Linear Discriminant Analysis Research Articles

Characterizing variability in geochemistry and mineralogy of western US dust sources

Dust events originate from multiple sources in arid and semi-arid regions, making it difficult to quantify source contributions. Dust geochemical/mineralogical composition, if the sources are sufficiently distinct, can be used to quantify the contributions from different sources. To test the viability of using geochemical and mineralogical measurements to separate dust-emitting sites, we used dust samples collected between 2018 and 2020 from ten National Wind Erosion Research Network (NWERN) sites that are representative of western United States (US) dust sources. Dust composition varied seasonally at many of the sites, but within-site variability was smaller than across-site variability, indicating that the geochemical signatures are robust over time. It was not possible to separate all the sites using commonly applied principal component analysis (PCA) and cluster analysis because of overlap in dust geochemistry. However, a linear discriminant analysis (LDA) successfully separated all sites based on their geochemistry, suggesting that LDA may prove useful for separating dust sources that cannot be separated using PCA or other methods. Further, an LDA based on mineralogical data separated most sites using only a limited number of mineral phases that were readily explained by the local geologic setting. Taken together, the geochemical and mineralogical measurements generated distinct signatures of dust emissions across NWERN sites. If expanded to include a broader range of sites across the western US, a library of geochemical and mineralogical data may serve as a basis to track and quantify dust contributions from these sources.

Comprehensive Performance Evaluation of Machine Learning Algorithms for detecting DDoS attacks in SDN

<div align="center">Software-Defined Networking (SDN) revolutionizes networking by separating control logic and data forwarding, enhancing security against threats like Distributed Denial of Service (DDoS) attacks. These attacks flood control plane bandwidth, causing SDN network failures. Recent studies emphasize the efficacy of machine learning and statistical approaches in identifying and mitigating these security risks. However, there has been a lack of focus on employing ensembling techniques, amalgamating diverse machine learning models, selecting pertinent features, and utilizing oversampling techniques to balance categorical data. Our study evaluates 20 machine-learning models, emphasizing feature engineering and addressing class imbalance using Synthetic Minority Oversampling TEchnique (SMOTE). The results indicate that Ensemble methods such as LGBM Classifier, Random Forest Classifier, XGB Classifier, Decision Tree Classifier attained near-perfect scores (almost 100%) across all metrics, suggesting potential overfitting. Conversely, models like AdaBoost Classifier, K-Neighbors Classifier, and SVC exhibited slightly lower (99%) but realistic performance, underscoring the intricacy of accurate prediction in cybersecurity. Simpler models, including Logistic Regression, Linear Discriminant Analysis, and Gaussian Naive Bayes, demonstrated moderate to low accuracy, approximately around 70%. These findings stress the imperative need for a nuanced approach in the selection and fine-tuning of machine learning models to ensure effective DDoS detection in SDN environments.</div>

Sex Estimation from the First and Second Ribs Using 3D Postmortem CT Images in a Japanese Population: A Comparison of Discriminant Analysis and Machine Learning Techniques

This study investigated the use of 3D postmortem computed tomography (PMCT) images of the first and second ribs for sex estimation in a Japanese population. Sex estimation models using conventional discriminant analysis and ten machine learning algorithms including logistic regression (LR), Naive Bayes (NB), K-Nearest Neighbors (KNN), decision tree (DT), random forest (RF), support vector machine (SVM), linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), artificial neural network (ANN), and extra tree (ET), were achieved from PMCT measurements of the first and second rib and the accuracy of models were compared. The results showed that ML algorithms, particularly LR, outperformed discriminant analysis, achieving an accuracy of 83.6% compared to 79.1% for stepwise discriminant analysis. This study highlights the potential of 3D PMCT and ML for accurate sex estimation in forensic anthropology.

