatological images uploaded by patients on an online teledermatology platform. This filter flags if a mole is detected, which can be employed as a triage system that helps the dermatologist diagnose and ease the follow-up treatment procedure in a physical setting if the patient has a mole in their image. A comparative study of the prediction performance of the different models has been provided for different performance metrics of interest. The results presented in this paper have been obtained from two sets of data, consisting of more than 26,000 clinical pictures with different conditions combined. Multiple experiments using different models yielded a macro average recall value as high as 0.955, along with overall accuracy and macro average precision values of 0.962 and 0.958, respectively.

Feasibility Study of the Prediction of Radiologist's Instructions with the Bi-LSTM Model Trained with Descriptions of MR Imaging Order-statement

Magnetic resonance (MR) images provide essential diagnostic information; however, it is also a very burdensome examination for patients. At our hospital, radiologists make imaging instructions for all MR examination orders, but this is a time-consuming task. If a natural language processing model can predict the imaging instructions, it will be possible to reduce the burden on radiologists and the instruction quality can be assured. The purpose of this study was to investigate the feasibility of using natural language processing to predict MR imaging instructions with the aim of assisting radiologists. Considering the uniqueness of the MR imaging protocols at each facility and the particularity of the test order text, we considered that the use of large datasets and pre-training models would be unsuitable. We focused on LSTM, which has been used for natural language processing, and built a 4-layer bi-LSTM model in combination with our own morphological preprocessing to predict MR imaging instructions. The proposed method achieved macro-average precision, recall, and F1-score of 70.9%, 65.4%, and 66.6%, respectively. Compared to the previous studies, the proposed method achieved satisfactory performance in the natural language analysis task for Japanese. It is considered that the proposed method improved the prediction accuracy of the minority class through direct and indirect effects of vocabulary reduction, optimization, and similarity learning. It is suggested that the proposed method is effective and that the prediction of MR imaging instructions using natural language analysis in combination with the proposed method is feasible.

Human monkeypox diagnose (HMD) strategy based on data mining and artificial intelligence techniques

In May 2022, monkeypox re-emerged as a rare zoonotic disease that is an important viral disease for public health. Monkeypox can be transmitted from animals to humans, between humans through close contact with an infected human, or with a virus stained substance. Through this paper, a new detection strategy based on artificial intelligence techniques is provided to early detect monkeypox patients. This strategy is called Human Monkeypox Detection (HMD) strategy and mainly consists of two main phases, which are; (i) Selection Phase (SP) and (ii) Detection Phase (DP). While SP tries to select the best features, DP tries to introduce fast and accurate detection based on valid data from SP. In SP, an Improved Binary Chimp Optimization (IBCO) algorithm as a new feature selection algorithm is introduced to select valuable features before learning an Ensemble Diagnosis (ED) model as a new diagnostic algorithm in the next phase called DP. In fact, the proposed IBCO algorithm is a hybrid selection algorithm that includes both filter and wrapper methods. IBCO consists of a filter layer called Filter Selection Layer (FSL) and a wrapper layer called Wrapper Selection Layer (WSL). At first, monkeypox dataset is entered into FSL to quickly select meaningful features by using ‘m’ filter selection techniques. Then, ‘m’ sets of selected features are fed into WSL to construct the initial population of Binary Chimp Optimization (BCO) algorithm to precisely choose the best set of features for the next phase (DP). Finally, the ED model will be correctly trained on the filtered data from FSL. This model consists of three diagnostic algorithms called Weighted Naïve Bayes (WNB), Weighted K-Nearest Neighbors (WKNN), and deep learning which are combined using a new weighted voting method to provide the best diagnostic results. The weighted values of WNB algorithm are determined by measuring the impact of each feature on the class categories while the Grey Wolf Optimization (GWO) algorithm is used to determine the weighted values of WKNN. Experimental results illustrated that the suggested feature selection algorithm called IBCO outperforms other modern feature selection methods and also the proposed ED model outperforms other modern diagnostic models. At the end, the HMD strategy gives the best results compared to other modern strategies with accuracy, precision, and recall values equal 98.48%, 91.1% and 88.91% respectively. Also, the HMD gives 92.56%,89.01%,88.01%,85.01%, 83.9%, and 5.4 s for micro-average precision, micro-average recall, macro-average precision, macro-average recall, F1-measure, and implementation time values respectively.

