Unknown Samples Research Articles

Machine learning-assisted fluorescence/fluorescence colorimetric sensor array for discriminating amyloid fibrils

Misfolding and aggregation of proteins often lead to the development of diseases, and amyloid has gained widespread attention as a biomarker for a variety of diseases. In this study, we developed a fluorescence/fluorescence colorimetric dual-mode sensor array for the detection of amyloid fibrils using several commercially available organic small molecular dyes and alkaloids, including Thioflavin T, Congo Red, 8-anilino-1-naphthalenesulfonic acid, Safranine T, berberine and coptisine, as the elements. Herein, the array could not only use the fluorescence intensities change before and after protein interaction as a pattern recognition signal, but also read the ΔR/ΔG/ΔB values of the photos taken in the UV dark box on a smartphone-based platform, which converted the chromaticity information into intuitive data. Five studied amyloid fibrils, i.e. insulin, lysozyme, bovine serum albumin, amyloid-β 42 and α-synuclein fibrils, were properly distinguished with data processing assisted by machine learning algorithms, i.e. linear discriminant analysis, principal component analysis and hierarchical cluster analysis. After reducing the number of elements by principal component analysis, a simplified array quantified individual amyloid fibrils at 0.05–5 μM and 0.5–10 μM with fluorescence and fluorescence colorimetric signals, respectively, and successfully identified 25 unknown samples with high accuracy in diluted human plasma matrix and artificial cerebrospinal fluid. The array had good selectivity and sensitivity, providing a simple and inexpensive method for amyloid discrimination.

High-accuracy quantitative model for phosphate anions in solution based on absorption spectroscopy and machine learning algorithms

The eutrophication pollution caused by excessive phosphorus seriously affects production and daily life of the people. Accurate and rapid detection of phosphate is crucial for water quality management and cleaner production. Currently, most analytical methods for phosphates involve the addition of multiple chemical reagents and are incapable of simultaneous quantification of total phosphorus (TP) and phosphate anions. The environmental friendliness and analytical performance of the methods need to be improved. Here, we established a green and high-accuracy quantitative model based on absorption spectroscopy and machine learning algorithms, which achieved the analysis of TP, H2PO4− and HPO42−. Based on the pH and absorption spectra of phosphate anions, classification and regression tree (CART), random forest (RF), and extreme gradient boosting (XGBoost) algorithms were used to train the original absorption spectra data, first-order derivative spectra data, and multiple feature parameter data of phosphate solutions with pH 4.7–10.0. By comparing the R2, root mean square error (RMSE) and mean absolute error (MAE) evaluation indicators, it was found that the phosphate anions quantitative model established using the CART based on the feature parameters of 187 nm performed the best, R2 > 0.9999, RMSE <3.4 × 10−6, MAE <8.1 × 10−7. The quantitative ranges of TP, H2PO4−, and HPO42− were 3.98 × 10−4—9.33 × 10−4 mol/L, 5.98 × 10−6—4.20 × 10−4 mol/L and 1.53 × 10−6—7.43 × 10−4 mol/L, respectively. Moreover, the quantitative model was optimized based on the spectral characteristics of 14 interfering ions, to improve the model analytical performance for unknown samples. The quantitative model developed in the study achieves accurate analysis of TP, H2PO4− and HPO42−, which is in line with green chemistry concept, and has the potential for prediction and early warning of water environment quality.

Simplified and Detailed Evaluations of the Uncertainty of the Measurement of Microbiological Contamination of Pharmaceutical Products.