Network intrusion detection in big datasets using Spark environment and incremental learning

<p class="Abstract">Internet of things (IoT) systems have experienced significant growth in data traffic, resulting in security and real-time processing issues. Intrusion detection systems (IDS) are currently an indispensable tool for self-protection against various attacks. However, IoT systems face serious challenges due to the functional diversity of attacks, resulting in detection methods with machine learning (ML) and limited static models generated by the linear discriminant analysis (LDA) algorithm. The process entails adjusting the model parameters in real time as new data arrives. This paper proposes a new method of an IDS based on the LDA algorithm with the incremental model. The model framework is trained and tested on the IoT intrusion dataset (UNSW-NB15) using the streaming linear discriminant analysis (SLDA) ML algorithm. Our approach increased model accuracy after each training, resulting in continuous model improvement. The comparison reveals that our dynamic model becomes more accurate after each batch and can detect new types of attacks.</p>

Classification of IPB variety of cayenne pepper genotypes using physical characteristics during the growing period until harvest using machine learning

In agrarian sectors, the prevalence of adulteration from seed selection to final agricultural outputs is a significant concern. This deceptive practice not only spans across distinct plant species but also extends to varieties within the same species, as observed in chili cultivation. Addressing this issue, the current study aims to develop a machine learning-based classification model that discriminates between cayenne pepper genotypes of IPB varieties and others. The model harnesses a range of plant physiological parameters monitored from the growth phase through to harvest. These parameters include plant and dichotomous heights, age at flowering, age at harvest, fruit stalk length, chili dimensions, individual chili weight, total number of chilies per plant, and overall yield productivity. Employing a dataset of 45 plant samples—30 from IPB varieties and 15 from non-IPB varieties—this research evaluates four machine learning algorithms: Linear Discriminant Analysis (LDA), k-Nearest Neighbors (kNN), Decision Tree (DT), and Random Forest (RF). To ascertain the robustness of the proposed model, the study also investigates the impact of varying data split ratios, including 90/10, 75/25, and 50/50, on model performance. Preliminary results indicate that the DT classifier, with an accuracy exceeding 80%, successfully differentiates the IPB cayenne pepper genotypes from other varieties. This promising conceptual model warrants further validation and enhancement through future research on a larger dataset, paving the way for its practical application in ensuring the authenticity of agricultural produce.

The effect of energy modality on tissue identification from surgical smoke by differential ion mobility spectrometry

Surgical smoke analysis offers a way to provide assisting information to the surgeon intraoperatively, which can be potentially used to assess cancer tumor margins in surgical oncology and alert the operator of accidental organ injuries caused by electrosurgical (ES) instruments. Surgical smoke content is affected by the energy instrument it is produced by. Classification of surgical smoke by differential ion mobility spectrometry (DMS) was evaluated with 6 porcine tissue types and 5 energy instruments. Instruments consisted of mono- and bipolar instruments, ultrasonic shears and a –blade. Machine learning was used to classify tissues by training binary classifier linear discriminant analysis (LDA) to distinguish a marked tissue class from the rest. The greatest binary classification accuracies were obtained with the monopolar instruments and the lowest with the bipolar instrument, 93.5 % and 77.5 % respectively. The analysis of surgical smoke with DMS is possible with a variety of energy instruments, however with varying performance. This implies that DMS based tissue identification is generalizable across different surgical instruments and surgical specialties.

