Search results for: classifier

Authors: Souvik Phadikar, Nidul Sinha, Rajdeep Ghosh

Abstract:

In brainwaves research using electroencephalogram (EEG) signals, finding the most relevant and effective feature set for identification of activities in the human brain is a big challenge till today because of the random nature of the signals. The feature extraction method is a key issue to solve this problem. Finding those features that prove to give distinctive pictures for different activities and similar for the same activities is very difficult, especially for the number of activities. The performance of a classifier accuracy depends on this quality of feature set. Further, more number of features result in high computational complexity and less number of features compromise with the lower performance. In this paper, a novel idea of the selection of optimal feature set using a heuristically optimized deep autoencoder is presented. Using various feature extraction methods, a vast number of features are extracted from the EEG signals and fed to the autoencoder deep neural network. The autoencoder encodes the input features into a small set of codes. To avoid the gradient vanish problem and normalization of the dataset, a meta-heuristic search algorithm is used to minimize the mean square error (MSE) between encoder input and decoder output. To reduce the feature set into a smaller one, 4 hidden layers are considered in the autoencoder network; hence it is called Heuristically Optimized Deep Autoencoder (HO-DAE). In this method, no features are rejected; all the features are combined into the response of responses of the hidden layer. The results reveal that higher accuracy can be achieved using optimal reduced features. The proposed HO-DAE is also compared with the regular autoencoder to test the performance of both. The performance of the proposed method is validated and compared with the other two methods recently reported in the literature, which reveals that the proposed method is far better than the other two methods in terms of classification accuracy.

Keywords: autoencoder, brainwave signal analysis, electroencephalogram, feature extraction, feature selection, optimization

Procedia PDF Downloads 90

Authors: Abdulaziz Alsadhan, Naveed Khan

Abstract:

In recent years intrusions on computer network are the major security threat. Hence, it is important to impede such intrusions. The hindrance of such intrusions entirely relies on its detection, which is primary concern of any security tool like Intrusion Detection System (IDS). Therefore, it is imperative to accurately detect network attack. Numerous intrusion detection techniques are available but the main issue is their performance. The performance of IDS can be improved by increasing the accurate detection rate and reducing false positive. The existing intrusion detection techniques have the limitation of usage of raw data set for classification. The classifier may get jumble due to redundancy, which results incorrect classification. To minimize this problem, Principle Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Local Binary Pattern (LBP) can be applied to transform raw features into principle features space and select the features based on their sensitivity. Eigen values can be used to determine the sensitivity. To further classify, the selected features greedy search, back elimination, and Particle Swarm Optimization (PSO) can be used to obtain a subset of features with optimal sensitivity and highest discriminatory power. These optimal feature subset used to perform classification. For classification purpose, Support Vector Machine (SVM) and Multilayer Perceptron (MLP) used due to its proven ability in classification. The Knowledge Discovery and Data mining (KDD’99) cup dataset was considered as a benchmark for evaluating security detection mechanisms. The proposed approach can provide an optimal intrusion detection mechanism that outperforms the existing approaches and has the capability to minimize the number of features and maximize the detection rates.

Keywords: Particle Swarm Optimization (PSO), Principle Component Analysis (PCA), Linear Discriminant Analysis (LDA), Local Binary Pattern (LBP), Support Vector Machine (SVM), Multilayer Perceptron (MLP)

Procedia PDF Downloads 336

Authors: Zachary Pan

Abstract:

Our general mission is to solve the problem of classifying images into different body part types and deciding if each of them is healthy or not. However, for now, we will determine healthiness for only one-sixth of the body parts, specifically the chest. We will detect pneumonia in X-ray scans of those chest images. With this type of AI, doctors can use it as a second opinion when they are taking CT or X-ray scans of their patients. Another ad-vantage of using this machine learning classifier is that it has no human weaknesses like fatigue. The overall ap-proach to this problem is to split the problem into two parts: first, classify the image, then determine if it is healthy. In order to classify the image into a specific body part class, the body parts dataset must be split into test and training sets. We can then use many models, like neural networks or logistic regression models, and fit them using the training set. Now, using the test set, we can obtain a realistic accuracy the models will have on images in the real world since these testing images have never been seen by the models before. In order to increase this testing accuracy, we can also apply many complex algorithms to the models, like multiplicative weight update. For the second part of the problem, to determine if the body part is healthy, we can have another dataset consisting of healthy and non-healthy images of the specific body part and once again split that into the test and training sets. We then use another neural network to train on those training set images and use the testing set to figure out its accuracy. We will do this process only for the chest images. A major conclusion reached is that convolutional neural networks are the most reliable and accurate at image classification. In classifying the images, the logistic regression model, the neural network, neural networks with multiplicative weight update, neural networks with the black box algorithm, and the convolutional neural network achieved 96.83 percent accuracy, 97.33 percent accuracy, 97.83 percent accuracy, 96.67 percent accuracy, and 98.83 percent accuracy, respectively. On the other hand, the overall accuracy of the model that de-termines if the images are healthy or not is around 78.37 percent accuracy.

Keywords: body part, healthcare, machine learning, neural networks

Procedia PDF Downloads 70

Authors: Ermias A. Tegegn, Million Meshesha

Abstract:

Human Immunodeficiency Virus and Acquired Immunodeficiency Syndrome (HIV/AIDS) is a major cause of death for most African countries. Ethiopia is one of the seriously affected countries in sub Saharan Africa. Previously in Ethiopia, having HIV/AIDS was almost equivalent to a death sentence. With the introduction of Antiretroviral Therapy (ART), HIV/AIDS has become chronic, but manageable disease. The study focused on a data mining technique to predict future living status of HIV/AIDS patients at the time of drug regimen change when the patients become toxic to the currently taking ART drug combination. The data is taken from University of Gondar Hospital ART program database. Hybrid methodology is followed to explore the application of data mining on ART program dataset. Data cleaning, handling missing values and data transformation were used for preprocessing the data. WEKA 3.7.9 data mining tools, classification algorithms, and expertise are utilized as means to address the research problem. By using four different classification algorithms, (i.e., J48 Classifier, PART rule induction, Naïve Bayes and Neural network) and by adjusting their parameters thirty-two models were built on the pre-processed University of Gondar ART program dataset. The performances of the models were evaluated using the standard metrics of accuracy, precision, recall, and F-measure. The most effective model to predict the status of HIV patients with drug regimen substitution is pruned J48 decision tree with a classification accuracy of 98.01%. This study extracts interesting attributes such as Ever taking Cotrim, Ever taking TbRx, CD4 count, Age, Weight, and Gender so as to predict the status of drug regimen substitution. The outcome of this study can be used as an assistant tool for the clinician to help them make more appropriate drug regimen substitution. Future research directions are forwarded to come up with an applicable system in the area of the study.

