Classification Results Research Articles

Visualization and classification of mushroom species with multi-feature fusion of metaheuristics-based convolutional neural network model

Determining the correct mushroom species with the necessary ecological characteristics is critical to continue mushroom production, which is essential in gastronomy. The mushroom farmers and collectors technique may help identify toxic mushrooms by detecting poisonous mushrooms using images of different mushroom species with distinctive morphological features. However, it can be not easy to distinguish between species. This paper used a dataset of 6714 mushroom images obtained from nine different mushroom species to classify the mushroom species. For a more straightforward comprehension of mushroom images and feature extraction by reanalysis of data sets, data visualization was performed using Grad-CAM, LIME, and Heatmap methods. Residual block-based Convolutional Neural Network (CNN) architectures are trained to automatically classify the concatenated feature map obtained from the Grad-CAM, LIME, and Heatmap methods. After extracting the deep features of the images from each architecture, the Atom Search Optimization (ASO) algorithm has been used to select the most distinctive features. The 6714×9000 size of the concatenated feature map was reduced to 6714×600 using the ASO algorithm. Classification results were evaluated using six different classifiers based on the feature map obtained to determine the mushroom species. The nine classes of mushroom species were classified successfully with 95.45 % accuracy using the proposed model with the ASO algorithm and KNN classifier. The methodology introduces novel visualization techniques for interpreting CNN-based models’ decisions in mushroom species classification tasks. Using metaheuristics-based CNN models with multi-feature fusion techniques allows the model to leverage diverse sources of information, potentially enhancing its ability to discriminate between mushroom species and achieve higher classification accuracy than existing methods. This study can advance the mushroom species classification field by introducing new methodologies, improving classification accuracy, providing insights into model interpretability, and facilitating knowledge transfer to related fields.

Classifying UK charities’ activities by charitable cause: a new classification system

This article presents a new system for classifying UK charities’ activities according to their charitable purposes. It also outlines our attempts to use keyword search rules to apply these classifications to the various UK charity registers. The classification results and code, which are made freely available online, help to address the limitations of existing classification schemes in the UK context. Depending on the scheme, these include a lack of detail and coverage of important subsectors, a lack of systematic data collection and limits on the number of classifications per charity. We discuss the pros and cons of different approaches and show that the keyword searching method provides a sufficiently accurate and transparent approach. We also present some preliminary results on how commonly each ‘tag’ is matched against UK charities, as well as exploring how the results compare to existing classifications in the register of charities for England and Wales.

Discrimination and analytical profiling of colored printed documents using ATR-FTIR spectroscopy coupled with explorative and predictive statistical analysis: Part I

Printed documents are a common form of evidence in forensic document examination. The integration of spectroscopy with chemometrics have evolved evidential analytical interpretation of printing inks. However, we report the first ever study that explores the examination of both black and colored printed documents combined with explorative Principal Component Analysis (PCA) and supervised techniques viz. Soft independent modelling of class analogy (SIMCA) and Partial Least Square- Discriminant Analysis (PLS-DA). The study investigated 74 (40 Ink-based and 34 Toner- based) colored printed document samples using ATR-FTIR to discriminate and determine the source of origin of an unknown printed document using a non-destructive approach. Qualitative analysis by ATR- FTIR indicated the presence of polystyrene, bisphenol A and acrylates as the common binder polymers in the samples. The study was also able to obtain pigment information like presence of PR 57 and PR 146 in magenta, Carbon black in black, Copper Phthalocyanine and PB 15 in Cyan and PY 74 in yellow colored printed samples. Further, PCA has been used as an explorative technique that showed a variance of 97 % in the dataset and indicating that the color Cyan contributes to the maximum classification accuracy. SIMCA has been used as a supervised method to classify the known and test samples to their respective defined classes. However, SIMCA could only classify Toner-based samples in their respective class and inconclusive results were obtained in case of Ink-based samples. Finally, PLS-DA was also used to classify the two class of samples which resulted in a discrimination accuracy of 98.6 %. The derived model was also used for validation study on blind test samples which provided 100 % classification results.