Dynamic gestational week prediction model for pre-eclampsia based on ID3 algorithm.

Pre-eclampsia (PE) is a type of hypertensive disorder during pregnancy, which is a serious threat to the life of mother and fetus. It is a placenta-derived disease that results in placental damage and necrosis due to systemic small vessel spasms that cause pathological changes such as ischemia and hypoxia and oxidative stress, which leads to fetal and maternal damage. In this study, four types of risk factors, namely, clinical epidemiology, hemodynamics, basic biochemistry, and biomarkers, were used for the initial selection of model parameters related to PE, and factors that were easily available and clinically recognized as being associated with a higher risk of PE were selected based on hospital medical record data. The model parameters were then further analyzed and screened in two subgroups: early-onset pre-eclampsia (EOPE) and late-onset pre-eclampsia (LOPE). Dynamic gestational week prediction model for PE using decision tree ID3 algorithm in machine learning. Performance of the model was: macro average (precision = 76%, recall = 73%, F1-score = 75%), weighted average (precision = 88%, recall = 89%, F1-score = 89%) and overall accuracy is 86%. In this study, the addition of the dynamic timeline parameter "gestational week" made the model more convenient for clinical application and achieved effective PE subgroup prediction.

Hybrid Framework for Diabetic Retinopathy Stage Measurement Using Convolutional Neural Network and a Fuzzy Rules Inference System

Diabetic retinopathy (DR) is an increasingly common eye disorder that gradually damages the retina. Identification at the early stage can significantly reduce the severity of vision loss. Deep learning techniques provide detection for retinal images based on data size and quality, as the error rate increases with low-quality images and unbalanced data classes. This paper proposes a hybrid intelligent framework of a conventional neural network and a fuzzy inference system to measure the stages of DR automatically, Diabetic Retinopathy Stage Measurement using Conventional Neural Network and Fuzzy Inference System (DRSM-CNNFIS). The fuzzy inference used human experts’ rules to overcome data dependency problems. At first, the Conventional Neural Network (CNN) model was used for feature extraction, and then fuzzy rules were used to measure diabetic retinopathy stage percentage. The framework is trained using images from Kaggle datasets (Diabetic Retinopathy Detection, 2022). The efficacy of this framework outperformed the other models with regard to accuracy, macro average precision, macro average recall, and macro average F1 score: 0.9281, 0.7142, 0.7753, and 0.7301, respectively. The evaluation results indicate that the proposed framework, without any segmentation process, has a similar performance for all the classes, while the other classification models (Dense-Net-201, Inception-ResNet ResNet-50, Xception, and Ensemble methods) have different levels of performance for each class classification.

Multiclass Classification for Self-Admitted Technical Debt Based on XGBoost

In software development, due to the demands from users or the limitations of time and resources, developers tend to adopt suboptimal solutions to achieve quick software development. In such a way, the released software usually involves not-quite-right code that is called technical debt, which will significantly decrease the quality of software and increase the maintenance cost. Recently, the concept of self-admitted technical debt (SATD) is proposed and refers to technical debt that is self-admitted by developers in code comments. Existing studies mainly focus on detecting technical debt by classifying code comments into either “SATD” or “non-SATD.” However, different types of SATD has different impacts on software maintenance and needs to be handled by different developers. Therefore, the detected SATD should be further classified so that developers can understand and remove technical debt better. In this article, we propose a new method based on eXtreme Gradient Boosting (XGBoost) to classify SATD into multiple classes. In our approach, we first preprocess the original code comments and adopt the easy data augmentation strategy to overcome the class unbalance problem. Then, chi-square is leveraged to select representative features from the textual feature set. Finally, we apply XGBoost to train a classifier and use the trained classifier to partition each comment into the corresponding class. We experimentally investigate the effectiveness of our approach on a public dataset, including 62 566 code comments from 10 open-source projects. Experimental results show that our approach achieves 56.66% in terms of macroaveraged precision, 59.07% in terms of macroaveraged recall, and 55.77% in terms of macroaveraged F-measure on average, and outperforms the natural language processing based method by 4.98%, 5.32%, and 3.17%, respectively. In addition, the experimental results also demonstrate that the data augmentation strategy is effective in improving the effectiveness of our approach.