Keywords: HIV drug regimen, data mining, hybrid methodology, predictive model

Procedia PDF Downloads 110

Authors: Thiago Spilborghs Bueno Meyer, Plinio Thomaz Aquino Junior

Abstract:

Through speech, which privileges the functional and interactive nature of the text, it is possible to ascertain the spatiotemporal circumstances, the conditions of production and reception of the discourse, the explicit purposes such as informing, explaining, convincing, etc. These conditions allow bringing the interaction between humans closer to the human-robot interaction, making it natural and sensitive to information. However, it is not enough to understand what is said; it is necessary to recognize emotions for the desired interaction. The validity of the use of neural networks for feature selection and emotion recognition was verified. For this purpose, it is proposed the use of neural networks and comparison of models, such as recurrent neural networks and deep neural networks, in order to carry out the classification of emotions through speech signals to verify the quality of recognition. It is expected to enable the implementation of robots in a domestic environment, such as the HERA robot from the RoboFEI@Home team, which focuses on autonomous service robots for the domestic environment. Tests were performed using only the Mel-Frequency Cepstral Coefficients, as well as tests with several characteristics of Delta-MFCC, spectral contrast, and the Mel spectrogram. To carry out the training, validation and testing of the neural networks, the eNTERFACE’05 database was used, which has 42 speakers from 14 different nationalities speaking the English language. The data from the chosen database are videos that, for use in neural networks, were converted into audios. It was found as a result, a classification of 51,969% of correct answers when using the deep neural network, when the use of the recurrent neural network was verified, with the classification with accuracy equal to 44.09%. The results are more accurate when only the Mel-Frequency Cepstral Coefficients are used for the classification, using the classifier with the deep neural network, and in only one case, it is possible to observe a greater accuracy by the recurrent neural network, which occurs in the use of various features and setting 73 for batch size and 100 training epochs.

Keywords: emotion recognition, speech, deep learning, human-robot interaction, neural networks

Procedia PDF Downloads 128

Authors: Reeta Devi, Hitender Kumar Tyagi, Dinesh Kumar

Abstract:

In present scenario, cardiovascular problems are growing challenge for researchers and physiologists. As heart disease have no geographic, gender or socioeconomic specific reasons; detecting cardiac irregularities at early stage followed by quick and correct treatment is very important. Electrocardiogram is the finest tool for continuous monitoring of heart activity. Heart rate variability (HRV) is used to measure naturally occurring oscillations between consecutive cardiac cycles. Analysis of this variability is carried out using time domain, frequency domain and non-linear parameters. This paper presents HRV analysis of the online dataset for normal sinus rhythm (taken as healthy subject) and sudden cardiac death (SCD subject) using all three methods computing values for parameters like standard deviation of node to node intervals (SDNN), square root of mean of the sequences of difference between adjacent RR intervals (RMSSD), mean of R to R intervals (mean RR) in time domain, very low-frequency (VLF), low-frequency (LF), high frequency (HF) and ratio of low to high frequency (LF/HF ratio) in frequency domain and Poincare plot for non linear analysis. To differentiate HRV of healthy subject from subject died with SCD, k –nearest neighbor (k-NN) classifier has been used because of its high accuracy. Results show highly reduced values for all stated parameters for SCD subjects as compared to healthy ones. As the dataset used for SCD patients is recording of their ECG signal one hour prior to their death, it is therefore, verified with an accuracy of 95% that proposed algorithm can identify mortality risk of a patient one hour before its death. The identification of a patient’s mortality risk at such an early stage may prevent him/her meeting sudden death if in-time and right treatment is given by the doctor.

Keywords: early stage prediction, heart rate variability, linear and non-linear analysis, sudden cardiac death

Procedia PDF Downloads 317

Authors: Xiaoyang Zhao, Kaiying Wang

Abstract:

The choice of stocking density in poultry farming is a potential way for determining welfare level of poultry. However, it is difficult to measure stocking densities in poultry farming because of a lot of variables such as species, age and weight, feeding way, house structure and geographical location in different broiler houses. A method was proposed in this paper to measure the differences of young broilers reared under different stocking densities by sound analysis. Vocalisations of broilers were recorded and analysed under different stocking densities to identify the relationship between sounds and stocking densities. Recordings were made continuously for three-week-old chickens in order to evaluate the variation of sounds emitted by the animals at the beginning. The experimental trial was carried out in an indoor reared broiler farm; the audio recording procedures lasted for 5 days. Broilers were divided into 5 groups, stocking density treatments were 8/m², 10/m², 12/m² (96birds/pen), 14/m² and 16/m², all conditions including ventilation and feed conditions were kept same except from stocking densities in every group. The recordings and analysis of sounds of chickens were made noninvasively. Sound recordings were manually analysed and labelled using sound analysis software: GoldWave Digital Audio Editor. After sound acquisition process, the Mel Frequency Cepstrum Coefficients (MFCC) was extracted from sound data, and the Support Vector Machine (SVM) was used as an early detector and classifier. This preliminary study, conducted in an indoor reared broiler farm shows that this method can be used to classify sounds of chickens under different densities economically (only a cheap microphone and recorder can be used), the classification accuracy is 85.7%. This method can predict the optimum stocking density of broilers with the complement of animal welfare indicators, animal productive indicators and so on.