The control of the microbial contamination of pharmaceutical products (PP) is crucial to ensure their safety and efficacy. The validity of the monitoring of such contamination depends on the uncertainty of this quantification. Highly uncertain quantifications due to the variability of determinations or the magnitude of systematic effects affecting microbial growth or other analytical operations make analysis unfit for the intended use. The quantification of the measurement uncertainty expressing the combined effects of all random and systematic effects affecting the analysis allows for a sound decision about quantification adequacy for their intended use. The complexity of the quantification of microbial analysis uncertainty led to the development of simplified ways of performing this evaluation. This work assesses the adequacy of the simplified quantification of the uncertainty of the determination of the microbial contamination of PP by log transforming microbial count and dilution factor of the test sample whose uncertainty is combined in a log scale using the uncertainty propagation law. This assessment is performed by a parallel novel bottom-up and accurate evaluation of microbial analysis uncertainty involving the Monte Carlo method simulation of the Poisson log-normal distribution of counts and of the normally distributed measured volumes involved in the analysis. Systematic effects are assessed and corrected on results to compensate for their impact on the determinations. Poisson regression is used to predict precision affecting determinations on unknown test samples. Simplified and detailed models of the uncertainty of the measurement of the microbial contamination of PP are provided, allowing objective comparisons of several determinations and those with a maximum contamination level. This work concludes that triplicate determinations are required to produce results with adequately low uncertainty and that simplified uncertainty quantification underevaluates or overevaluates the uncertainty from determinations based on low or high colony numbers, respectively. Therefore, detailed uncertainty evaluations are advised for determinations between 50 and 200% of PP's maximum admissible contamination value. User-friendly tools for detailed and simplified evaluations of the uncertainty of the measurement of microbial contamination of PP are provided together with the understanding of when simplifications are adequate.

The Homology Determination System for APT Samples Based on Gene Maps

At present, there are fewer types of homology determination methods for advanced persistent threat (APT) samples detection, and most existing determination schemes have problems such as high cost, low accuracy, and difficulty in identifying unknown APT samples. Therefore, we proposed a homology determination system for APT samples based on gene maps by integrating deep learning and gene maps. Firstly, we extract the software gene features from the samples uploaded by the user and apply the TF-IDF algorithm to clean the extracted software genes. The Word2Vec algorithm is used to vectorize all the genes to construct the gene sample vectors. And we use a LSTM-based classifier to detect APT attack samples. Finally, the K-nearest neighbor algorithm is used to determine the homology of gene-sharing APT samples. The detailed construction process of the scheme is given in this paper, including APT sample gene extraction, cleaning, clustering, sample detection, and homology determination. Experimental validation showcases our model outperforming existing methodologies with an accuracy of 95%, precision of 94%, and recall of 95%. When compared to previous models, the superiority of our approach is evident. These results underscore our model’s high efficiency and accuracy, confirming its potential for significant application in the field of cybersecurity.

Individualization of petrol sources by high field nuclear magnetic resonance spectroscopy

In the forensic science context petrol is considered the most common fire accelerant. However, the identification and classification of petrol sources through the years has been proven to be a challenge in the investigation of fire related incidents. This research explored the possibility of identification and classification of petrol sources using high field NMR spectroscopy. In this study, 1H NMR profiling, using specific pulse sequences to analyse neat aliquot petrol samples of different brands collected at different times across the UK and Ireland is shown, for the first time, to provide a diagnostic ‘fingerprint’ with specific chemical compounds that can be used for identification and classification of petrol samples. This enables linkage of unknown petrol samples to a source and in addition provides a tool which allows exclusion of potential petrol sources. A new, innovative method using 1H selTOCSY is described for the individualization and classification of petrol samples through the identification of olefinic markers in the samples. Those markers were identified as (i) 3-methyl-1-butene, (ii) a mixture of 1-pentene and 3-methyl-1-butene, (iii) 2-methyl-2-butene and (iv) a mixture of cis and trans-2-pentene.

Evaluating Effectiveness of Multi-Component Waste Plastic Bags on Bitumen Properties: Physical, Rheological, and Aging.