Development of Soundness Diagnostic Model for Concrete Slab Using Bridge Inspection Data

With the aging of bridges, the efficiency of periodic inspections has become a problem. As issues with the continuing close visual inspection of bridges are surfacing, remote imaging systems are expected to become a new inspection method that replaces close visual inspection. The objective of the study is to develop a classification model of countermeasure categories using the results of past periodic inspections of bridges conducted by skilled inspectors. Focusing on concrete slabs, a model was constructed to classify the countermeasure categories based on the characteristics of the damage maps by random forest classification. As a result, it was possible to classify two classes of countermeasure categories with a macro-average precision rate of about 88%. It became clear that the degree of crack development and the number of cracks are the most important factors in the classification of judgment categories.

Automatic Multichannel Electrocardiogram Record Classification Using XGBoost Fusion Model.

There is an increasing demand for automatic classification of standard 12-lead electrocardiogram signals in the medical field. Considering that different channels and temporal segments of a feature map extracted from the 12-lead electrocardiogram record contribute differently to cardiac arrhythmia detection, and to the classification performance, we propose a 12-lead electrocardiogram signal automatic classification model based on model fusion (CBi-DF-XGBoost) to focus on representative features along both the spatial and temporal axes. The algorithm extracts local features through a convolutional neural network and then extracts temporal features through bi-directional long short-term memory. Finally, eXtreme Gradient Boosting (XGBoost) is used to fuse the 12-lead models and domain-specific features to obtain the classification results. The 5-fold cross-validation results show that in classifying nine categories of electrocardiogram signals, the macro-average accuracy of the fusion model is 0.968, the macro-average recall rate is 0.814, the macro-average precision is 0.857, the macro-average F1 score is 0.825, and the micro-average area under the curve is 0.919. Similar experiments with some common network structures and other advanced electrocardiogram classification algorithms show that the proposed model performs favourably against other counterparts in F1 score. We also conducted ablation studies to verify the effect of the complementary information from the 12 leads and the auxiliary information of domain-specific features on the classification performance of the model. We demonstrated the feasibility and effectiveness of the XGBoost-based fusion model to classify 12-lead electrocardiogram records into nine common heart rhythms. These findings may have clinical importance for the early diagnosis of arrhythmia and incite further research. In addition, the proposed multichannel feature fusion algorithm can be applied to other similar physiological signal analyses and processing.

Malicious code classification based on opcode sequences and textCNN network

A malicious code classification problem is essential for the network security. Malicious code is the most common means of network attack, which threatens user information and property security. An effective malicious code classification method can improve the efficiency of malicious code detection and the ability to discover new malicious code families. This study proposes a new malicious code classification method to analyze, classify, and detect malicious code. The semantic features of opcode sequences are extracted effectively by introducing the concept of word vectors. Furthermore, the extracted sequence is regarded as a text sentence and then introduced to a text convolutional neural network (textCNN) to identify malicious code families. The experimental results revealed that the model has more than 98% accuracy (with macro-average precision above 98.65% and macro-average recall approximately 98.66%) on the Microsoft Malware Challenge dataset conducted in 2015. Meanwhile, the accuracy of the model on the SOREL-20M dataset is 91.93%. Mostly call instructions are used to call the API, library functions, and other user-defined functions through which the behavior of malicious code is generally realized. Thus, selecting the block that contains call instructions as the key block will reduce the model training speed. After selecting the key block, on average, the number of opcodes on Microsoft Malware Challenge dataset is reduced by 39.07% and has a 98.18% accuracy rate, which is slightly lower than the result obtained by using all opcodes. The number of opcodes on the SOREL-20M dataset is reduced by 30.49% on average, and the accuracy is increased to 93.46%. Experimental results show that the proposed algorithm works well and outperforms the results obtained by using byte n-gram representation.