Exposure to greenspaces sourced soils improves mice gut microbiota

Greenspaces are important components of our living environment and have been linked to various human health. However, the mechanisms underlying the linkages remain unclear. Enriching microbiota has emerged as a novel mechanism, but the corresponding evidence is still limited. We collected soil samples from forest land, grassland, and barren land in Zunyi City, southwestern China and prepared soil solutions. A total of 40 BALB/c mice were evenly divided into normal control group, model control group, forest soil group, grassland soil group, and barren land soil group. After establishing the pseudo germ-free mouse model, different soil solutions were administered through gavage, lasting for seven weeks. Fecal samples were collected and a 16S rRNA high-throughput sequencing analysis was performed. Then, alpha- and beta-diversity were calculated and employed to estimate the effects of soil exposures on mice gut microbial diversity and composition. Further, Linear Discriminant Analysis Effect Size (LEfSe) analysis was carried out to evaluate the effects of soil exposures on gut microbiota specific genera abundances and functional pathways. Compared to mice exposed to barren land soils, those exposed to soils sourced from forest land showed an increase of 0.43 and 70.63 units in the Shannon index and the Observed ASVs, respectively. In addition, exposure to soils sourced from forest land and grassland resulted in healthier changes (i.e., more short-chain fatty acids (SCFAs)-producing bacteria) in gut microbiota than those from barren land. Furthermore, mice exposed to forest soil and grassland soil showed enrichment in 5 and 3 pathways (e.g., butanoate metabolism) compared to those exposed to barren land soil, respectively. In conclusion, exposure to various greenspaces soils may modify the gut microbial communities of mice, potentially fostering a more beneficial microbiota profile. Further research is needed to explore the metabolites of these gut microbiota changes and their correlations with human health.

Response of bacterial and fungal composition in tailings to Mn pollution

Microorganisms are crucial for natural remediation of heavy metal pollution in mining areas. The regional survey and process analysis of Mn mine microbes is still limited. We investigated microbial species composition in tailings and adjacent soils of seven typical Mn mining areas in wet mid-subtropical China. The Mn bioavailable content in tailings was 55 times higher than in soils. Compared to soils, the heavy metal pollution in tailings reduced the hydrolase activities and microbial species diversity by 97% and 38%, respectively. The co-occurrence network of bacterial and fungal species in tailings was dominated by symbiosis and synergism, and their network complexity was lower than soils. Linear discriminant analysis of effect size revealed that Ralstonia, Acidisoma, and Talaromyces were the species most stimulated by Mn pollution because their relative and absolute abundance in tailings was much higher than in soils (p < 0.001). These key species defined the co-occurrence networks and affected metabolic pathways of microbial communities. Electrical conductivity and its interaction with Mn bioavailability strongly affected tailings microbial key species. This work identified the key species adapted to extreme Mn pollution in tailings, that can be used by bioremediation and maintenance of ecosystem functions in Mn-contaminated soils.

16S rRNA Sequencing Reveals Alterations of Gut Bacteria in Hirschsprung-Associated Enterocolitis.

Hirschsprung-associated enterocolitis (HAEC) stands as most common and serious complication of Hirschsprung's disease. Variations in the microbiota composition may account for the differences observed between HAEC and healthy individuals, offering crucial insights into the disease's pathogenesis. Here, we performed a study to changes in the gut microbiome using 16sRNA amplicon sequencing in a cohort of HAEC patients ( n = 16) and healthy controls ( n = 14). Our result revealed a significant disparity in beta diversity between the two groups. Following correction for false discovery rate, a rank-sum test at the genus level indicated a notable decrease in the relative abundance of Bifidobacterium , Lactobacillus , and Veillonella , whereas the Enterococcus genus exhibited a substantial increase in HAEC, a finding further supported by additional linear discriminant analysis effect size analysis. Functional analysis showed that putative transport and catabolism, digestive system, and metabolism of cofactors and vitamins were proved to be some abundant KOs (Kyoto Encyclopedia of Genes and Genomes [KEGG] orthologs) in healthy group, whereas infectious disease, membrane transport, and carbohydrate metabolism were the three KOs with the higher abundance in the HAEC group. Our data increased our insight into the HAEC, which may shed further light on HAEC pathogenesis. Our study firstly demonstrated the difference between fecal microbiota of HAEC patients and healthy individuals, which made a step forward in the understanding of the pathophysiology of HAEC.