Association between METS-IR and heart failure: a cross-sectional study.

Prior research has indicated the importance of insulin resistance in the development of heart failure (HF). The metabolic score for insulin resistance (METS-IR), a novel measure for assessing insulin resistance, has been found to be associated with cardiovascular disease (CVD). Nevertheless, the relationship between METS-IR and heart failure remains uncertain. This cross-sectional study collected data from the 2007-2018 National Health and Nutrition Examination Survey (NHANES). Multivariable logistic regression analysis and smoothing curve fitting were performed to explore the relationship between METS-IR and the risk of heart failure. Subgroup analysis and receiver operating characteristic (ROC) curve analysis were also conducted. A total of 14772 patients were included, of whom 485 (3.28%) had heartfailure. We observed a significant positive association between METS-IR and the risk of heart failure in a fully adjusted model (per 1-unit increment in METS-IR: OR: 2.44; 95% CI: 1.38, 4.32). Subgroup analysis and interaction tests revealed nosignificant influence on this relationship. A saturation effect and nonlinearrelationship between METS-IR and heart failure risk were found using a smoothing curve fitting analysis. The relationship was represented by a J-shaped curve with an inflection point at 40.966. The results of our study indicated a J-shaped association between METS-IR and HF in adults in the United States. METS-IR may be a promising novel index for predicting the risk of heart failure. More longitudinal studies are needed to further verify causal relationships and validate the results in different classifications of heart failure populations.

Analysis of spatiotemporal changes mechanism of cell wall biopolymers and monosaccharide components in kiwifruit during Botryosphaeria dothidea infection

To further reveal the interaction mechanism between plants and pathogens, this study used confocal Raman microscopy spectroscopy (CRM) combined with chemometrics to visualize the biopolymers distribution of kiwifruit cell walls at different infection stages at the cellular micro level. Simultaneously, the changes in the content of various monosaccharides in fruit were studied at the molecular level using high-performance liquid chromatography (HPLC). There were significant differences in the composition of various nutrient components in the cell wall structure of kiwifruit at different infection times after infection by Botryosphaeria dothidea. PCA could cluster samples with infection time of 0–9 d into different infection stages, and SVM was used to predict the PCA classification results, the accuracy >96 %. Multivariate curve resolution-alternating least squares (MCR-ALS) helped to identify single substance spectra and concentration signals from mixed spectral signals. The pure substance chemical imaging maps of low methylated pectin (LMP), high methylated pectin (HMP), cellulose, hemicellulose, and lignin were obtained by analyzing the resolved concentration data. The imaging results showed that the lignin content in the kiwifruit cell wall increased significantly to resist pathogens infection after the infection of B. dothidea. With the development of infection, B. dothidea decomposed various substances in the host cell walls, allowing them to penetrate the interior of fruit cells. This caused significant changes in the form, structure, and distribution of various chemicals on the fruit cell walls in time and space. HPLC showed that glucose was the main carbon source and energy substance obtained by pathogens from kiwifruit during infection. The contents of galactose and arabinose, which maintained the structure and function of the fruit cell walls, decreased significantly and the cell wall structure was destroyed in the late stage of pathogens infection. This study provided a new perspective on the cellular structure changes caused by pathogenic infection of fruit and the defense response process of fruit and provided effective references for further research on the mechanisms of host-pathogen interactions in fruit infected by pathogens.

Spatiotemporal Properties of Common Semantic Categories for Words and Pictures.

The timing of semantic processing during object recognition in the brain is a topic of ongoing discussion. One way of addressing this question is by applying multivariate pattern analysis to human electrophysiological responses to object images of different semantic categories. However, although multivariate pattern analysis can reveal whether neuronal activity patterns are distinct for different stimulus categories, concerns remain on whether low-level visual features also contribute to the classification results. To circumvent this issue, we applied a cross-decoding approach to magnetoencephalography data from stimuli from two different modalities: images and their corresponding written words. We employed items from three categories and presented them in a randomized order. We show that if the classifier is trained on words, pictures are classified between 150 and 430 msec after stimulus onset, and when training on pictures, words are classified between 225 and 430 msec. The topographical map, identified using a searchlight approach for cross-modal activation in both directions, showed left lateralization, confirming the involvement of linguistic representations. These results point to semantic activation of pictorial stimuli occurring at ≈150 msec, whereas for words, the semantic activation occurs at ≈230 msec.