Generating decision support for alarm processing in cold supply chains using a hybrid k-NN algorithm

Real-time temperature monitoring is necessary in cold pharmaceutical supply chains (SCs), where exposures to extreme temperatures can lead to product quality deterioration. Temperature alarms (TAs) triggered by the current rule-based systems still require lengthy examinations before a suitable corrective measure (CM) can be chosen. However, provision of additional information relevant to TAs can expedite the examination process.In the related areas of recommender systems and false alarm/anomaly detection, k-nearest neighbors (k-NN) algorithm has proven to be successful because of its interpretability and ease of use. However, in the context of TA processing, it may suffer from some inherent limitations (i.e., varying neighborhood radius, unreliable classifications in sparse and noisy regions, and blindness to natural class boundaries). To overcome these limitations, we propose a hybrid k-NN (Hk-NN) algorithm based on the principles of local similarity and neighborhood homogeneity. It incorporates a two-step voting procedure with an entropy-optimized k-NN radius, decision trees with k-constrained leaves, and nearest neighbor predictions.We investigate 16,525 comments by alarm personnel for TAs in a pharmaceutical SC and encode them in terms of deviation causes and CMs (target features). We use SC data on cargo location, SC phase, sensor role, and temperature characteristics as predictor features for TA similarity estimation. In eight experimental setups, Hk-NN consistently outperforms k-NN with an optimized k in terms of accuracy, balanced accuracy, macro-average precision, recall, and specificity. At the same time, Hk-NN refrains from predicting observations, for which k-NN’s accuracy is close to a random guess.

A low-cost and efficient electronic nose system for quantification of multiple indoor air contaminants utilizing HC and PLSR

A novel quantification technique is presented for electronic nose (E-nose), which is based on a double-step strategy combined with hierarchical classifier (HC) and partial least squares regression (PLSR). With the tree structure of HC, the complexity of classifier training process can be reduced in the case of unbalanced samples. For each level of the class hierarchy, the extreme learning machine-based artificial neural network (ELM-ANN) is applied for classification. In order to improve the classification performance of ELM-ANN, the multiple time-domain features are selected as training inputs, and a novel optimization method of the number of hidden layer neurons is given. To validate the effectiveness of the presented quantification technique, an E-nose system is designed to quantify the gases including six toxic gases (hydrogen sulfide, carbon monoxide, ammonia, toluene, formaldehyde, acetone) and three kinds of binary gas mixtures. This presented hierarchical classifier has demonstrated outstanding performance for the identification of target gases, such as the macro-averaged precision for unlabeled data is improved from 80% to 92% compared with non-hierarchical classifiers. Furthermore, an excellent performance of concentration estimation is obtained utilizing PLSR, where average values of the coefficient of determination for training and test samples are equal to 0.957 and 0.927, respectively. Overall, our work demonstrates that the proposed approach is applicable in E-nose-based odor quantification.

Citation recommendation using semantic representation of cited papers’ relations and content

Citation recommendation can help researchers quickly find supplementary or alternative references in massive academic resources. Current research on citation recommendation mainly focuses on the citing papers, resulting in the enormous cited papers are ignored, including the relations among cited papers and their citation context cited in citing papers. Moreover, cited paper’s content is often denoted with its original title and abstract, which is hard to acquire and rarely considers different citation motivations. Furthermore, the most appropriate method for semantic representation of cited papers’ relations and content is uncertain. Therefore, this paper studies citation recommendation from the perspective of semantic representation of cited papers’ relations and content. Firstly, four forms of citation context are designed and extracted as cited papers’ content considering citation motivations, as well as co-citation relationships are extracted as cited papers’ relations. Secondly, 132 methods are designed for generating semantic vector of cited paper, including four network embedding methods, 16 methods by combining four text representation algorithms with four forms of citation content, and 112 fusion methods. Finally, similarity among cited papers is calculated for citation recommendation and a quantitative evaluation method based on link prediction is designed, to find the most appropriate form of citation content and the optimal method. The result shows that doc2vecC (Document to Vector through Corruption) with the form of CS&SS (Current Sentences and Surrounding Sentences) performs best, in which the AUC (Area Under Curve) and MAP (Macro Average Precision) reach 0.877 and 0.889 and have increased by 0.462 and 0.370 compared with the worst-performing method. This performance is slightly improved by parameters adjustment, and a case study is performed whose results have further proved the effectiveness of this method. In addition, among four forms of cited papers’ content, CS&SS performs best in almost all methods. Furthermore, the fusion methods not always perform better than the single methods, where doc2vecC (CS&SS) performs better than the best fusion method GCN (Graph Convolutional Network). These results not only prove the effectiveness of citation recommendation from the perspective of cited paper, but also provide helpful and useful suggestions for method selection and citation content selection. The data and conclusions can be extended to other text mining-related tasks. Simultaneously, it is a preliminary research which needs to be further studied in other domains using emerging semantic representation methods.