Keywords: broiler, stocking density, poultry farming, sound monitoring, Mel Frequency Cepstrum Coefficients (MFCC), Support Vector Machine (SVM)

Procedia PDF Downloads 118

Authors: Anželika Teresė

Abstract:

In order to solve issues in sign language linguistics, address matters pertaining to maintaining high quality of sign language (SL) translation, contribute to dispelling misconceptions about SL and deaf people, and raise awareness and understanding of the deaf community heritage, this presentation discusses literature in Lithuanian Sign Language (LSL) and inherent metaphors that are created by using the phonological parameter –handshape, location, movement, palm orientation and nonmanual features. The study covered in this presentation is twofold, involving both the micro-level analysis of metaphors in terms of phonological parameters as a sub-lexical feature and the macro-level analysis of the poetic context. Cognitive theories underlie research of metaphors in sign language literature in a range of SL. The study follows this practice. The presentation covers the qualitative analysis of 34 pieces of LSL literature. The analysis employs ELAN software widely used in SL research. The target is to examine how specific types of each phonological parameter are used for the creation of metaphors in LSL literature and what metaphors are created. The results of the study show that LSL literature employs a range of metaphors created by using classifier signs and by modifying the established signs. The study also reveals that LSL literature tends to create reference metaphors indicating status and power. As the study shows, LSL poets metaphorically encode status by encoding another meaning in the same sign, which results in creating double metaphors. The metaphor of identity has been determined. Notably, the poetic context has revealed that the latter metaphor can also be identified as a metaphor for life. The study goes on to note that deaf poets create metaphors related to the importance of various phenomena significance of the lyrical subject. Notably, the study has allowed detecting locations, nonmanual features and etc., never mentioned in previous SL research as used for the creation of metaphors.

Keywords: Lithuanian sign language, sign language literature, sign language metaphor, metaphor at the phonological level, cognitive linguistics

Procedia PDF Downloads 104

Authors: Massimo Miccoli, Luca Marangoni, Alberto Aniello Scaringi, Alessandro Marceddu, Alessandro Amicone

Abstract:

The susceptibility of deep neural networks (DNNs) to adversarial manipulations is a recognized challenge within the computer vision domain. Adversarial examples, crafted by adding subtle yet malicious alterations to benign images, exploit this vulnerability. Various defense strategies have been proposed to safeguard DNNs against such attacks, stemming from diverse research hypotheses. Building upon prior work, our approach involves the utilization of autoencoder models. Autoencoders, a type of neural network, are trained to learn representations of training data and reconstruct inputs from these representations, typically minimizing reconstruction errors like mean squared error (MSE). Our autoencoder was trained on a dataset of benign examples; learning features specific to them. Consequently, when presented with significantly perturbed adversarial examples, the autoencoder exhibited high reconstruction errors. The architecture of the autoencoder was tailored to the dimensions of the images under evaluation. We considered various image sizes, constructing models differently for 256x256 and 512x512 images. Moreover, the choice of the computer vision model is crucial, as most adversarial attacks are designed with specific AI structures in mind. To mitigate this, we proposed a method to replace image-specific dimensions with a structure independent of both dimensions and neural network models, thereby enhancing robustness. Our multi-modal autoencoder reconstructs the spectral representation of images across the red-green-blue (RGB) color channels. To validate our approach, we conducted experiments using diverse datasets and subjected them to adversarial attacks using models such as ResNet50 and ViT_L_16 from the torch vision library. The autoencoder extracted features used in a classification model, resulting in an MSE (RGB) of 0.014, a classification accuracy of 97.33%, and a precision of 99%.

Keywords: adversarial attacks, malicious images detector, binary classifier, multimodal transformer autoencoder

Procedia PDF Downloads 38

Authors: Victor N. Anisimov, Igor E. Serov, Ksenia M. Kolkova, Natalia V. Galkina

Abstract:

The problem of memorability of the advertising content is closely connected with the key issues of neuromarketing. The memorability of the advertising content contributes to the marketing effectiveness of the promoted product. Significant directions of studying the phenomenon of memorability are the memorability of the brand (detected through the memorability of the logo) and the memorability of the product offer (detected through the memorization of dynamic audiovisual advertising content - commercial). The aim of this work is to reveal the predictors of memorization of static and dynamic audiovisual stimuli (logos and commercials). An important direction of the research was revealing differences in psychophysiological correlates of memorability between static and dynamic audiovisual stimuli. We assumed that static and dynamic images are perceived in different ways and may have a difference in the memorization process. Objective methods of recording psychophysiological parameters while watching static and dynamic audiovisual materials are well suited to achieve the aim. The electroencephalography (EEG) method was performed with the aim of identifying correlates of the memorability of various stimuli in the electrical activity of the cerebral cortex. All stimuli (in the groups of statics and dynamics separately) were divided into 2 groups – remembered and not remembered based on the results of the questioning method. The questionnaires were filled out by survey participants after viewing the stimuli not immediately, but after a time interval (for detecting stimuli recorded through long-term memorization). Using statistical method, we developed the classifier (statistical model) that predicts which group (remembered or not remembered) stimuli gets, based on psychophysiological perception. The result of the statistical model was compared with the results of the questionnaire. Conclusions: Predictors of the memorability of static and dynamic stimuli have been identified, which allows prediction of which stimuli will have a higher probability of remembering. Further developments of this study will be the creation of stimulus memory model with the possibility of recognizing the stimulus as previously seen or new. Thus, in the process of remembering the stimulus, it is planned to take into account the stimulus recognition factor, which is one of the most important tasks for neuromarketing.

Keywords: memory, commercials, neuromarketing, EEG, branding

Procedia PDF Downloads 226

Authors: Kishor Chandra Kandpal, Amit Kumar

Abstract:

The Indian Himalaya region harbours approximately 1748 plants of medicinal importance, and as per International Union for Conservation of Nature (IUCN), the 112 plant species among these are threatened and endangered. To ease the pressure on these plants, the government of India is encouraging its in-situ cultivation. The Saussurea costus, Valeriana jatamansi, and Picrorhiza kurroa have also been prioritized for large scale cultivation owing to their market demand, conservation value and medicinal properties. These species are found from 1000 m to 4000 m elevation ranges in the Indian Himalaya. Identification of these plants in the field requires taxonomic skills, which is one of the major bottleneck in the conservation and management of these plants. In recent years, Hyperspectral remote sensing techniques have been precisely used for the discrimination of plant species with the help of their unique spectral signatures. In this background, a spectral library of the above 03 medicinal plants was prepared by collecting the spectral data using a handheld spectroradiometer (325 to 1075 nm) from farmer’s fields of Himachal Pradesh and Uttarakhand states of Indian Himalaya. The Random forest (RF) model was implied on the spectral data for the classification of the medicinal plants. The 80:20 standard split ratio was followed for training and validation of the RF model, which resulted in training accuracy of 84.39 % (kappa coefficient = 0.72) and testing accuracy of 85.29 % (kappa coefficient = 0.77). This RF classifier has identified green (555 to 598 nm), red (605 nm), and near-infrared (725 to 840 nm) wavelength regions suitable for the discrimination of these species. The findings of this study have provided a technique for rapid and onsite identification of the above medicinal plants in the field. This will also be a key input for the classification of hyperspectral remote sensing images for mapping of these species in farmer’s field on a regional scale. This is a pioneer study in the Indian Himalaya region for medicinal plants in which the applicability of hyperspectral remote sensing has been explored.