This study examines the applicability of an unknown composition waste plastic bag sample as bitumen modifier. The waste components were initially characterized to identify the type of plastics and the level of impurity. Asphalt binder performance was examined for rutting, thermal, and age resistance. The results revealed that the waste plastic bags, predominantly consisted of Low-Density Polyethylene (LDPE) and Linear Low-Density Polyethylene (LLDPE) and contained 6.1% impurities. The binder tests indicated that the waste plastic bags enhanced the rutting resistance of bitumen by one grade, with its modification more similar to LLDPE, rather than LDPE. The thermal degradation and aging properties of the modified binders demonstrated that the bitumen modified by the waste plastic bags exhibited slightly lower resistance to temperature and aging compared to virgin LDPE and LLDPE. This was attributed to the impurities contained in the waste plastic. In conclusion, the analyzed waste plastic bags proved to be suitable for use in binder modification, presenting a viable alternative to virgin LLDPE.

Data-science-guided calibration curve prediction of an MLCT-based ee determination assay for chiral amines

Circular dichroism (CD) based enantiomeric excess (ee) determination assays are optical alternatives to chromatographic ee determination in high-throughput screening (HTS) applications. However, the implementation of these assays requires calibration experiments using enantioenriched materials. We present a data-driven approach that circumvents the need for chiral resolution and calibration experiments for an octahedral Fe(II) complex (1) used for the ee determination of α-chiral primary amines. By computationally parameterizing the imine ligands formed in the assay conditions, a model of the circular dichroism (CD) response of the Fe(II) assembly was developed. Using this model, calibration curves were generated for four analytes and compared to experimentally generated curves. In a single-blind ee determination study, the ee values of unknown samples were determined within 9% mean absolute error, which rivals the error using experimentally generated calibration curves.

Integrating intrinsic information: A novel open set domain adaptation network for cross-domain fault diagnosis with multiple unknown faults

Existing popular domain adaptation approaches typically assume that the source and target domains share the same label set. However, in industrial scenarios, the equipment may encounter unknown fault modes because of the harsh operating environment, which limits the application of existing diagnostic methods. To address the above problem, an intrinsic information-guided open set domain adaptation network is proposed for cross-domain fault diagnosis with unknown faults. First, a similarity-based discrimination framework is constructed to enhance the robustness of the model for unknown samples, which learns the similarity between samples and fault prototypes to enhance the classification performance. Then, a multi-information integrated weighting module is designed to quantify the transferability of samples through enhanced domain similarity information learning and prediction uncertainty information learning methods. Additionally, a self-supervised neighborhood clustering learning method is constructed, which enables the model to learn structural information about the target domain and encourages the target samples to cluster closely for better separability. Finally, the weighted open set adversarial training framework effectively facilitates diagnostic knowledge transfer and unknown fault recognition. Comprehensive experimental results on two datasets demonstrate the effectiveness of the proposed method in addressing the open set cross-domain diagnosis problem, which achieves promising performance over the comparison methods.

Strengthening LLM ecosystem security: Preventing mobile malware from manipulating LLM-based applications

Large language model (LLM) platform vendors have begun to make their models available for developers to build for different use cases. However, the emergence of LLM-based applications may raise security and privacy issues, and even LLM-based applications may be susceptible to malware. To strengthen LLM ecosystem security, it's crucial to develop malware detection algorithms for various platforms. We pay attention to Android malware because the Android platform is widely used and vulnerable. Existing single feature based-solutions cannot effectively describe applications, and aged models fail to detect new malware as Android platform develops and malware evolves. Therefore, existing detection methods are ill-suited for evolved malware that may manipulate LLM-based applications. To tackle the above problems, we design EvolveDroid, an anti-aging Android malware detection system. On the one hand, EvolveDroid utilizes different view features to reflect malware behavior from multiple dimensions, and maximizes the advantages of each feature type through feature aggregation. On the other hand, EvolveDroid learns good representation of applications through contrastive learning and generates pseudo labels by measuring the distance between unknown samples and existing samples for model updating. Extensive evaluations show that EvolveDroid outperforms state-of-the-art (sota) solutions in detection performance and slowing model aging.