Automatic Diagnosis of Sleep Apnea and Its Type Using Linear Discriminant Analysis and Support Vector Machine Based on the Characteristics of Two ECG and PPG Signals

Automatic Diagnosis of Sleep Apnea and Its Type Using Linear Discriminant Analysis and Support Vector Machine Based on the Characteristics of Two ECG and PPG Signals

Preclinical Implementation of matRadiomics: A Case Study for Early Malformation Prediction in Zebrafish Model

Radiomics provides a structured approach to support clinical decision-making through key steps; however, users often face difficulties when switching between various software platforms to complete the workflow. To streamline this process, matRadiomics integrates the entire radiomics workflow within a single platform. This study extends matRadiomics to preclinical settings and validates it through a case study focused on early malformation differentiation in a zebrafish model. The proposed plugin incorporates Pyradiomics and streamlines feature extraction, selection, and classification using machine learning models (linear discriminant analysis—LDA; k-nearest neighbors—KNNs; and support vector machines—SVMs) with k-fold cross-validation for model validation. Classifier performances are evaluated using area under the ROC curve (AUC) and accuracy. The case study indicated the criticality of the long time required to extract features from preclinical images, generally of higher resolution than clinical images. To address this, a feature analysis was conducted to optimize settings, reducing extraction time while maintaining similarity to the original features. As a result, SVM exhibited the best performance for early malformation differentiation in zebrafish (AUC = 0.723; accuracy of 0.72). This case study underscores the plugin’s versatility and effectiveness in early biological outcome prediction, emphasizing its applicability across biomedical research fields.

Soluble Urokinase-Type Plasminogen Activator Receptor and Inflammatory Biomarker Response with Prognostic Significance after Acute Neuronal Injury - a Prospective Cohort Study.

Aneurysmal subarachnoid hemorrhage (aSAH), ischemic stroke (IS), and traumatic brain injury (TBI) are severe conditions impacting individuals and society. Identifying reliable prognostic biomarkers for predicting survival or recovery remains a challenge. Soluble urokinase type plasminogen activator receptor (suPAR) has gained attention as a potential prognostic biomarker in acute sepsis. This study evaluates suPAR and related neuroinflammatory biomarkers in serum for brain injury prognosis. This prospective study included 31 aSAH, 30 IS, 13 TBI, and three healthy controls (n = 77). Serum samples were collected on average 5.9days post-injury, analyzing suPAR, IL-1β, cyclophilin A, and TNFα levels using ELISA. Outcomes were assessed 90days post-injury with the modified Rankin Scale (mRS), categorized as favorable (mRS 0-2) or unfavorable (mRS 3-6). Statistical analyses included 2-tailed t-tests, Pearson's correlations, and machine learning linear discriminant analysis (LDA) for biomarker combinations. Elevated suPAR levels were found in brain injury patients compared to controls (p = 0.017). Increased suPAR correlated with unfavorable outcomes (p = 0.0018) and showed prognostic value (AUC = 0.66, p = 0.03). IL-1β levels were higher in the unfavorable group (p = 0.0015). LDA combinatory analysis resulted a fair prognostic accuracy with canonical equation = 0.775[suPAR] + 0.667[IL1-β] (AUC = 0.77, OR 0.296, sensitivity 93.1%, specificity 53.1%, p = 0.0007). No correlation was found between suPAR and CRP or infection status. Elevated suPAR levels in acute brain injury patients were associated with poorer outcomes, highlighting suPAR's potential as a prognostic biomarker across different brain injury types. Combining IL-1β with suPAR improved prognostic accuracy, supporting a multimodal biomarker approach for predicting outcomes.

RF sensing enabled tracking of human facial expressions using machine learning algorithms.