Personality Traits Recognition on Social Network - Facebook

For the natural and social interaction it is necessary to understand human behavior. Personality is one of the fundamental aspects, by which we can understand behavioral dispositions. It is evident that there is a strong correlation between users’ personality and the way they behave on online social network (e.g., Facebook). This paper presents automatic recognition of Big-5 personality traits on social network (Facebook) using users’ status text. For the automatic recognition we studied different classification methods such as SMO (Sequential Minimal Optimization for Support Vector Machine), Bayesian Logistic Regression (BLR) and Multinomial Naïve Bayes (MNB) sparse modeling. Performance of the systems had been measured using macro-averaged precision, recall and F1; weighted average accuracy (WA) and un-weighted average accuracy (UA). Our comparative study shows that MNB performs better than BLR and SMO for personality traits recognition on the social network data.

Performance evaluation of a prescription medication image classification model: an observational cohort

Technology assistance of pharmacist verification tasks through the use of machine intelligence has the potential to detect dangerous and costly pharmacy dispensing errors. National Drug Codes (NDC) are unique numeric identifiers of prescription drug products for the United States Food and Drug Administration. The physical form of the medication, often tablets and capsules, captures the unique features of the NDC product to help ensure patients receive the same medication product inside their prescription bottle as is found on the label from a pharmacy. We report and evaluate using an automated check to predict the shape, color, and NDC for images showing a pile of pills inside a prescription bottle. In a test set containing 65,274 images of 345 NDC classes, overall macro-average precision was 98.5%. Patterns of incorrect NDC predictions based on similar colors, shapes, and imprints of pills were identified and recommendations to improve the model are provided.

A comparative study of machine learning classifiers for injury severity prediction of crashes involving three-wheeled motorized rickshaw

Motorcycles and motorcyclists have a variety of attributes that have been found to be a potential contributor to the high liability of vulnerable road users (VRUs). Vulnerable Road Users (VRUs) that include pedestrians, bicyclists, cycle-rickshaw occupants, and motorcyclists constitute by far the highest share of road traffic accidents in developing countries. Motorized three-wheeled Rickshaws (3W-MR) is a popular public transport mode in almost all Pakistani cities and is used primarily for short trips to carry passengers and small-scale goods movement. Despite being an important mode of public transport in the developing world, little work has been done to understand the factors affecting the injury severity of three-wheeled motorized vehicles. Crash injury severity prediction is a promising research target in traffic safety. Traditional statistical models have underlying assumptions and predefined associations, which can yield misleading results if flouted. Machine learning(ML) is an emerging non-parametric method that can effectively capture the non-linear effects of both continuous and discrete variables without prior assumptions and achieve better prediction accuracy. This research analyzed injury severity of three-wheeled motorized rickshaws (3W-MR) using various machine learning-based identification algorithms, i.e., Decision jungle (DJ), Random Forest (RF), and Decision Tree (DT). Three years of crash data (from 2017 to 2019) was collected from Provincial Emergency Response Service RESCUE 1122 for Rawalpindi city, Pakistan. A total of 2,743 3W-MR crashes were reported during the study period that resulted in 258 fatalities. The predictive performance of proposed ML models was assessed using several evaluation metrics such as overall accuracy, macro-average precision, macro-average recall, and geometric means of individual class accuracies. Results revealed that DJ with an overall accuracy of 83.7 % outperformed the DT and RF-based on a stratified 10-fold cross-validation approach. Finally, Spearman correlation analysis showed that factors such as the lighting condition, crashes involving young drivers (aged 20–30 years), facilities with high-speed limits (over 60 mph), weekday, off-peak, and shiny weather conditions were more likely to worsen injury severity of 3W-MR crashes. The outcomes of this study could provide necessary and essential guidance to road safety agencies, particularly in the study area, for proactive implementation of appropriate countermeasures to curb road safety issues pertaining to three-wheeled motorized vehicles.