Development and evaluation of an ensemble model to identify host-related metadata from fecal microbiota of zoo-housed mammals

Much research has been conducted to describe the factors that determine the fecal microbiome, with diet and host phylogeny as the main drivers. The influence of diet has been described at different levels. Firstly, there are major differences in the microbiomes of herbivorous and carnivorous species and secondly the morphology of the digestive system also determines the composition and diversity of the microbiota. In this study, we aim to describe the influence of the three factors – diet, digestive system and host - on the microbiota in order to develop a model that is able to characterize host-specific metadata from an unknown fecal sample. We therefore analyzed the 16s rRNA from 525 fecal samples of 14 zoo-housed species belonging to different phylogenetic groups including herbivores, carnivores and omnivores. We found significant differences in the bacterial taxa correlated with these groups. While herbivores show positive correlations with a large number of bacterial taxa, we found fewer taxa correlating with carnivores or omnivores. We also detected considerable differences in the microbiota of the ruminant, hindgut fermenting and simple digestive system. Based on these results, we developed a logistic ensemble model, that predicts the diet and based on these findings either the herbivorous digestive system or the carnivorous host-family from a given fecal microbiota composition. This model is able to effectively discriminate herbivores, omnivores and carnivores. It also excels at predicting the herbivore-specific digestive system with 98% accuracy, further reinforcing the strong link between microbiota and digestive system morphology. Carnivorous host-family identification achieves an overall accuracy of 79%, although this performance varies between families. We provide this trained model as a tool to enable users to generate host-specific information from their microbiome data. In future research, tools such as the one presented here could lead to a combined approach of microbiome and host-specific analyses which would be a great advantage in non-invasive wildlife monitoring.

Machine Learning for Characterizing Biofuels Based on Femtosecond Laser Ionization Mass Spectrometry.

A sample mixture of fatty acid methyl esters (FAMEs) was measured by femtosecond laser ionization mass spectrometry (fsLIMS) using the fifth (206 nm) and fourth (257 nm) harmonic emissions of an ytterbium (Yb) laser (1030 nm). Molecular ions were observed as the major signals in this technique, providing valuable information concerning the molecular weight and the number of double bonds in the molecule. The mass spectral data were then used as explanatory variables in machine learning based on artificial intelligence (AI) to correlate with objective variables such as the cetane number, kinematic viscosity, specific gravity, a higher heating value, an iodine value, flash point, oxidative stability index, and a cloud point measured for reference biofuel samples containing various FAMEs. The properties of biofuels, i.e., the objective variables, were evaluated from the mass spectral data obtained for unknown samples. The errors in the evaluation were a few percent when the distribution of the FAMEs in the unknown biofuel sample was similar to those of the biofuels used for machine learning. As demonstrated herein, the present approach, involving a combination of fsLIMS and AI, has the potential for use in evaluating the properties of a biofuel and then in solving of environmental issues associated with global warming.

Reversible Fusion-Fission MXene Fiber-Based Microelectrodes for Target-Specific Gram-Positive and Gram-Negative Bacterium Discrimination.

Inaccurate or cumbersome clinical pathogen diagnosis between Gram-positive bacteria (G+) and Gram-negative (G-) bacteria lead to delayed clinical therapeutic interventions. Microelectrode-based electrochemical sensors exhibit the significant advantages of rapid response and minimal sample consumption, but the loading capacity and discrimination precision are weak. Herein, we develop reversible fusion-fission MXene-based fiber microelectrodes for G+/G- bacteria analysis. During the fissuring process, the spatial utilization, loading capacity, sensitivity, and selectivity of microelectrodes were maximized, and polymyxin B and vancomycin were assembled for G+/G- identification. The surface-tension-driven reversible fusion facilitated its reusability. A deep learning model was further applied for the electrochemical impedance spectroscopy (EIS) identification in diverse ratio concentrations of G+ and G- of (1:100-100:1) with higher accuracy (>93%) and gave predictable detection results for unknown samples. Meanwhile, the as-proposed sensing platform reached higher sensitivity toward E. coli (24.3 CFU/mL) and S. aureus (37.2 CFU/mL) in 20 min. The as-proposed platform provides valuable insights for bacterium discrimination and quantification.