ANALISIS PERKEMBANGAN LAHAN TERBANGUN BERDASARKAN METODE SUPERVISED CLASSIFICATION MENGGUNAKAN GOOGLE EARTH ENGINE (STUDI KASUS: DESA CIPUTI, KECAMATAN PACET, KAB.CIANJUR)

Monitoring the development of built-up areas can be done by observing remote sensing time series data such as Satellite Imagery. Google Earth Engine (GEE) makes it easy for users to access satellite image data, data processing and data analysis. GEE provides various machine learning algorithms to extract land cover data. This research aims to analyze the development of built-up areas using time series of remote sensing data, namely Sentinel 2A images recorded in 2019 and 2023 and comparing Random Forest (RF), Classification and Regression Tree (CART), Support Vector Machine (SVM) and Gradient Tree Boost (GTB) algorithms and predicts built-up areas in 2027. Based on the results of this research, RF is the algorithm with the highest accuracy in mapping land cover in Ciputri Village with an Overall Accuracy (OA) of 92% and a Kappa Coefficent (KC) of 0.89 in both the 2019 and 2023 classification results, while the lowest accuracy is the SVM algorithm. A comparison of the built-up land area between the 2019 and 2023 classification results shows a decrease in the built-up land area of 3.08 ha. Meanwhile, the prediction results for 2027 show an increase in built-up areas to 114.72 ha.

Comparison of Naive Bayes Classification Methods Without and With Kernel Density Estimation

Halal certification is important to give confidence to Muslim consumers around the world regarding the halalness of products. The Halal Product Assurance Organizing Body (BPJPH) is the official auditor in Indonesia that is responsible for the halal certification process. This study aims to address the need for verification and validation of data for halal certification applications in Indonesia by using the data science approach and machine learning technology. In this study, the Naïve Bayes classification method was used to optimize the data verification and validation process. However, this method needs to be improved by applying optimization methods such as Kernel Density Estimation (KDE) to improve classification results. The results showed that the Naïve Bayes classification method with KDE optimization produced better performance than the Naïve Bayes method without optimization. The performance of the Naïve Bayes classification model without optimization achieves 87.6% Accuracy, 85.4% Recall, 88.8% Precision, and 87.1% Fmeasure. Meanwhile, the Naïve Bayes classification model with KDE optimization achieves 97.5% Accuracy, 95.9% Recall, 98.9% Precision, and 97.8% Fmeasure. Thus, it can be concluded that the Naïve Bayes classification algorithm with KDE optimization results in a performance increase of 9.9% compared to the Naïve Bayes method without optimization. This research has important implications in handling complex and non-normally distributed data and providing solutions for BPJPH in the process of verifying halal certification.

An Approach for Efficient and Accurate Phishing Website Prediction Using Improved ML Classifier Performance for Feature Selection

The article discusses the use of machine learning (ML) to combat phishing websites, which are deceptive sites that mimic trusted entities to steal sensitive information. This is why the continued invention of methods of identifying and counteracting phishing threats is beneficial. Such attacks pose significant risks to the integrity of online security. To enhance the success rate and specificity of predicting phishing websites, this study proposes a new approach that utilizes machine learning algorithms. To enhance the methods mentioned above and achieve better results in classification and better prediction of customer behaviour, the main points exposed to further transformations are increasing classifier accuracy and selecting an optimal feature space. Traditional anti-phishing strategies like blacklisting and heuristic searches often have slow detection times and high false positive rates. The article introduces a novel feature selection method to extract highly correlated features from datasets, thereby enhancing classifier accuracy. Using six feature selection techniques on a phishing dataset, it evaluates eight classifiers, including SVM, Logistic Regression, Random Forest, and others. The study finds that the Random Forest classifier combined with the Chi-2 feature selection method significantly improves model accuracy, achieving up to 96.99%.

