Multi-class Approach Research Articles

Face spoofing detection ensemble via multistage optimisation and pruning

Despite the recent improvements in facial recognition, face spoofing attacks can still pose a serious security threat to biometric systems. As fraudsters are coming up with novel spoofing attacks, anomaly-based detectors, compared to the binary spoofing attack counterparts, have certain generalisation performance advantages. In this work, we investigate the merits of fusing multiple anomaly classifiers using weighted averaging (WA) fusion. The design of the entire system is based on genuine-access data only. To optimise the parameters of WA, we propose a novel three-stage optimisation method with the following contributions: (a) A new hybrid optimisation method using Genetic Algorithm (GA) and Pattern Search (PS) to explore the weight space more effectively (b) a novel two-sided score normalisation method to improve the anomaly detection performance (c) a new ensemble pruning method to improve the generalisation performance. To further boost the performance of the proposed anomaly detection ensemble, we incorporate client-specific information to train the proposed model. We evaluate the capability of the proposed model on publicly available face spoofing databases including Replay-Attack, Replay-Mobile and Rose-Youtu. The experimental results demonstrate that the proposed WA fusion outperforms the state-of-the-art anomaly-based and multiclass approaches.

Defining user spectra to classify Ethereum users based on their behavior

PurposeIn this paper, we define the concept of user spectrum and adopt it to classify Ethereum users based on their behavior.Design/methodology/approachGiven a time period, our approach associates each user with a spectrum showing the trend of some behavioral features obtained from a social network-based representation of Ethereum. Each class of users has its own spectrum, obtained by averaging the spectra of its users. In order to evaluate the similarity between the spectrum of a class and the one of a user, we propose a tailored similarity measure obtained by adapting to this context some general measures provided in the past. Finally, we test our approach on a dataset of Ethereum transactions.FindingsWe define a social network-based model to represent Ethereum. We also define a spectrum for a user and a class of users (i.e., token contract, exchange, bancor and uniswap), consisting of suitable multivariate time series. Furthermore, we propose an approach to classify new users. The core of this approach is a metric capable of measuring the similarity degree between the spectrum of a user and the one of a class of users. This metric is obtained by adapting the Eros distance (i.e., Extended Frobenius Norm) to this scenario.Originality/valueThis paper introduces the concept of spectrum of a user and a class of users, which is new for blockchains. Differently from past models, which represented user behavior by means of univariate time series, the user spectrum here proposed exploits multivariate time series. Moreover, this paper shows that the original Eros distance does not return satisfactory results when applied to user and class spectra, and proposes a modified version of it, tailored to the reference scenario, which reaches a very high accuracy. Finally, it adopts spectra and the modified Eros distance to classify Ethereum users based on their past behavior. Currently, no multi-class automatic classification approach tailored to Ethereum exists yet, albeit some single-class ones have been recently proposed. Therefore, the only way to classify users in Ethereum are online services (e.g., Etherscan), where users are classified after a request from them. However, the fraction of users thus classified is low. To address this issue, we present an automatic approach for a multi-class classification of Ethereum users based on their past behavior.

A machine learning approach to identify hand actions from single-channel sEMG signals.

Surface Electromyographic (sEMG) signal is a prime source of information to activate prosthetic hand such that it is able to restore a few basic hand actions of amputee, making it suitable for rehabilitation. In this work, a non-invasive single channel sEMG amplifier is developed that captures sEMG signal for three typical hand actions from the lower elbow muscles of able bodied subjects and amputees. The recorded sEMG signal detrends and has frequencies other than active frequencies. The Empirical Mode Decomposition Detrending Fluctuation Analysis (EMD-DFA) is attempted to de-noise the sEMG signal. A feature vector is formed by extracting eight features in time domain, seven features each in spectral and wavelet domain. Prominent features are selected by Fuzzy Entropy Measure (FEM) to ease the computational complexity and reduce the recognition time of classification. Classification of different hand actions is attempted based on multi-class approach namely Partial Least Squares Discriminant Analysis (PLS-DA) to control the prosthetic hand. It is inferred that an accuracy of 89.72% & 84% is observed for the pointing action whereas the accuracy for closed fist is81.2% & 79.54% while for spherical grasp it is 80.6% & 76% respectively for normal subjects and amputees. The performance of the classifier is compared with Linear Discriminant Analysis (LDA) and an improvement of 5% in mean accuracy is observed for both normal subjects and amputees. The mean accuracy for all the three different hand actions is significantly high (83.84% & 80.18%) when compared with LDA. The proposed work frame provides afair mean accuracy in classifying the hand actions of amputees. This methodology thus appears to be useful in actuating the prosthetic hand.