Automatic analysis of facial expressions has emerged as a prominent research area in the past decade. Facial expressions serve as crucial indicators for understanding human behavior, enabling the identification and assessment of positive and negative emotions. Moreover, facial expressions provide insights into various aspects of mental activities, social connections, and physiological information. Currently, most facial expression detection systems rely on cameras and wearable devices. However, these methods have drawbacks, including privacy concerns, issues with poor lighting and line of sight blockage, difficulties in training with longer video sequences, computational complexities, and disruptions to daily routines. To address these challenges, this study proposes a novel and privacy-preserving human behavior recognition system that utilizes Frequency Modulated Continuous Wave (FMCW) radar combined with Machine Learning (ML) techniques for classifying facial expressions. Specifically, the study focuses on five common facial expressions: Happy, Sad, Fear, Surprise, and Neutral. The recorded data is obtained in the form of a Micro-Doppler signal, and state-of-the-art ML models such as Super Learner, Linear Discriminant Analysis, Random Forest, K-Nearest Neighbor, Long Short-Term Memory, and Logistic Regression are employed to extract relevant features. These extracted features from the radar data are then fed into ML models for classification. The results show a highly promising classification accuracy of 91%. The future applications of the proposed work will lead to advancements in technology, healthcare, security, and communication, thereby improving overall human well-being and societal functioning.

Comparison of automatic and physiologically-based feature selection methods for classifying physiological stress using heart rate and pulse rate variability indices.

This study evaluates the effectiveness of four machine Learning algorithms in classifying physiological stress using Heart Rate Variability (HRV) and Pulse Rate Variability (PRV) time series, comparing an automatic feature selection based on Akaike's criterion to a physiologically-based feature selection approach. Linear Discriminant Analysis, Support Vector Machines, K-Nearest Neighbors and Random Forest were applied on ten HRV and PRV indices from time, frequency and information domains, selected with the two feature selection approaches. Data were collected from 127 healthy individuals during different stress conditions (rest, postural and mental stress). Our results highlight that, while specific stress classification is feasible, distinguishing between postural and mental stress remains challenging. The used classifiers exhibited similar performance, with automatic Akaike Information Criterion-based feature selection proving overall better than the physiology-driven approach. Additionally, PRV-based features performed comparably to HRV-based ones, indicating their potential in outpatient monitoring using wearable devices. The obtained findings help to determine the most relevant HRV/PRV features for stress classification, potentially useful to highlight different physiological mechanisms involved during both challenges accompanied by a shift in the sympathovagal balance. The proposed approach may have implications for advancing stress assessment methodologies in clinical settings and real-world contexts for well-being evaluation.

Hybrid Functional Near-Infrared Spectroscopy System and Electromyography for Prosthetic Knee Control

The increasing number of individuals with limb loss worldwide highlights the need for advancements in prosthetic knee technology. To improve control and quality of life, integrating brain–computer communication with motor imagery offers a promising solution. This study introduces a hybrid system that combines electromyography (EMG) and functional near-infrared spectroscopy (fNIRS) to address these limitations and enhance the control of knee movements for individuals with above-knee amputations. The study involved an experiment with nine healthy male participants, consisting of two sessions: real execution and imagined execution using motor imagery. The OpenBCI Cyton board collected EMG signals corresponding to the desired movements, while fNIRS monitored brain activity in the prefrontal and motor cortices. The analysis of the simultaneous measurement of the muscular and hemodynamic responses demonstrated that combining these data sources significantly improved the classification accuracy compared to using each dataset alone. The results showed that integrating both the EMG and fNIRS data consistently achieved a higher classification accuracy. More specifically, the Support Vector Machine performed the best during the motor imagery tasks, with an average accuracy of 49.61%, while the Linear Discriminant Analysis excelled in the real execution tasks, achieving an average accuracy of 89.67%. This research validates the feasibility of using a hybrid approach with EMG and fNIRS to enable prosthetic knee control through motor imagery, representing a significant advancement potential in prosthetic technology.

Comparison of dimensionality reduction methods on hyperspectral images for the identification of heathlands and mires.