Keywords: himalaya, hyperspectral remote sensing, machine learning; medicinal plants, random forests

Procedia PDF Downloads 172

Authors: Shishay Kidanu, Abdullah Alhaj

Abstract:

Sinkholes in the karst terrain of Greene County, Missouri, pose significant geohazards, imposing challenges on construction and infrastructure development, with potential threats to lives and property. To address these issues, understanding the influencing factors and modeling sinkhole susceptibility is crucial for effective mitigation through strategic changes in land use planning and practices. This study utilizes geographic information system (GIS) software to collect and process diverse data, including topographic, geologic, hydrogeologic, and anthropogenic information. Nine key sinkhole influencing factors, ranging from slope characteristics to proximity to geological structures, were carefully analyzed. The Frequency Ratio method establishes relationships between attribute classes of these factors and sinkhole events, deriving class weights to indicate their relative importance. Weighted integration of these factors is accomplished using the Analytic Hierarchy Process (AHP) and the Weighted Linear Combination (WLC) method in a GIS environment, resulting in a comprehensive sinkhole susceptibility index (SSI) model for the study area. Employing Jenk's natural break classifier method, the SSI values are categorized into five distinct sinkhole susceptibility zones: very low, low, moderate, high, and very high. Validation of the model, conducted through the Area Under Curve (AUC) and Sinkhole Density Index (SDI) methods, demonstrates a robust correlation with sinkhole inventory data. The prediction rate curve yields an AUC value of 74%, indicating a 74% validation accuracy. The SDI result further supports the success of the sinkhole susceptibility model. This model offers reliable predictions for the future distribution of sinkholes, providing valuable insights for planners and engineers in the formulation of development plans and land-use strategies. Its application extends to enhancing preparedness and minimizing the impact of sinkhole-related geohazards on both infrastructure and the community.

Keywords: sinkhole, GIS, analytical hierarchy process, frequency ratio, susceptibility, Missouri

Procedia PDF Downloads 36

Authors: Amanina Iymia Jeffree, Reena Thriumani, Mohammad Iqbal Omar, Ammar Zakaria, Yumi Zuhanis Has-Yun Hashim, Ali Yeon Md Shakaff

Abstract:

Metabolite profiling is a strategy to be approached in the pattern recognition method focused on three types of cancer cell line that driving the most to death specifically lung, breast, and colon cancer. The purpose of this study was to discriminate the VOCs pattern among cancerous and control group based on metabolite profiling. The sampling was executed utilizing the cell culture technique. All culture flasks were incubated till 72 hours and data collection started after 24 hours. Every running sample took 24 minutes to be completed accordingly. The comparative metabolite patterns were identified by the implementation of headspace-solid phase micro-extraction (HS-SPME) sampling coupled with gas chromatography-mass spectrometry (GCMS). The optimizations of the main experimental variables such as oven temperature and time were evaluated by response surface methodology (RSM) to get the optimal condition. Volatiles were acknowledged through the National Institute of Standards and Technology (NIST) mass spectral database and retention time libraries. To improve the reliability of significance, it is of crucial importance to eliminate background noise which data from 3rd minutes to 17th minutes were selected for statistical analysis. Targeted metabolites, of which were annotated as known compounds with the peak area greater than 0.5 percent were highlighted and subsequently treated statistically. Volatiles produced contain hundreds to thousands of compounds; therefore, it will be optimized by chemometric analysis, such as principal component analysis (PCA) as a preliminary analysis before subjected to a pattern classifier for identification of VOC samples. The volatile organic compound profiling has shown to be significantly distinguished among cancerous and control group based on metabolite profiling.

Keywords: in vitro cancer cell line, metabolite profiling, pattern recognition, volatile organic compounds

Procedia PDF Downloads 343

Authors: Michal Krecichwost, Zauzanna Miodonska, Pawel Badura

Abstract:

The diagnosis of sigmatism is mostly based on the observation of articulatory organs. It is, however, not always possible to precisely observe the vocal apparatus, in particular in the oral cavity of the patient. Speech processing can allow to objectify the therapy and simplify the verification of its progress. In the described study the methodology for classification of incorrectly pronounced phoneme [s] is proposed. The recordings come from adults. They were registered with the speech recorder at the sampling rate of 44.1 kHz and the resolution of 16 bit. The database of pathological and normative speech has been collected for the study including reference assessments provided by the speech therapy experts. Ten adult subjects were asked to simulate a certain type of stigmatism under the speech therapy expert supervision. In the recordings, the analyzed phone [s] was surrounded by vowels, viz: ASA, ESE, ISI, SPA, USU, YSY. Thirteen MFCC (mel-frequency cepstral coefficients) and RMS (root mean square) values are calculated within each frame being a part of the analyzed phoneme. Additionally, 3 fricative formants along with corresponding amplitudes are determined for the entire segment. In order to aggregate the information within the segment, the average value of each MFCC coefficient is calculated. All features of other types are aggregated by means of their 75th percentile. The proposed method of features aggregation reduces the size of the feature vector used in the classification. Binary SVM (support vector machine) classifier is employed at the phoneme recognition stage. The first group consists of pathological phones, while the other of the normative ones. The proposed feature vector yields classification sensitivity and specificity measures above 90% level in case of individual logo phones. The employment of a fricative formants-based information improves the sole-MFCC classification results average of 5 percentage points. The study shows that the employment of specific parameters for the selected phones improves the efficiency of pathology detection referred to the traditional methods of speech signal parameterization.