An integrated approach on verification of signatures using multiple classifiers (SVM and Decision Tree): A multi-classification approach

A signature is a handwritten representation that is commonly used to validate and recognize the writer individually. An automated verification system is mandatory to verify the identity. The signature essentially displays a variety of dynamics and the static characteristics differ with time and place. Many scientists have already found different algorithms to boost the signature verification system function extraction point. The paper is aimed at multiplying two different ways to solve the problem in digital, manual, or some other means of verifying signatures. The various characteristics of the signature were found through the most adequately implemented methods of machine learning (support vector and decision tree). In addition, the characteristics were listed after measuring the effects. An experiment was performed in various language databases. More precision was obtained from the feature.

Multilabel convolution neural network for facial expression recognition and ordinal intensity estimation.

Facial Expression Recognition (FER) has gained considerable attention in affective computing due to its vast area of applications. Diverse approaches and methods have been considered for a robust FER in the field, but only a few works considered the intensity of emotion embedded in the expression. Even the available studies on expression intensity estimation successfully assigned a nominal/regression value or classified emotion in a range of intervals. Most of the available works on facial expression intensity estimation successfully present only the emotion intensity estimation. At the same time, others proposed methods that predict emotion and its intensity in different channels. These multiclass approaches and extensions do not conform to man heuristic manner of recognising emotion and its intensity estimation. This work presents a Multilabel Convolution Neural Network (ML-CNN)-based model, which could simultaneously recognise emotion and provide ordinal metrics as the intensity estimation of the emotion. The proposed ML-CNN is enhanced with the aggregation of Binary Cross-Entropy (BCE) loss and Island Loss (IL) functions to minimise intraclass and interclass variations. Also, ML-CNN model is pre-trained with Visual Geometric Group (VGG-16) to control overfitting. In the experiments conducted on Binghampton University 3D Facial Expression (BU-3DFE) and Cohn Kanade extension (CK+) datasets, we evaluate ML-CNN’s performance based on accuracy and loss. We also carried out a comparative study of our model with some popularly used multilabel algorithms using standard multilabel metrics. ML-CNN model simultaneously predicts emotion and intensity estimation using ordinal metrics. The model also shows appreciable and superior performance over four standard multilabel algorithms: Chain Classifier (CC), distinct Random K label set (RAKEL), Multilabel K Nearest Neighbour (MLKNN) and Multilabel ARAM (MLARAM).

Multi-Class Strategies for Joint Building Footprint and Road Detection in Remote Sensing