Hyperspectral data and machine learning offer great potential for identifying valuable open ecosystems. Due to the large volume of data, preprocessing of hyperspectral images must involve dimensionality reduction. The main goal of this study was to test the effectiveness of various types of feature reduction (feature selection and feature extraction) when performing classification using the Random Forest algorithm. A comparison was conducted between two ecosystems - heathlands and mires protected as Natura 2000 habitats. Two transformations of feature extraction were chosen, namely Minimum Noise Fraction (MNF) and Principal Component Analysis (PCA), while Linear Discriminant Analysis (LDA) was used as a feature selection method. It was proven that irrespective of the class type, accuracy is higher with the feature extraction method (mean F1 accuracy of 0.922) than with feature selection (mean F1 accuracy of 0.787). At the same time, no significant differences in accuracies were found between the MNF and PCA methods. Although LDA resulted in lower accuracies (0.816 for heathland and 0.750 for mires), the method could also be used due to relatively high F1 values. The effectiveness of the LDA method for feature reduction in open ecosystem identification was confirmed for the first time for open natural vegetation.

Bimodal Array-Based Fluorescence Sensor and Microfluidic Technology for Protein Fingerprinting and Clinical Diagnosis.

Proteins play a crucial role in determining disease states in humans, making them prime targets for the development of diagnostic sensors. The developed sensor array is used to investigate global proteomic changes by fingerprinting multifactorial disease states in model urine simulating phenylketonuria and in serum from preeclamptic pregnant women. Here, we report a fluorescence-based chemical sensing array that exploits the host-guest interaction between cucurbit[7]uril (CB[7]) and fluorescent triphenylamine derivatives (TPA) to detect a range of proteins. Using linear discriminant analysis, we identify fluorescence fingerprints of 14 proteins with over 98% accuracy in buffer and human serum. The array is optimized on an automated droplet microfluidic-based platform, for high-throughput sensing with controlled composition and lower sample volumes. This sensor enables the discrimination of proteins in physiological buffer and human serum, with promising applications in disease diagnosis.

Immunoglobulin-coating patterns reveal altered humoral responses to gut bacteria in pediatric cow milk allergies.

Pediatric cow milk allergies (CMA) can occur in immunoglobulin (Ig) E and non-IgE-mediated forms. Unlike IgE-mediated allergies, the mechanisms of disease pathogenesis in non-IgE-mediated food allergy and an association with microbiome has not been well established. Previous studies have identified the presence of altered humoral responses to gut bacteria in IgE mediated allergies. Here, we analyzed IgA, IgE and IgG responses to gut bacteria in subjects with either IgE or non-IgE mediated CMA to identify relative proportions of Ig-coated bacteria and characterize unique disease specific microbial signatures. Multi-parametric flow cytometry analysis was used to identify IgA, IgE and IgG responses to gut bacteria in CMA patients. Cell sorting of Ig coated gut bacteria was subsequently performed followed by high throughput 16S rRNA gene sequencing and specific patterns of humoral responses to gut bacteria assessed in each study group. We identified significant alterations in IgA and IgG gut bacterial coating patterns in CMA subjects. Proportions of IgA-coated bacteria were decreased in IgE mediated CMA subjects without atopic dermatitis (ALL) and non-IgE mediated CMA subjects (ENP), compared to healthy controls (CON). In comparison, IgG-coated bacteria was significantly elevated in CMA subjects with atopic dermatitis (AD). Alpha and beta diversities displayed significant differences in IgA-, IgE-, and IgG-coated bacteria in AD and ENP groups. Significant differences in bacteria coated by IgA, IgE and IgG were detected at Phyla, Genus and Species levels and associated bacterial dysbiosis in IgE and non-IgE mediated allergies were identified. Linear discriminant analysis (LDA) effect size (LEFse) revealed unique disease associated bacterial signatures, including several pathogenic bacteria namely Bacteroides fragilis, Ruminococcus gnavus, Eubacterium dolichum, Fusobacterium, Clostridium neonatale and Robinsoniella peoriensis. Receiver operating characteristic curve analysis confirmed the efficiency of using the bacterial signatures identified as biomarkers for disease. Altered IgA and IgG responses to gut bacteria were identified in CMA subjects. The disease-specific responses were associated with alterations in bacterial diversity and concomitant dysbiosis of Ig-coated bacteria in IgE-mediated and non-IgE-mediated CMA pediatric subjects. The identification of pathogenic bacteria uniquely associated with different classes of allergic disease indicates a role of these bacteria in driving disease-specific pathological phenotypes.