경계선급 지적 기능성(BIF) 초등학생의 학교생활적응에 필요한 지원요소 탐색

The purpose of this study is to explore the support elements necessary for the adaptation of border-level intellectual functional elementary school students to school life recognized by field educators. One-on-one individual interviews were conducted with 10 public elementary school teachers and 10 teachers from alternative schools operated for border-level intellectual functional students. The results of similarity classification and importance evaluation were both visualized by a multidimensional scale analysis and hierarchical cluster analysis based on the extracted statements. As a result of this study, first, support in the relational aspect should be expanded. Support for border-level intellectual functional students, which has thus far been biased towards the learning aspect, should be provided in a balanced way when regarding the relational aspect. Second, support for establishing awareness of the educational community object is necessary; this is so that the educational community can fully understand the characteristics of border-level intellectual functionality that are distinguished from other disabilities. Third, cooperative support between several educational entities with homeroom teachers and parents as the core subjects is needed. In order for border-level intellectual functional elementary school students to adapt to school life, it should be recognized that support focus needs to be tailored towards how to create an educational environment including homes and schools surrounding them, rather than how to improve their functions.

Deep Learning Algorithm Analysis of Potato Disease Classification for System on Chip Implementation

Recently, every aspect of human existence has been affected by modern technologies including agriculture. The broad range of crops has witnessed setbacks in different capacities due to climate change among other factors, hence leading to diseases and infections; thereby leading to negatively impacted nutrition. This work uses a deep learning algorithm to investigate the classification of potato diseases as a case study which can be leveraged on by other agricultural products for reference. Potatoes play a vital role in global agriculture, constituting a significant portion of fresh produce consumption. In southern Africa, potatoes hold particular importance, comprising 39% of total fresh produce consumption. However, the industry faces challenges from diseases such as early blight and late blight, resulting in substantial yield losses. Early detection is crucial, prompting the exploration of Convolutional Neural Networks (CNNs) for their disease classification capabilities. CNNs have shown excellent results in plant disease classification based on image data set. This study proposes the potential of aligning existing software-based Central Processing Units (CPUs) and Graphic Processing Units (GPUs) with FPGA-based potato disease classification using CNNs. To this end, five CNN models were trained on the Plant Village dataset for analysis. The models were compared using various performance metrics such as sensitivity, F1-Score, precision and recall. Based on the analysis the most suitable CNN model for FPGA implementation was selected. The chosen model was optimized for size using quantization technique. Subsequently the model was converted to Hardware Description Language (HDL) using Vitis High level synthesis (HLS) tool performance in terms of resources required to implement the model in System on Chip (SoC) was analysed.

A COMPUTER-AIDED BRAIN GLIOMA IDENTIFICATION USING MRI BASED TEXTURE ANALYSIS

Brain gliomas are deadly tumors that are discovered after a stressful, lengthy, and difficult procedure. Radiology is a broad and varied field that detects brain tumors, but interpreting radiological pictures of the brain needs advanced training, experience, and subject-matter expertise. The endeavor is challenging due to the wide variation in brain tumor tissues among individuals and similar cases within normal tissues. Magnetic resonance imaging (MRI)-based computer-aided biomedical image processing solves the challenges associated with brain tumor localization and identification while simultaneously addressing the shortage of qualified radiologists. This study proposes a computer-aided brain glioma identification (CABGD) model for brain glioma identification using the texture analysis of brain MRIs. The proposed model makes use of machine learning classifiers and brain MRI data. There were 200 MRIs of both normal and glioma brains in the experimental dataset. Firstly, the MRI dataset was pre-processed to crop the MRIs, size equalization, and gray-level conversion. Next, noises were removed by applying filters. Two ROIs of the sizes (10 ×10) were taken on each MRI after the tumor region was segmented. After extracting COM texture features from each ROI, thirty optimal features were obtained through a compound supervised feature selection, blend of Fisher (FSHR), probabilistic model of error (POE), average correlation (AVC), and mutual information (MUI). The optimal feature brain MRI dataset was classified into glioma and normal brain by applying machine learning classifiers named Bayas Net (BN), logistic model tree (LMT), and partial decision tree (PART); the classification results of the NB, LMT, and PART classifiers were 79.75%, 82.75, and 85%, respectively.