Building footprints and road networks are important inputs for a great deal of services. For instance, building maps are useful for urban planning, whereas road maps are essential for disaster response services. Traditionally, building and road maps are manually generated by remote sensing experts or land surveying, occasionally assisted by semi-automatic tools. In the last decade, deep learning-based approaches have demonstrated their capabilities to extract these elements automatically and accurately from remote sensing imagery. The building footprint and road network detection problem can be considered a multi-class semantic segmentation task, that is, a single model performs a pixel-wise classification on multiple classes, optimizing the overall performance. However, depending on the spatial resolution of the imagery used, both classes may coexist within the same pixel, drastically reducing their separability. In this regard, binary decomposition techniques, which have been widely studied in the machine learning literature, are proved useful for addressing multi-class problems. Accordingly, the multi-class problem can be split into multiple binary semantic segmentation sub-problems, specializing different models for each class. Nevertheless, in these cases, an aggregation step is required to obtain the final output labels. Additionally, other novel approaches, such as multi-task learning, may come in handy to further increase the performance of the binary semantic segmentation models. Since there is no certainty as to which strategy should be carried out to accurately tackle a multi-class remote sensing semantic segmentation problem, this paper performs an in-depth study to shed light on the issue. For this purpose, open-access Sentinel-1 and Sentinel-2 imagery (at 10 m) are considered for extracting buildings and roads, making use of the well-known U-Net convolutional neural network. It is worth stressing that building and road classes may coexist within the same pixel when working at such a low spatial resolution, setting a challenging problem scheme. Accordingly, a robust experimental study is developed to assess the benefits of the decomposition strategies and their combination with a multi-task learning scheme. The obtained results demonstrate that decomposing the considered multi-class remote sensing semantic segmentation problem into multiple binary ones using a One-vs.-All binary decomposition technique leads to better results than the standard direct multi-class approach. Additionally, the benefits of using a multi-task learning scheme for pushing the performance of binary segmentation models are also shown.

Visualized Malware Multi-Classification Framework Using Fine-Tuned CNN-Based Transfer Learning Models

There is a massive growth in malicious software (Malware) development, which causes substantial security threats to individuals and organizations. Cybersecurity researchers makes continuous efforts to defend against these malware risks. This research aims to exploit the significant advantages of Transfer Learning (TL) and Fine-Tuning (FT) methods to introduce efficient malware detection in the context of imbalanced families without the need to apply complex features extraction or data augmentation processes. Therefore, this paper proposes a visualized malware multi-classification framework to avoid false positives and imbalanced datasets’ challenges through using the fine-tuned convolutional neural network (CNN)-based TL models. The proposed framework comprises eight different FT CNN models including VGG16, AlexNet, DarkNet-53, DenseNet-201, Inception-V3, Places365-GoogleNet, ResNet-50, and MobileNet-V2. First, the binary files of different malware families were transformed into 2D images and then forwarded to the FT CNN models to detect and classify the malware families. The detection and classification performance was examined on a benchmark Malimg imbalanced dataset using different, comprehensive evaluation metrics. The evaluation results prove the FT CNN models’ significance in detecting malware types with high accuracy that reached 99.97% which also outperforms the performance of related machine learning (ML) and deep learning (DL)-based malware multi-classification approaches tested on the same malware dataset.

Mapping Alkaline Fens, Transition Mires and Quaking Bogs Using Airborne Hyperspectral and Laser Scanning Data

The aim of this study is to evaluate the effectiveness of the identification of Natura 2000 wetland habitats (Alkaline fens—code 7230, and Transition mires and quaking bogs—code 7140) depending on various remotely sensed (RS) data acquired from an airborne platform. Both remote sensing data and botanical reference data were gathered for mentioned habitats in the Lower (LB) and Upper Biebrza (UB) River Valley and the Janowskie Forest (JF) in different seasonal stages. Several different classification scenarios were tested, and the ones that gave the best results for analyzed habitats were indicated in each campaign. In the final stage, a recommended term of data acquisition, as well as a list of remote sensing products, which allowed us to achieve the highest accuracy mapping for these two types of wetland habitats, were presented. Designed classification scenarios integrated different hyperspectral products such as Minimum Noise Fraction (MNF) bands, spectral indices and products derived from Airborne Laser Scanning (ALS) data representing topography (developed in SAGA), or statistical products (developed in OPALS—Orientation and Processing of Airborne Laser Scanning). The image classifications were performed using a Random Forest (RF) algorithm and a multi-classification approach. As part of the research, the correlation analysis of the developed remote sensing products was carried out, and the Recursive Feature Elimination with Cross-Validation (RFE-CV) analysis was performed to select the most important RS sub-products and thus increase the efficiency and accuracy of developing the final habitat distribution maps. The classification results showed that alkaline fens are better identified in summer (mean F1-SCORE equals 0.950 in the UB area, and 0.935 in the LB area), transition mires and quaking bogs that evolved on/or in the vicinity of alkaline fens in summer and autumn (mean F1-SCORE equals 0.931 in summer, and 0.923 in autumn in the UB area), and transition mires and quaking bogs that evolved on dystrophic lakes in spring and summer (mean F1-SCORE equals 0.953 in spring, and 0.948 in summer in the JF area). The study also points out that the classification accuracy of both wetland habitats is highly improved when combining selected hyperspectral products (MNF bands, spectral indices) with ALS topographical and statistical products. This article demonstrates that information provided by the synergetic use of data from different sensors can be used in mapping and monitoring both Natura 2000 wetland habitats for its future functional assessment and/or protection activities planning with high accuracy.