Review prognosis system to predict employees job satisfaction using deep neural network

AbstractWith the multitude of companies that flourish today, job seekers want to join companies with highly satisfied employees. So, job satisfaction prediction is an important task that helps companies in sustaining or redesigning employee policies. Such predictions not only help in reducing employee attrition but also affect the goodwill and reputation of a company. The higher satisfaction level of current employees attracts potential new employees and confirms the positive policies of a company toward its employees. Job satisfaction prediction can be performed using employee reviews either manually or via automated machine learning algorithms. This study first evaluates four widely used machine learning algorithms, that is, random forest, logistic regression, support vector classifier, and gradient boosting, and then proposes a deep learning model to predict employee job satisfaction level. Experiments are carried out on a dataset that contains text reviews from the employees of Google, Facebook, Amazon, Microsoft, and Apple. Three feature extraction methods are analyzed as well including term frequency‐inverse document frequency (TF‐IDF), bag‐of‐words (BOW), and global vector for word representation (GloVe). Performance is evaluated using accuracy, precision, recall, F1 score, as well as, macro average precision, and weighted average. The performance of the proposed model is compared with state‐of‐the‐art deep learning models. Results demonstrate that the proposed model performs better than both the machine learning and state‐of‐the‐art approaches.

Alert Classification for the ALeRCE Broker System: The Light Curve Classifier

Abstract We present the first version of the Automatic Learning for the Rapid Classification of Events (ALeRCE) broker light curve classifier. ALeRCE is currently processing the Zwicky Transient Facility (ZTF) alert stream, in preparation for the Vera C. Rubin Observatory. The ALeRCE light curve classifier uses variability features computed from the ZTF alert stream and colors obtained from AllWISE and ZTF photometry. We apply a balanced random forest algorithm with a two-level scheme where the top level classifies each source as periodic, stochastic, or transient, and the bottom level further resolves each of these hierarchical classes among 15 total classes. This classifier corresponds to the first attempt to classify multiple classes of stochastic variables (including core- and host-dominated active galactic nuclei, blazars, young stellar objects, and cataclysmic variables) in addition to different classes of periodic and transient sources, using real data. We created a labeled set using various public catalogs (such as the Catalina Surveys and Gaia DR2 variable stars catalogs, and the Million Quasars catalog), and we classify all objects with ≥6 g-band or ≥6 r-band detections in ZTF (868,371 sources as of 2020 June 9), providing updated classifications for sources with new alerts every day. For the top level we obtain macro-averaged precision and recall scores of 0.96 and 0.99, respectively, and for the bottom level we obtain macro-averaged precision and recall scores of 0.57 and 0.76, respectively. Updated classifications from the light curve classifier can be found at the ALeRCE Explorer website (http://alerce.online).

Improving blood cells classification in peripheral blood smears using enhanced incremental training

Peripheral Blood Smear (PBS) analysis is a vital routine test carried out by medical specialists to assess some health aspects of individuals. The automation of blood analysis has attracted the attention of researchers in recent years, as it will not only save time, money and reduce errors, but also protect and save lives of front-line workers, especially during pandemics. In this work, deep neural networks are trained on a synthetic blood smears dataset to classify fifteen different white blood cell and platelet subtypes and morphological abnormalities. For classifying platelets, a hybrid approach of deep learning and image processing techniques is proposed. This approach improved the platelet classification accuracy and macro-average precision from 82.6% to 98.6% and 76.6%–97.6% respectively.Moreover, for white blood cell classification, a novel scheme for training deep networks is proposed, namely, Enhanced Incremental Training, that automatically recognises and handles classes that confuse and negatively affect neural network predictions. To handle the confusable classes, we also propose a procedure called “training revert”. Application of the proposed method has improved the classification accuracy and macro-average precision from 61.5% to 95% and 76.6%–94.27% respectively.