Effect of Bio-fortified Vitamin A Cassava on Poverty Status of Farming Households in Nigeria, Evidence from Oyo State.

ABSTRACT Profitable farming is a prime target of all farmers regardless of the scale of production adopted and the veritable opportunity to attain the threshold of bumper harvest is not unconnected with the adoption of new technology. Bio-fortified vitamin A cassava remains the best option for farmers to experience improvement in cassava performance in both the quality and quantity terms and they are encouraged to adopt it in order to improve their households’ welfare. The survey was launched to investigate the effect of bio-fortified vitamin A cassava on the poverty status of the farming households in Nigeria using Oyo State as a case study. A Multistage sampling technique was used to sample a total of 186 cassava farmers comprising 92 adopters and 94 non-adopters of bio-fortified vitamin A cassava variety using copies of well-structured questionnaires. Parametric tools used were descriptive statistics and probit regression for socioeconomic characteristics and adoption status analysis. The socioeconomic results showed that both the adopters and non-adopters were in the productive ages of 42 years and 59 years respectively. The majorities of farmers in the two categories were married and had formal education. Years of experience possessed by them were 11 years and 18 years while both groups were members of associations. Poverty classification result showed 41.3% of adopters being poor while 83.0% of non-adopters were poor. Marital status (2.1271<p=0.05), education (1.1510; p<0.1), seasonal income (0.0757; p=0.05) and poverty status (I.6019; p<0.1) had influence on the farmers’ adoption status while lack of credit was voted highest among the factors militating against adoption of bio-fortified vitamin A cassava. It is therefore recommended that stakeholders in the crop enterprise should make frantic efforts in providing credit and other basic inputs that will enhance improved vitamin A adoption.

Multiclass Text Classification of Indonesian Short Message Service (SMS) Spam using Deep Learning Method and Easy Data Augmentation

The ease of using Short Message Service (SMS) has brought the issue of SMS spam, characterized by unsolicited and unwanted. Many studies have been conducted utilizing machine learning methods to build models capable of classifying SMS Spam to overcome this problem. However, most of these studies still rely on traditional methods, with limited exploration of deep learning-based approaches. Whereas traditional methods have a limitation compared to deep learning, which performs manual feature extraction. Moreover, many of these studies only focus on binary classification rather than multiclass SMS classification which can provide more detailed classification results. The aim of this research is to analyze deep learning model for multiclass Indonesian SMS spam classification with six categories and to assess the effectiveness of the text augmentation method in addressing data imbalace issues arising from the increased number of SMS categories. The research method used were Indonesian version of Bidirectional Encoder Representations from Transformers (IndoBERT) model and exploratory data analysis (EDA) augmentation technique to address imbalance dataset issue. The evaluation is conducted by comparing the performance of the IndoBERT model on the dataset and applying EDA techniques to enhance the representation of minority classes. The result of this research shows that IndoBERT achieves 91% accuracy rate in classifying SMS spam. Furthermore, the use of EDA technique results in significant improvement in f1-score, with an average 12% increase in minority classes. Overall model accuracy also improves to 93% after EDA implementation. This research concludes that IndoBERT is effective for multiclass SMS spam classification, and the EDA is beneficial in handling imbalanced data, contributing to the enhancement of model performances.