A Machine Learning Multi-Class Approach for Fall Detection Systems Based on Wearable Sensors with a Study on Sampling Rates Selection.

Falls are dangerous for the elderly, often causing serious injuries especially when the fallen person stays on the ground for a long time without assistance. This paper extends our previous work on the development of a Fall Detection System (FDS) using an inertial measurement unit worn at the waist. Data come from SisFall, a publicly available dataset containing records of Activities of Daily Living and falls. We first applied a preprocessing and a feature extraction stage before using five Machine Learning algorithms, allowing us to compare them. Ensemble learning algorithms such as Random Forest and Gradient Boosting have the best performance, with a Sensitivity and Specificity both close to 99%. Our contribution is: a multi-class classification approach for fall detection combined with a study of the effect of the sensors’ sampling rate on the performance of the FDS. Our multi-class classification approach splits the fall into three phases: pre-fall, impact, post-fall. The extension to a multi-class problem is not trivial and we present a well-performing solution. We experimented sampling rates between 1 and 200 Hz. The results show that, while high sampling rates tend to improve performance, a sampling rate of 50 Hz is generally sufficient for an accurate detection.

Fault Identification of Photovoltaic Array Based on Machine Learning Classifiers

Fault identification in Photovoltaic (PV) array is a contemporary research topic motivated by the higher penetration levels of PV systems in recent electrical grids. Therefore, this work aims to define an optimal Machine learning (ML) structure of automatic detection and diagnosis algorithm for common PV array faults, namely, permanent (Arc Fault, Line-to-Line, Maximum Power Point Tracking unit failure, and Open-Circuit faults), and temporary (Shading) under a wide range of climate datasets, fault impedances, and shading scenarios. To achieve the best-fit ML structure, three distinct ML classifiers are compared, namely, Decision Tree (DT) based on different splitting criteria, K-Nearest Neighbors (KNN) based on the different metrics of distance and weighting functions, and Support Vector Machine (SVM) based on different Kernel functions and multi-classification approaches. Also, Bayesian Optimization is adopted to assign the optimal hyperparameters to the fault classifiers. To investigate the performance of classifiers reported, both simulation and experimental case studies are carried out and presented.

A Hybrid Artificial Intelligence Model for Skin Cancer Diagnosis

Melanoma or skin cancer is the most dangerous and deadliest disease. As the incidence and mortality rate of skin cancer increases worldwide, an automated skin cancer detection/classification system is required for early detection and prevention of skin cancer. In this study, a Hybrid Artificial Intelligence Model (HAIM) is designed for skin cancer classification. It uses diverse multi-directional representation systems for feature extraction and an efficient Exponentially Weighted and Heaped Multi-Layer Perceptron (EWHMLP) for the classification. Though the wavelet transform is a powerful tool for signal and image processing, it is unable to detect the intermediate dimensional structures of a medical image. Thus the proposed HAIM uses Curvelet (CurT), Contourlet (ConT) and Shearlet (SheT) transforms as feature extraction techniques. Though MLP is very flexible and well suitable for the classification problem, the learning of weights is a challenging task. Also, the optimization process does not converge, and the model may not be stable. To overcome these drawbacks, EWHMLP is developed. Results show that the combined qualities of each transform in a hybrid approach provides an accuracy of 98.33% in a multi-class approach on PH2 database.