Keywords: computer-aided pronunciation evaluation, sibilants, sigmatism diagnosis, speech processing

Procedia PDF Downloads 256

Authors: Monica Gonzalez Machorro

Abstract:

Dementia is hard to diagnose because of the lack of early physical symptoms. Early dementia recognition is key to improving the living condition of patients. Speech technology is considered a valuable biomarker for this challenge. Recent works have utilized conventional acoustic features and machine learning methods to detect dementia in speech. BERT-like classifiers have reported the most promising performance. One constraint, nonetheless, is that these studies are either based on human transcripts or on transcripts produced by automatic speech recognition (ASR) systems. This research contribution is to explore a method that does not require transcriptions to detect early Alzheimer’s disease (AD) and mild cognitive impairment (MCI). This is achieved by fine-tuning a pre-trained ASR model for the downstream early AD and MCI tasks. To do so, a subset of the thoroughly studied Pitt Corpus is customized. The subset is balanced for class, age, and gender. Data processing also involves cropping the samples into 10-second segments. For comparison purposes, a baseline model is defined by training and testing a Random Forest with 20 extracted acoustic features using the librosa library implemented in Python. These are: zero-crossing rate, MFCCs, spectral bandwidth, spectral centroid, root mean square, and short-time Fourier transform. The baseline model achieved a 58% accuracy. To fine-tune HuBERT as a classifier, an average pooling strategy is employed to merge the 3D representations from audio into 2D representations, and a linear layer is added. The pre-trained model used is ‘hubert-large-ls960-ft’. Empirically, the number of epochs selected is 5, and the batch size defined is 1. Experiments show that our proposed method reaches a 69% balanced accuracy. This suggests that the linguistic and speech information encoded in the self-supervised ASR-based model is able to learn acoustic cues of AD and MCI.

Keywords: automatic speech recognition, early Alzheimer’s recognition, mild cognitive impairment, speech impairment

Procedia PDF Downloads 97

Authors: Lily Ingsrisawang, Tasanee Nacharoen

Abstract:

Introduction: The problems of unbalanced data sets generally appear in real world applications. Due to unequal class distribution, many research papers found that the performance of existing classifier tends to be biased towards the majority class. The k -nearest neighbors’ nonparametric discriminant analysis is one method that was proposed for classifying unbalanced classes with good performance. Hence, the methods of discriminant analysis are of interest to us in investigating misclassification error rates for class-imbalanced data of three diabetes risk groups. Objective: The purpose of this study was to compare the classification performance between parametric discriminant analysis and nonparametric discriminant analysis in a three-class classification application of class-imbalanced data of diabetes risk groups. Methods: Data from a healthy project for 599 staffs in a government hospital in Bangkok were obtained for the classification problem. The staffs were diagnosed into one of three diabetes risk groups: non-risk (90%), risk (5%), and diabetic (5%). The original data along with the variables; diabetes risk group, age, gender, cholesterol, and BMI was analyzed and bootstrapped up to 50 and 100 samples, 599 observations per sample, for additional estimation of misclassification error rate. Each data set was explored for the departure of multivariate normality and the equality of covariance matrices of the three risk groups. Both the original data and the bootstrap samples show non-normality and unequal covariance matrices. The parametric linear discriminant function, quadratic discriminant function, and the nonparametric k-nearest neighbors’ discriminant function were performed over 50 and 100 bootstrap samples and applied to the original data. In finding the optimal classification rule, the choices of prior probabilities were set up for both equal proportions (0.33: 0.33: 0.33) and unequal proportions with three choices of (0.90:0.05:0.05), (0.80: 0.10: 0.10) or (0.70, 0.15, 0.15). Results: The results from 50 and 100 bootstrap samples indicated that the k-nearest neighbors approach when k = 3 or k = 4 and the prior probabilities of {non-risk:risk:diabetic} as {0.90:0.05:0.05} or {0.80:0.10:0.10} gave the smallest error rate of misclassification. Conclusion: The k-nearest neighbors approach would be suggested for classifying a three-class-imbalanced data of diabetes risk groups.

Keywords: error rate, bootstrap, diabetes risk groups, k-nearest neighbors

Procedia PDF Downloads 408

Authors: Rosa Figueroa, Christopher Flores

Abstract:

Text categorization is the problem of assigning a new document to a set of predetermined categories, on the basis of a training set of free-text data that contains documents whose category membership is known. To train a classification model, it is necessary to extract characteristics in the form of tokens that facilitate the learning and classification process. In text categorization, the feature extraction process involves the use of word sequences also known as N-grams. In general, it is expected that documents belonging to the same category share similar features. The Smith-Waterman (SW) algorithm is a dynamic programming algorithm that performs a local sequence alignment in order to determine similar regions between two strings or protein sequences. This work explores the use of SW algorithm as an alternative to feature extraction in text categorization. The dataset used for this purpose, contains 2,610 annotated documents with the classes Obese/Non-Obese. This dataset was represented in a matrix form using the Bag of Word approach. The score selected to represent the occurrence of the tokens in each document was the term frequency-inverse document frequency (TF-IDF). In order to extract features for classification, four experiments were conducted: the first experiment used SW to extract features, the second one used unigrams (single word), the third one used bigrams (two word sequence) and the last experiment used a combination of unigrams and bigrams to extract features for classification. To test the effectiveness of the extracted feature set for the four experiments, a Support Vector Machine (SVM) classifier was tuned using 20% of the dataset. The remaining 80% of the dataset together with 5-Fold Cross Validation were used to evaluate and compare the performance of the four experiments of feature extraction. Results from the tuning process suggest that SW performs better than the N-gram based feature extraction. These results were confirmed by using the remaining 80% of the dataset, where SW performed the best (accuracy = 97.10%, weighted average F-measure = 97.07%). The second best was obtained by the combination of unigrams-bigrams (accuracy = 96.04, weighted average F-measure = 95.97) closely followed by the bigrams (accuracy = 94.56%, weighted average F-measure = 94.46%) and finally unigrams (accuracy = 92.96%, weighted average F-measure = 92.90%).