Multimodal Machine Learning for Prognosis and Survival Prediction in Renal Cell Carcinoma Patients: A Two-Stage Framework with Model Fusion and Interpretability Analysis

Current medical limitations in predicting cancer survival status and time necessitate advancements beyond traditional methods and physical indicators. This research introduces a novel two-stage prognostic framework for renal cell carcinoma, addressing the inadequacies of existing diagnostic approaches. In the first stage, the framework accurately predicts the survival status (alive or deceased) with metrics Accuracy, Precision, Recall, and F1 score to evaluate the effects of the classification results, while the second stage focuses on forecasting the future survival time of deceased patients with Root Mean Square Error and Mean Absolute Error to evaluate the regression results. Leveraging popular machine learning models, such as Adaptive Boosting, Extra Trees, Gradient Boosting, Random Forest, and Extreme Gradient Boosting, along with fusion models like Voting, Stacking, and Blending, our approach significantly improves prognostic accuracy as shown in our experiments. The novelty of our research lies in the integration of a logistic regression meta-model for interpreting the blending model’s predictions, enhancing transparency. By the SHapley Additive exPlanations’ interpretability, we provide insights into variable contributions, aiding understanding at both global and local levels. Through modal segmentation and multimodal fusion applied to raw data from the Surveillance, Epidemiology, and End Results program, we enhance the precision of renal cell carcinoma prognosis. Our proposed model provides an interpretable analysis of model predictions, highlighting key variables influencing classification and regression decisions in the two-stage renal cell carcinoma prognosis framework. By addressing the black-box problem inherent in machine learning, our proposed model helps healthcare practitioners with a more reliable and transparent basis for applying machine learning in cancer prognostication.

Pulmonary High Resolution Computed Tomography Findings of COVID-19 Patients Using CO-RADS Classification: A Study on Pakistani Population

Background: This study aimed to use high-resolution computed tomography (HRCT) and the CO-RADS grading system to diagnose COVID-19 patients and determine severity. Methodology: A retrospective study analyzed data from 280 patients at Aziz Bhatti Shaheed Hospital Gujrat between January and August 2021. Informed consent and ethical approval were secured. Data collection used a 64-slice multi-detector CT scanner, including patients with positive and negative COVID-19 polymerase chain reaction (PCR) findings. Data were analyzed using SPSS version 23. Results: The study included more males (170, 60.7%) than females (110, 39.3%), with the most affected age group being 45-64 years. HRCT findings showed ground glass opacities in 212 cases (35.6%), consolidations in 158 cases (26.5%), Crazy paving sign in 54 cases (9.1%), pulmonary lesions in 54 cases (9.1%), interstitial lung disease in 42 cases (7.0%), pulmonary infiltrates in 32 cases (5.4%), atelectasis in 26 cases (4.4%), and pulmonary fibrosis in 18 cases (3.0%). CO-RADS classification results were: CO-RADS 1 (5, 1.8%), CO-RADS 2 (11, 3.9%), CO-RADS 3 (101, 36.1%), CO-RADS 4 (68, 24.3%), CO-RADS 5 (56, 20%), and CO-RADS 6 (39, 13.9%). PCR results were positive only in severe cases (starting from CO-RADS 2), while HRCT detected pulmonary findings even in less severe cases (starting from CO-RADS 1). Conclusion: High-resolution computed tomography accurately and quickly identifies COVID-19 infections, even when PCR findings are negative. The CO-RADS technique effectively determines the severity and spread of COVID-19. CO-RADS 3 was the most frequently reported category, with typical findings including peripheral ground glass opacities and bilateral pleural consolidations.

Physics-informed interactive network for hyperspectral image classification

In hyperspectral image (HSI) classification, convolutional neural networks (CNNs) have made great progress. In recent CNN-based HSI classification methods, spatial attention mechanism has been widely used to emphasize more salient features and suppress less useful ones. However, these methods tend to utilize plug-and-play attention modules which ignore the physical meaning of HSI pixels. To reflect the physical construction of HSI pixels in the attention mask, we propose a physics-informed interactive network (PI2Net) integrating a spatial guided spectral component stream (SGSC-stream) and a spectral constrained spatial attention stream (SCSA-stream). The SGSC-stream can obtain more expressive abundance features of the input HSI cubes through a component analysis block. The SCSA-stream completes the prediction of hyperspectral data in three-stage features extraction cells and a classification cell. To connect and interact between these two streams, a component-aware attention module (CAAM) and a class-wise endmember activation gate (CWEAG) are designed. The CAAM measures the similarity of abundance features to generate a component-aware attention mask, so as to learn the component-based relationships between pixels. Then, the multiscale attention masks are embedded into different stages of the SCSA-stream. The CWEAG is proposed to iteratively generate more accurate abundance features with the classification result of SCSA-stream. Experimental results based on three common hyperspectral data sets demonstrate that PI2Net outperforms several state-of-the-art methods in terms of qualitative and quantitative analysis.