A methodical evaluation of classifiers in predicting academic performance for a multi-class approach

A methodical evaluation of classifiers in predicting academic performance for a multi-class approach

Validation of a Multiclass Method for the Screening of 15 Antibiotic Residues in Milk Using Biochip Multi-array Technology and Its application to Monitor Real Samples.

Veterinary drugs are extensively and legally consumed to treat and prevent disease in chattels but some are also used illegally as growth-stimulating agents. Inappropriate or intensive use of antibiotics can cause allergic reactions and, above all, antibiotic resistance. A multiclass approach for the screening of antimicrobial substances in milk was validated in consonance with Commission Decision 2002/657/EC and to the European guideline for the validation of screening methods for veterinary medicines. This biochip-based approach enables the simultaneous determination of a total of 13 sulphonamide, dapsone and trimethoprim. For monitoring of antibiotic residues, 53 UHT milk samples collected from Tehran, IR Iran were screened applying this technology. The result showed that for all antibiotic residues, the positivity threshold T was much more than the cut-off value Fm. A false positive rate of less than 5% was found for all antibiotics which are satisfactory. All detection capabilities (CCβ) were well below the Maximum Residue Level (MRL) set by the European Commission (100 µg/kg for the sum of all sulphonamides and 50 µg/kg for trimethoprim in milk). The screening results of 53 milk samples showed that 71.7% of samples were compliant and all positive samples were below the MRL set by European Commission. This study showed that the biochip-based technique is valid to identify and quantify antibiotic residues in milk at the studied validation levels. The method was rapid, easy, safe, and able to screen 13 sulphonamide, dapsone and trimethoprim from a single milk sample simultaneously with no sample preparation procedure (or just one-step centrifugation).

A methodical evaluation of classifiers in predicting academic performance for a multi-class approach

Predictive analytics has gained importance in recent years as it helps to proactively identify factors that contribute to the success or failure of an event in the relevant field. Academic achievements of students can be predicted early by employing algorithms and analysing relevant data thereby devising solutions to improve performance. In this process choosing the right algorithm is very crucial since performance of algorithms vary depending on the distribution of data and the way it is tuned to handle the data. In order to enhance the performance of algorithms their hyper-parameters were tuned. Many multi-class classifiers were examined and the prediction accuracy of each model developed by employing them was compared. Depending on their classification accuracy the models developed were used to predict the performance of the students. This was done by using micro and macro averaging because of multi-class features. The results show that ensemble classifiers performed well than their individual counterparts.

Algorithmically generated malicious domain names detection based on n-grams features

Botnets are one of the major cyber infections used in several criminal activities. In most botnets, a Domain Generation Algorithm (DGA) is used by bots to make DNS queries aimed at establishing the connection with the Command and Control (C&C) server. The identification of such queries by monitoring the network DNS traffic is then crucial for bot detection. In this paper we present a methodology to detect DGA generated domain names based on a supervised machine learning process, trained with a dataset of known benign and malicious domain names. The proposed approach represents the domain names through a set of features which express the similarity between the 2-grams and 3-grams in a single unclassified domain name and those in domain names known as malicious or benign. We used the Kullback-Leibner divergence and the Jaccard Index to estimate the similarity, and we tested different machine learning algorithms to classify each domain name as benign or DGA-based (with both binary and multi-class approach). The results of our experiments demonstrate that the proposed methodology, which only exploits lexical features of domain names, attains a good level of accuracy and results in a general model able to classify previously unseen domains in an effective way. It is also able to outperform some of the state-of-the-art featur eless classification methods based on deep learning.