Keywords: comorbidities, machine learning, obesity, Smith-Waterman algorithm

Procedia PDF Downloads 269

Authors: Mohammad Tarek Al Muallim, Ozhan Duzenli, Ceyhun Ilguy

Abstract:

Harbor protection systems are so important. The need for automatic protection systems has increased over the last years. Diver detection active sonar has great significance. It used to detect underwater threats such as divers and autonomous underwater vehicle. To automatically detect such threats the sonar image is processed by algorithms. These algorithms used to detect, track and classify of underwater objects. In this work, divers tracking and classification algorithm is improved be proposing a robust wake detection method. To detect objects the sonar images is normalized then segmented based on fixed threshold. Next, the centroids of the segments are found and clustered based on distance metric. Then to track the objects linear Kalman filter is applied. To reduce effect of noise and creation of false tracks, the Kalman tracker is fine tuned. The tuning is done based on our active sonar specifications. After the tracks are initialed and updated they are subjected to a filtering stage to eliminate the noisy and unstable tracks. Also to eliminate object with a speed out of the diver speed range such as buoys and fast boats. Afterwards the result tracks are subjected to a classification stage to deiced the type of the object been tracked. Here the classification stage is to deice wither if the tracked object is an open circuit diver or a close circuit diver. At the classification stage, a small area around the object is extracted and a novel wake detection method is applied. The morphological features of the object with his wake is extracted. We used support vector machine to find the best classifier. The sonar training images and the test images are collected by ARMELSAN Defense Technologies Company using the portable diver detection sonar ARAS-2023. After applying the algorithm to the test sonar data, we get fine and stable tracks of the divers. The total classification accuracy achieved with the diver type is 97%.

Keywords: harbor protection, diver detection, active sonar, wake detection, diver classification

Procedia PDF Downloads 207

Authors: Christian Neunzig, Simon Fahle, Jürgen Schulz, Matthias Möller, Bernd Kuhlenkötter

Abstract:

Manufacturing companies are facing global competition and enormous cost pressure. The use of machine learning applications can help reduce production costs and create added value. Predictive quality enables the securing of product quality through data-supported predictions using machine learning models as a basis for decisions on test results. Furthermore, machine learning methods are able to process large amounts of data, deal with unfavourable row-column ratios and detect dependencies between the covariates and the given target as well as assess the multidimensional influence of all input variables on the target. Real production data are often subject to highly fluctuating boundary conditions and unbalanced data sets. Changes in production data manifest themselves in trends, systematic shifts, and seasonal effects. Thus, Machine learning applications require intensive pre-processing and feature selection. Data preprocessing includes rule-based data cleaning, the application of dimensionality reduction techniques, and the identification of comparable data subsets. Within the used real data set of Bosch hydraulic valves, the comparability of the same production conditions in the production of hydraulic valves within certain time periods can be identified by applying the concept drift method. Furthermore, a classification model is developed to evaluate the feature importance in different subsets within the identified time periods. By selecting comparable and stable features, the number of features used can be significantly reduced without a strong decrease in predictive power. The use of cross-process production data along the value chain of hydraulic valves is a promising approach to predict the quality characteristics of workpieces. In this research, the ada boosting classifier is used to predict the leakage of hydraulic valves based on geometric gauge blocks from machining, mating data from the assembly, and hydraulic measurement data from end-of-line testing. In addition, the most suitable methods are selected and accurate quality predictions are achieved.

Keywords: classification, achine learning, predictive quality, feature selection

Procedia PDF Downloads 134

Authors: H. Jung, N. Kim, B. Kang, J. Choe

Abstract:

History matching is a crucial procedure for predicting reservoir performances and making future decisions. However, it is difficult due to uncertainties of initial reservoir models. Therefore, it is important to have reliable initial models for successful history matching of highly heterogeneous reservoirs such as channel reservoirs. In this paper, we proposed a novel scheme for regenerating geological models using support vector machine (SVM) and principal component analysis (PCA). First, we perform PCA for figuring out main geological characteristics of models. Through the procedure, permeability values of each model are transformed to new parameters by principal components, which have eigenvalues of large magnitude. Secondly, the parameters are projected into two-dimensional plane by multi-dimensional scaling (MDS) based on Euclidean distances. Finally, we train an SVM classifier using 20% models which show the most similar or dissimilar well oil production rates (WOPR) with the true values (10% for each). Then, the other 80% models are classified by trained SVM. We select models on side of low WOPR errors. One hundred channel reservoir models are initially generated by single normal equation simulation. By repeating the classification process, we can select models which have similar geological trend with the true reservoir model. The average field of the selected models is utilized as a probability map for regeneration. Newly generated models can preserve correct channel features and exclude wrong geological properties maintaining suitable uncertainty ranges. History matching with the initial models cannot provide trustworthy results. It fails to find out correct geological features of the true model. However, history matching with the regenerated ensemble offers reliable characterization results by figuring out proper channel trend. Furthermore, it gives dependable prediction of future performances with reduced uncertainties. We propose a novel classification scheme which integrates PCA, MDS, and SVM for regenerating reservoir models. The scheme can easily sort out reliable models which have similar channel trend with the reference in lowered dimension space.

Keywords: history matching, principal component analysis, reservoir modelling, support vector machine

Procedia PDF Downloads 133

Authors: Manutchanok Jongprasithporn, Rawiphorn Srivilai, Paweena Pongsopha

Abstract:

Fall is the most serious accident leading to increased unintentional injuries and mortality. Falls are not only the cause of suffering and functional impairments to the individuals, but also the cause of increasing medical cost and days away from work. The early detection of falls could be an advantage to reduce fall-related injuries and consequences of falls. Smartphones, embedded accelerometer, have become a common device in everyday life due to decreasing technology cost. This paper explores a physical activity monitoring and fall detection application in smartphones which is a non-invasive biomedical device to determine physical activities and fall event. The combination of application and sensors could perform as a biomedical sensor to monitor physical activities and recognize a fall. We have chosen Android-based smartphone in this study since android operating system is an open-source and no cost. Moreover, android phone users become a majority of Thai’s smartphone users. We developed Thai 3 Axis (TH3AX) as a physical activities and fall detection application which included command, manual, results in Thai language. The smartphone was attached to right hip of 10 young, healthy adult subjects (5 males, 5 females; aged< 35y) to collect accelerometer and gyroscope data during performing physical activities (e.g., walking, running, sitting, and lying down) and falling to determine threshold for each activity. Dependent variables are including accelerometer data (acceleration, peak acceleration, average resultant acceleration, and time between peak acceleration). A repeated measures ANOVA was performed to test whether there are any differences between DVs’ means. Statistical analyses were considered significant at p<0.05. After finding threshold, the results were used as training data for a predictive model of activity recognition. In the future, accuracies of activity recognition will be performed to assess the overall performance of the classifier. Moreover, to help improve the quality of life, our system will be implemented with patients and elderly people who need intensive care in hospitals and nursing homes in Thailand.