Non-destructive prediction of fertility and sex in chicken eggs using the short wave near-infrared region

The objective of this study was to evaluate the ability of a handheld near-infrared device (900–1600 nm) to predict fertility and sex (male and female) traits in-ovo. The NIR reflectance spectra of the egg samples were collected on days 0, 7, 14 and 18 of incubation and the data was analysed using principal component analysis (PCA), linear discriminant analysis (LDA) and support vector machines classification (SVM). The overall classification rates for the prediction of fertile and infertile egg samples ranged from 73 % to 84 % and between 93 % to 95 % using LDA and SVM classification, respectively. The highest classification rate was obtained on day 7 of incubation. The classification between male and female embryos achieved lower classification rates, between 62 % and 68 % using LDA and SVM classification, respectively. Although the classification rates for in-ovo sexing obtained in this study are higher than those obtained by chance (50 %), the classification results are currently not sufficient for industrial in-ovo sexing of chicken eggs. These results demonstrated that short wavelengths in the NIR range may be useful to distinguish between fertile and infertile egg samples at days 7 and 14 during incubation.

Research on the classification of ancient silicate glass artifacts based on machine learning

AbstractClassifying cultural relics has always been a major challenge for archaeologists. Using glass artifacts as the research object, a classification model for glass artifacts was constructed using decision trees, support vector machines, and logistic regression methods based on their patterns, colors, surface weathering conditions, types, and composition ratios. Three models were used to identify the types of unknown glass artifacts. A subclassification model for high‐potassium glass and lead barium glass was established using the K‐means clustering method. The elbow method and average contour method were used to determine the optimal number of clusters, and the decision tree model was named based on the characteristics of the cluster center components. The research results indicate that the three models yield consistent identification results for unknown types of glass relics, and the classification results are good. Lead barium glass and high‐potassium glass can be divided into three and six subclasses, respectively, and the naming of the subclass decision tree is reasonable. The identification method for ancient glass relics in this article is highly practical and can provide a reference for the classification and identification of other component data.

Fluorescence excitation-scanning hyperspectral imaging with scalable 2D–3D deep learning framework for colorectal cancer detection

Colorectal cancer is one of the top contributors to cancer-related deaths in the United States, with over 100,000 estimated cases in 2020 and over 50,000 deaths. The most common screening technique is minimally invasive colonoscopy using either reflected white light endoscopy or narrow-band imaging. However, current imaging modalities have only moderate sensitivity and specificity for lesion detection. We have developed a novel fluorescence excitation-scanning hyperspectral imaging (HSI) approach to sample image and spectroscopic data simultaneously on microscope and endoscope platforms for enhanced diagnostic potential. Unfortunately, fluorescence excitation-scanning HSI datasets pose major challenges for data processing, interpretability, and classification due to their high dimensionality. Here, we present an end-to-end scalable Artificial Intelligence (AI) framework built for classification of excitation-scanning HSI microscopy data that provides accurate image classification and interpretability of the AI decision-making process. The developed AI framework is able to perform real-time HSI classification with different speed/classification performance trade-offs by tailoring the dimensionality of the dataset, supporting different dimensions of deep learning models, and varying the architecture of deep learning models. We have also incorporated tools to visualize the exact location of the lesion detected by the AI decision-making process and to provide heatmap-based pixel-by-pixel interpretability. In addition, our deep learning framework provides wavelength-dependent impact as a heatmap, which allows visualization of the contributions of HSI wavelength bands during the AI decision-making process. This framework is well-suited for HSI microscope and endoscope platforms, where real-time analysis and visualization of classification results are required by clinicians.