Exploring the rumen microbial function in Angus bulls with divergent residual feed intake

This study leverages Shotgun metagenomics to assess the rumen microbial community and functionality in Angus bulls with differing residual feed intake-expected progeny difference (RFI-EPD) values, aiming to elucidate the microbial contributions to feed efficiency. Negative RFI-EPD bulls (NegRFI: n=10; RFI-EPD= -0.3883 kg/d) and positive RFI-EPD bulls (PosRFI: n=10; RFI-EPD=0.2935 kg/d) were selected from a group of 59 Angus bulls (average body weight (BW) = 428 ± 18.8 kg; 350 ± 13.4 d of age) fed a high-forage total mixed ration after a 60-d testing period. At the end of the 60-d period, rumen fluid samples were collected for bacterial DNA extraction and subsequent shotgun metagenomic sequencing. Results of the metagenome analysis revealed greater gene richness in NegRFI bulls, compared to PosRFI. Analysis of similarity revealed a small but noticeable difference (P =0.052; R-value = 0.097) in the rumen microbial community of NegRFI and PosRFI bulls. Linear Discriminant Analysis effect size (Lefse) was utilized to identify the differentially abundant taxa. The Lefse results showed that class Fibrobacteria (LDA = 5.1) and genus Fibrobacter (LDA = 4.8) were greater in NegRFI bulls, compared to PosRFI bulls. Relative abundance of the carbohydrate-active enzymes was also compared using Lefse. The results showed greater relative abundance of glycoside hydrolases and carbohydrate-binding modules such as GH5, CBM86, CBM35, GH43, and CBM6 (LDA &amp;gt; 3.0) in NegRFI bulls whereas GH13 and GT2 were greater in PosRFI bulls. The distinct metabolic and microbial profiles observed in NegRFI, compared to PosRFI bulls, characterized by greater gene richness and specific taxa such as Fibrobacter, and variations in carbohydrate-active enzymes, underscore the potential genetic and functional differences in their rumen microbiome. These findings contribute to a deeper understanding of the interplay between rumen microbiota and feed efficiency in Angus bulls, opening avenues for targeted interventions and advancements in livestock management practices.

Spectral discrimination of crops and weeds using deep learning assisted by wavelet transform and statistical preprocessing

Abstract Automatic detection and removal of weeds is a challenging task that requires precise sensors. While crops and weeds possess similar features in terms of appearance, they can be discriminated based on spectral information. This can be done because any object has its own specific spectral signature based on its physical structure and chemical contents. This study examined the use of wavelet transform and deep learning for discrimination of weeds from crops. A total of 626 spectral reflectances in the range of 380 to 1,000 nm were obtained for three crops (cucumber [Cucumis sativus L.], tomato [Solanum lycopersicum L.], and bell pepper [Capsicum annuum L.]) and five different weeds (bindweed [Convolvulus spp.], purple nutsedge [Cyperus rotundus L.], narrowleaf plantain [Plantago lanceolata L.], common cinquefoil [Potentilla simplex Michx.], and garden sorrel [Rumex acetosa L.]). Morse wavelet was employed to decompose the spectra and extract the scalograms, which are the RGB representations of the spectral data. Two deep convolutional neural networks (i.e., GoogLeNet and SqueezNet) were employed for the recognition of crops and weeds. In addition, six common classifiers, including linear discriminant analysis, quadratic discriminant analysis, linear support vector machine, quadratic support vector machine, artificial neural networks, and k-nearest neighbors classifier (KNN), were used for the task of crop/weed discrimination to build the comparison with the proposed method. The error of prediction gradually decreased, and a 100% correct classification was achieved after 258 iterations. Analysis showed that SqueezNet provided classification of 100% accuracy, while GoogLeNet’s accuracy was 97.8% for the test set. Among the common classifiers, KNN provided the highest accuracy (i.e., 100%). This study showed that the proposed method can be successfully utilized for classification of crops and weeds.