High-Oleic Sunflower Oil as a Potential Substitute for Palm Oil in Sugar Coatings-A Comparative Quality Determination Using Multispectral Imaging and an Electronic Nose.

Palm oil has a bad reputation due to the exploitation of farmers and the destruction of endangered animal habitats. Therefore, many consumers wish to avoid the use of palm oil. Decorative sugar contains a small amount of palm oil to prevent the sugar from melting on hot bakery products. High-oleic sunflower oil used as a substitute for palm oil was analyzed in this study via multispectral imaging and an electronic nose, two methods suitable for potential large-batch analysis of sugar/oil coatings. Multispectral imaging is a nondestructive method for comparing the wavelength reflections of the surface of a sample. Reference samples enabled the estimation of the quality of unknown samples, which were confirmed via acid value measurements. Additionally, for quality determination, volatile compounds from decorative sugars were measured with an electronic nose. Both applications provide comparable data that provide information about the quality of decorative sugars.

Pixel-Level Recognition of Trace Mycotoxins in Red Ginseng Based on Hyperspectral Imaging Combined with 1DCNN-Residual-BiLSTM-Attention Model.

Red ginseng is widely used in food and pharmaceuticals due to its significant nutritional value. However, during the processing and storage of red ginseng, it is susceptible to grow mold and produce mycotoxins, generating security issues. This study proposes a novel approach using hyperspectral imaging technology and a 1D-convolutional neural network-residual-bidirectional-long short-term memory attention mechanism (1DCNN-ResBiLSTM-Attention) for pixel-level mycotoxin recognition in red ginseng. The "Red Ginseng-Mycotoxin" (R-M) dataset is established, and optimal parameters for 1D-CNN, residual bidirectional long short-term memory (ResBiLSTM), and 1DCNN-ResBiLSTM-Attention models are determined. The models achieved testing accuracies of 98.75%, 99.03%, and 99.17%, respectively. To simulate real detection scenarios with potential interfering impurities during the sampling process, a "Red Ginseng-Mycotoxin-Interfering Impurities" (R-M-I) dataset was created. The testing accuracy of the 1DCNN-ResBiLSTM-Attention model reached 96.39%, and it successfully predicted pixel-wise classification for other unknown samples. This study introduces a novel method for real-time mycotoxin monitoring in traditional Chinese medicine, with important implications for the on-site quality control of herbal materials.

A kind of multi-dot ensemble regression AI detector for lubricating oil additive content based on lambert-beer law

In this work, we propose a Multi-dot Ensemble Regression AI detector (MER) based on the Lambert-Beer law. We pre-trained a model using the infrared spectral data of target additives collected in advance to detect the target additives in unknown oil samples. The algorithm's feasibility was validated by assessing the content of additives in a series of simulated commercial oil samples that were not part of the training set. We established models for three common lubricating oil additives (anti-friction, anti-wear, and antioxidant agents), demonstrating their effectiveness in oil sample detection. Additionally, by comparing with other algorithms, we established the superiority of MER in small-sample learning scenarios.

Research on the composition of glass relics based on CART model

Based on each index of glass samples, the indexes of all kinds of glass sample cluster analysis, and the class divide, it is to predict the chemical composition of glass samples which before weathering, the unknown glass samples for testing, and a correlation analysis of chemical composition in glass samples was carried out through the appropriate method. As for the relationship between the weathering degree of cultural relics surface and the type, ornamentation and color of glass samples, the weathering degree and the type, ornamentation and color of glass samples were corresponding to unilateral factors, and the relationship was preliminarily explored by bar charts. It was concluded that the weathering degree had a certain multiple relationship with the three indexes. In order to further study the relationship quantitatively, we used the correlation thermogram to analyze. It can be concluded that the type of cultural relics has the strongest correlation with surface weathering, the correlation degree reaches 0.4, while the correlation of color and ornamentation are 0.013 and 0.12, which are relatively low.