Challenges and future directions in LC-MS-based multiclass method development for the quantification of food contaminants

Monitoring of food contaminants and residues has undergone a significant improvement in recent years and is now performed in an intensive manner. Achievements in the area of chromatography-mass spectrometry coupling techniques enabled the development of quantitative multi-target approaches covering several hundred analytes. Although the majority of methods are focusing on the analysis of one specific group of substances, such as pesticides, mycotoxins, or veterinary drugs, current trends are going towards the simultaneous determination of multiclass compounds from several families of contaminants and residues. This work provides an overview of relevant multiclass concepts based on LC-MS/MS and LC-HRMS instruments. Merits and shortcomings will be critically discussed based on current performance characteristics of the EU legislation system. In addition, the discussion of a recently developed multiclass approach covering >1000 substances is presented as a case study to illustrate the current developments in this area.

Multi-class classification of biomechanical data: A functional LDA approach based on multi-class penalized functional PLS

A functional linear discriminant analysis approach to classify a set of kinematic data (human movement curves of individuals performing different physical activities) is performed. Kinematic data, usually collected in linear acceleration or angular rotation format, can be identified with functions in a continuous domain (time, percentage of gait cycle, etc.). Since kinematic curves are measured in the same sample of individuals performing different activities, they are a clear example of functional data with repeated measures. On the other hand, the sample curves are observed with noise. Then, a roughness penalty might be necessary in order to provide a smooth estimation of the discriminant functions, which would make them more interpretable. Moreover, because of the infinite dimension of functional data, a reduction dimension technique should be considered. To solve these problems, we propose a multi-class approach for penalized functional partial least squares (FPLS) regression. Then linear discriminant analysis (LDA) will be performed on the estimated FPLS components. This methodology is motivated by two case studies. The first study considers the linear acceleration recorded every two seconds in 30 subjects, related to three different activities (walking, climbing stairs and down stairs). The second study works with the triaxial angular rotation, for each joint, in 51 children when they completed a cycle walking under three conditions (walking, carrying a backpack and pulling a trolley). A simulation study is also developed for comparing the performance of the proposed functional LDA with respect to the corresponding multivariate and non-penalized approaches.

Binary and multi-class classification for androgen receptor agonists, antagonists and binders

Androgens and androgen receptor regulate a variety of biological effects in the human body. The impaired functioning of androgen receptor may have different adverse health effects from cancer to infertility. Therefore, it is important to determine whether new chemicals have any binding activity and act as androgen agonists or antagonists before commercial use. Due to the large number of chemicals that require experimental testing, the computational methods are a viable alternative. Therefore, the aim of the present study was to develop predictive QSAR models for classifying compounds according to their activity at the androgen receptor. A large data set of chemicals from the CoMPARA project was used for this purpose and random forest classification models have been developed for androgen binding, agonistic, and antagonistic activity. In addition, a unique effort has been made for multi-class approach that discriminates between inactive compounds, agonists and antagonists simultaneously. For the evaluation set, the classification models predicted agonists with 80% of accuracy and for the antagonists’ and binders’ the respective metrics were 72% and 78%. Combining agonists, antagonists and inactive compounds into a multi-class approach added complexity to the modelling task and resulted to 64% prediction accuracy for the evaluation set. Considering the size of the training data sets and their imbalance, the achieved evaluation accuracy is very good. The final classification models are available for exploring and predicting at QsarDB repository (https://doi.org/10.15152/QDB.236).

Realizing the simultaneous liquid chromatography-tandem mass spectrometry based quantification of >1200 biotoxins, pesticides and veterinary drugs in complex feed

The first quantitative multiclass approach enabling the accurate quantification of >1200 biotoxins, pesticides and veterinary drugs in complex feed using liquid chromatography tandem mass spectrometry (LC-MS/MS) has been developed. Optimization of HPLC/UHPLC (chromatographic column, flow rate and injection volume) and MS/MS conditions (dwell time and cycle time) were carried out in order to allow the combination of five major substance classes and the high number of target analytes with different physico-chemical properties. Cycle times and retention windows were carefully optimized and ensured appropriate dwell times reducing the overall measurement error. Validation was carried out in two compound feed matrices according to the EU SANTE validation guideline. Apparent recoveries matching the acceptable range of 60-140% accounted 60% and 79% for all analytes in cattle and chicken feed, respectively. High extraction efficiencies were obtained for all analyte/matrix combinations and revealed matrix effects as the main source for deviation of the targeted performance criteria. Concerning the methods repeatability 99% of all analytes in chicken and 96% in cattle feed complied with the acceptable RSD ≤ 20% criterion. Limits of quantification were between 1-10 µg/kg for the vast majority of compounds. Finally, the methods applicability was tested in >130 real compound feed samples and provides first insights into co-exposure of agro-contaminants in animal feed.