Keywords: activity recognition, accelerometer, fall, gyroscope, smartphone

Procedia PDF Downloads 661

Authors: Kristina Sakalauskiene, Renaldas Raisutis, Gintare Linkeviciute, Skaidra Valiukeviciene

Abstract:

Cutaneous melanoma is a melanocytic skin tumour, which has a very poor prognosis while is highly resistant to treatment and tends to metastasize. Thickness of melanoma is one of the most important biomarker for stage of disease, prognosis and surgery planning. In this study, we hypothesized that the automatic analysis of spectrophotometric images and high-frequency ultrasonic 2D data can improve differential diagnosis of cutaneous melanoma and provide additional information about tumour penetration depth. This paper presents the novel complex automatic system for non-invasive melanocytic skin tumour differential diagnosis and penetration depth evaluation. The system is composed of region of interest segmentation in spectrophotometric images and high-frequency ultrasound data, quantitative parameter evaluation, informative feature extraction and classification with linear regression classifier. The segmentation of melanocytic skin tumour region in ultrasound image is based on parametric integrated backscattering coefficient calculation. The segmentation of optical image is based on Otsu thresholding. In total 29 quantitative tissue characterization parameters were evaluated by using ultrasound data (11 acoustical, 4 shape and 15 textural parameters) and 55 quantitative features of dermatoscopic and spectrophotometric images (using total melanin, dermal melanin, blood and collagen SIAgraphs acquired using spectrophotometric imaging device SIAscope). In total 102 melanocytic skin lesions (including 43 cutaneous melanomas) were examined by using SIAscope and ultrasound system with 22 MHz center frequency single element transducer. The diagnosis and Breslow thickness (pT) of each MST were evaluated during routine histological examination after excision and used as a reference. The results of this study have shown that automatic analysis of spectrophotometric and high frequency ultrasound data can improve non-invasive classification accuracy of early-stage cutaneous melanoma and provide supplementary information about tumour penetration depth.

Keywords: cutaneous melanoma, differential diagnosis, high-frequency ultrasound, melanocytic skin tumours, spectrophotometric imaging

Procedia PDF Downloads 244

Authors: Nassima Noufail, Sara Bouhali

Abstract:

In this work, we develop a semi-supervised solution for the purpose of action detection in videos and propose an efficient algorithm for video segmentation. The approach is divided into video segmentation, feature extraction, and classification. In the first part, a video is segmented into clips, and we used the K-means algorithm for this segmentation; our goal is to find groups based on similarity in the video. The application of k-means clustering into all the frames is time-consuming; therefore, we started by the identification of transition frames where the scene in the video changes significantly, and then we applied K-means clustering into these transition frames. We used two image filters, the gaussian filter and the Laplacian of Gaussian. Each filter extracts a set of features from the frames. The Gaussian filter blurs the image and omits the higher frequencies, and the Laplacian of gaussian detects regions of rapid intensity changes; we then used this vector of filter responses as an input to our k-means algorithm. The output is a set of cluster centers. Each video frame pixel is then mapped to the nearest cluster center and painted with a corresponding color to form a visual map. The resulting visual map had similar pixels grouped. We then computed a cluster score indicating how clusters are near each other and plotted a signal representing frame number vs. clustering score. Our hypothesis was that the evolution of the signal would not change if semantically related events were happening in the scene. We marked the breakpoints at which the root mean square level of the signal changes significantly, and each breakpoint is an indication of the beginning of a new video segment. In the second part, for each segment from part 1, we randomly selected a 16-frame clip, then we extracted spatiotemporal features using convolutional 3D network C3D for every 16 frames using a pre-trained model. The C3D final output is a 512-feature vector dimension; hence we used principal component analysis (PCA) for dimensionality reduction. The final part is the classification. The C3D feature vectors are used as input to a multi-class linear support vector machine (SVM) for the training model, and we used a multi-classifier to detect the action. We evaluated our experiment on the UCF101 dataset, which consists of 101 human action categories, and we achieved an accuracy that outperforms the state of art by 1.2%.

Keywords: video segmentation, action detection, classification, Kmeans, C3D

Procedia PDF Downloads 48

Authors: Zeinab Jafari A., Ali Sharifnezhad B., Mohammad Razi C., Mohammad Haghpanahi D., Arash Maghsoudi

Abstract:

Background and Objective: Sports-related rupture of the anterior cruciate ligament (ACL) and following injuries have been associated with various disorders, such as long-lasting changes in muscle activation patterns in athletes, which might last after ACL reconstruction (ACLR). The rupture of the ACL might result in abnormal patterns of movement execution, extending the treatment period and delaying athletes’ return to sports (RTS). As ACL injury is especially prevalent among athletes, the lengthy treatment process and athletes’ absence from sports are of great concern to athletes and coaches. Thus, estimating safe time of RTS is of crucial importance. Therefore, using a deep neural network (DNN) to classify the health levels of ACL in injured athletes, this study aimed to estimate the safe time for athletes to return to competitions. Methods: Ten athletes with ACLR and fourteen healthy controls participated in this study. Three health levels of ACL were defined: healthy, six-month post-ACLR surgery and nine-month post-ACLR surgery. Athletes with ACLR were tested six and nine months after the ACLR surgery. During the course of this study, surface electromyography (sEMG) signals were recorded from five knee muscles, namely Rectus Femoris (RF), Vastus Lateralis (VL), Vastus Medialis (VM), Biceps Femoris (BF), Semitendinosus (ST), during single-leg drop landing (SLDL) and forward hopping (SLFH) tasks. The Pseudo-Wigner-Ville distribution (PWVD) was used to produce three-dimensional (3-D) images of the energy distribution patterns of sEMG signals. Then, these 3-D images were converted to two-dimensional (2-D) images implementing the heat mapping technique, which were then fed to a deep convolutional neural network (DCNN). Results: In this study, we estimated the safe time of RTS by designing a DCNN classifier with an accuracy of 90 %, which could classify ACL into three health levels. Discussion: The findings of this study demonstrate the potential of the DCNN classification technique using sEMG signals in estimating RTS time, which will assist in evaluating the recovery process of ACLR in athletes.