Speed of sound measurement and mapping in transparent materials by impulsive stimulated Brillouin microscopy

Impulsive stimulated Brillouin scattering (ISBS) is a variant of stimulated Brillouin scattering, which can overcome the shortcomings of the long acquisition time of traditional Brillouin microscopy. We introduce the difference between ISBS and other Brillouin microscopies in calculating longitudinal modulus. The Brillouin frequency shift obtained by ISBS is only related to the system parameters and the speed of sound (SOS) in the sample, not to the refractive index. Non-contact SOS measurement of homogeneous samples is an important application of Brillouin scattering, used in the early study of Brillouin spectroscopy and the mechanical properties of liquids. However, the measurement requires prior knowledge of the sample refractive index, which limits the measurement of the unknown refractive index sample. Here, we propose a method to measure the SOS based on ISBS, which in principle avoids the need for refractive index parameters. The SOS of several liquids are measured and compared with the standard values. The mean relative standard deviation is 1.13%. Moreover, we measure the SOS of a mixture of ethanol and water to demonstrate an application of measuring SOS without refractive index information. We also demonstrate the high spatial resolution of ISBS with a methanol-filled PDMS sample.

Contrastive learning based open-set recognition with unknown score

Open-set recognition (OSR) is designed to classify the seen classes and identify the unseen classes as unknown. However, existing open-set classifiers rely on deep networks trained using supervised learning techniques for known classes present in the training samples. This factor tends to specialize the learned features of the model toward known classes, making it challenging to differentiate between unknown classes. In this study, we propose a novel open-set recognition framework based on contrastive learning with an unknown score. First, we propose a novel training framework based on contrastive learning to learn more informative features and preserve beneficial information to separate unknown from known. Second, we propose an unknown score function based on multi-layer features to detect unknown samples by considering the difference between the known and unknown classes of intermediate-layer features. Based on extensive experiments conducted on multiple benchmark datasets, the proposed method has been demonstrated to outperform existing methods and achieve state-of-the-art results.

Multi-source domain adaptation handling inaccurate label spaces

Domain adaptation with inaccurate label is a challenging and interesting topic in transfer learning, dealing with source and target domains with shift label spaces. Most existing domain adaptation methods assume aware label distributions among source and target domains. However, this cannot always be guaranteed in reality. Furthermore, existing multi-domain adaptation methods rarely deal with label heterogeneity among source domains. Thus, in this paper, we propose a multi-source domain adaptation method handling Inaccurate label (IncLabDA) during transfer. The proposed method designs a module that can transfer knowledge from multi-source domains with both homogeneous and heterogeneous label spaces in universal scenario. Anchors are generated from pre-trained model to build data-matching via a contrastive method avoiding to referring original data. In addition, class center consistency combined with clustering strategy considering both global and local confidences is adopted to recognize out-of-distribution samples. By removing source private classes and target unknown samples, highly confident target samples are collected to self-supervise the adaptation. At the same time, constraints enlarging the distance among target known classes and between the known and unknown samples are applied to enhance the performance of the proposed model. Experiments on real-world datasets validate the superiority of the IncLabDA model.

Synthetic unknown class learning for learning unknowns

This paper addresses the open set recognition (OSR) problem, where the goal is to correctly classify samples of known classes while detecting unknown samples to reject. In the OSR problem, “unknown” is assumed to have infinite possibilities because we have no knowledge about unknowns until they emerge. Intuitively, the more an OSR system explores the possibilities of unknowns, the more likely it is to detect unknowns. Even though several generative OSR models have been proposed to explore more by generating synthetic samples and learning them as unknowns, the generated samples are limited to a small subspace of the known classes. Thus, this paper proposes a novel synthetic unknown class learning method that constantly generates unknown-like samples while maintaining diversity between the generated samples. By learning the unknown-like samples and known samples in an alternating manner, the proposed method can not only experience diverse synthetic unknowns but also reduce overgeneralization with respect to known classes. Experiments on several benchmark datasets show that the proposed method significantly outperforms other state-of-the-art approaches by generating diverse realistic unknown samples.