Abstract 257: Whole miRNome profiling in fecal and plasma exosome samples for the diagnosis of colorectal cancer

Abstract Colorectal cancer (CRC) ranks as the second most common non-sex-specific cancer worldwide and the second cause of cancer deaths. Screening programs for CRC based on the fecal immunochemical test (FIT) are effective in average-risk population for the tumor detection but has low sensitivity for the recognition of advanced adenomas. We identified microRNA (miRNA) signatures by next-generation sequencing in fecal samples that could detect patients with adenomas or CRC. The same signatures were also evaluated in plasma extracellular vesicles (EVs) from the same patients. A genome-wide miRNA expression profiling was initially performed in 221 subjects (80 healthy subjects; 43 adenoma patients; 41 individuals with inflammatory bowel disease [IBD]; and 57 CRC) recruited in a cross-sectional study in Italy where stool and plasma samples have been consecutively collected at colonoscopy. Several differentially expressed miRNAs, mostly in stool samples, have been identified both in common or peculiar of specific comparison among the 3 categories of patients vs healthy subjects. Subsequently, a similar profiling has been performed for a set of 24 paired colorectal tumor and adjacent non-tumor samples from the group of CRC patients already samples for stool and plasma. Several miRNAs (n=137) resulted altered in cancer tissues compared to non-tumor; on the other hand, 91 miRNAs resulted altered in stool and only 6 in plasma EVs. Analyzing stool samples from the whole Italian cohort we observed a good overlap of the altered miRNAs observed in tissues (miR-148a-3p, miR-182-5p, miR-143-3p, miR-12136-3p; miR-378a-3p, miR-1290-5p, miR-21-5p, miR-7704-3p). The identified differentially expressed miRNAs in stool and plasma EVs were validated in an independent cohort from the Czech Republic (22 healthy, 19 Inflammation, 21 Adenoma, 34 CRC) with a similar approach. Preliminary results in this independent cohort showed that 19, 12 and 7 miRNAs were differentially expressed when comparing CRC with healthy, CRC and adenoma and inflammation individuals, respectively. In the comparison of CRC versus healthy subjects we observed a good overlap of miRNAs differentially expressed between the Czech and Italian cohorts: 10 miRNAs (miR-148a-3p, miR-1181-3p, miR-1290-5p, miR-4488-3p, miR-1246-3p, let-7i-5p, let-7b-5p, miR-4497-3p, miR-21-5p, miR-149-3p) were significantly altered in the same directions in the 2 cohorts. Moreover, hsa-miR-1290-5p and hsa-miR-4488-3p resulted significantly upregulated when going from healthy, inflammation, adenoma and CRC both in the Italian and in the Czech cohorts. Finally, with the application of a multi-class machine learning approach, we identified a signature of 13 miRNAs whose expression was able to accurately classify (AUC = 0.90) the four classes of subjects in combination with the information of patient age and sex. These results showed that there could be a potential application of the analyses of nucleic acids in stool as adjuvant in non-invasive screening. Citation Format: Sonia Tarallo, Antonio Francavilla, Giulio Ferrero, Francesca Cordero, Gaetano Gallo, Andrew Maltez Thomas, Paolo Manghi, Veronika Vymetalkova, Ludmila Vodickova, Nicola Segata, Pavel Vodicka, Barbara Pardini, Alessio Naccarati. Whole miRNome profiling in fecal and plasma exosome samples for the diagnosis of colorectal cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 257.