Keywords: anterior cruciate ligament reconstruction, return to sports, surface electromyography, deep convolutional neural network

Procedia PDF Downloads 46

Authors: Maxence Queyrel, Edi Prifti, Alexandre Templier, Jean-Daniel Zucker

Abstract:

Analysis of the human microbiome using metagenomic sequencing data has demonstrated high ability in discriminating various human diseases. Raw metagenomic sequencing data require multiple complex and computationally heavy bioinformatics steps prior to data analysis. Such data contain millions of short sequences read from the fragmented DNA sequences and stored as fastq files. Conventional processing pipelines consist in multiple steps including quality control, filtering, alignment of sequences against genomic catalogs (genes, species, taxonomic levels, functional pathways, etc.). These pipelines are complex to use, time consuming and rely on a large number of parameters that often provide variability and impact the estimation of the microbiome elements. Training Deep Neural Networks directly from raw sequencing data is a promising approach to bypass some of the challenges associated with mainstream bioinformatics pipelines. Most of these methods use the concept of word and sentence embeddings that create a meaningful and numerical representation of DNA sequences, while extracting features and reducing the dimensionality of the data. In this paper we present an end-to-end approach that classifies patients into disease groups directly from raw metagenomic reads: metagenome2vec. This approach is composed of four steps (i) generating a vocabulary of k-mers and learning their numerical embeddings; (ii) learning DNA sequence (read) embeddings; (iii) identifying the genome from which the sequence is most likely to come and (iv) training a multiple instance learning classifier which predicts the phenotype based on the vector representation of the raw data. An attention mechanism is applied in the network so that the model can be interpreted, assigning a weight to the influence of the prediction for each genome. Using two public real-life data-sets as well a simulated one, we demonstrated that this original approach reaches high performance, comparable with the state-of-the-art methods applied directly on processed data though mainstream bioinformatics workflows. These results are encouraging for this proof of concept work. We believe that with further dedication, the DNN models have the potential to surpass mainstream bioinformatics workflows in disease classification tasks.

Keywords: deep learning, disease prediction, end-to-end machine learning, metagenomics, multiple instance learning, precision medicine

Procedia PDF Downloads 96

Authors: Alvaro Figueira, Bruno Cabral

Abstract:

Nowadays, students are used to be assessed through an online platform. Educators have stepped up from a period in which they endured the transition from paper to digital. The use of a diversified set of question types that range from quizzes to open questions is currently common in most university courses. In many courses, today, the evaluation methodology also fosters the students’ online participation in forums, the download, and upload of modified files, or even the participation in group activities. At the same time, new pedagogy theories that promote the active participation of students in the learning process, and the systematic use of problem-based learning, are being adopted using an eLearning system for that purpose. However, although there can be a lot of feedback from these activities to student’s, usually it is restricted to the assessments of online well-defined tasks. In this article, we propose an automatic system that informs students of abnormal deviations of a 'correct' learning path in the course. Our approach is based on the fact that by obtaining this information earlier in the semester, may provide students and educators an opportunity to resolve an eventual problem regarding the student’s current online actions towards the course. Our goal is to prevent situations that have a significant probability to lead to a poor grade and, eventually, to failing. In the major learning management systems (LMS) currently available, the interaction between the students and the system itself is registered in log files in the form of registers that mark beginning of actions performed by the user. Our proposed system uses that logged information to derive new one: the time each student spends on each activity, the time and order of the resources used by the student and, finally, the online resource usage pattern. Then, using the grades assigned to the students in previous years, we built a learning dataset that is used to feed a machine learning meta classifier. The produced classification model is then used to predict the grades a learning path is heading to, in the current year. Not only this approach serves the teacher, but also the student to receive automatic feedback on her current situation, having past years as a perspective. Our system can be applied to online courses that integrate the use of an online platform that stores user actions in a log file, and that has access to other student’s evaluations. The system is based on a data mining process on the log files and on a self-feedback machine learning algorithm that works paired with the Moodle LMS.

Keywords: data mining, e-learning, grade prediction, machine learning, student learning path

Procedia PDF Downloads 97

Authors: Zunaira Zaman, David Bohannon

Abstract:

This study explores the application of machine learning (ML) for detecting security vulnerabilities in source code. The research aims to assist organizations with large application portfolios and limited security testing capabilities in prioritizing security activities. ML-based approaches offer benefits such as increased confidence scores, false positives and negatives tuning, and automated feedback. The initial approach using natural language processing techniques to extract features achieved 86% accuracy during the training phase but suffered from overfitting and performed poorly on unseen datasets during testing. To address these issues, the study proposes using the abstract syntax tree (AST) for Java and C++ codebases to capture code semantics and structure and generate path-context representations for each function. The Code2Vec model architecture is used to learn distributed representations of source code snippets for training a machine-learning classifier for vulnerability prediction. The study evaluates the performance of the proposed methodology using two datasets and compares the results with existing approaches. The Devign dataset yielded 60% accuracy in predicting vulnerable code snippets and helped resist overfitting, while the Juliet Test Suite predicted specific vulnerabilities such as OS-Command Injection, Cryptographic, and Cross-Site Scripting vulnerabilities. The Code2Vec model achieved 75% accuracy and a 98% recall rate in predicting OS-Command Injection vulnerabilities. The study concludes that even partial AST representations of source code can be useful for vulnerability prediction. The approach has the potential for automated intelligent analysis of source code, including vulnerability prediction on unseen source code. State-of-the-art models using natural language processing techniques and CNN models with ensemble modelling techniques did not generalize well on unseen data and faced overfitting issues. However, predicting vulnerabilities in source code using machine learning poses challenges such as high dimensionality and complexity of source code, imbalanced datasets, and identifying specific types of vulnerabilities. Future work will address these challenges and expand the scope of the research.

Keywords: code embeddings, neural networks, natural language processing, OS command injection, software security, code properties

Procedia PDF Downloads 74

Authors: Arindam Chaudhuri

Abstract:

Classification of data has been actively used for most effective and efficient means of conveying knowledge and information to users. The prima face has always been upon techniques for extracting useful knowledge from data such that returns are maximized. With emergence of huge datasets the existing classification techniques often fail to produce desirable results. The challenge lies in analyzing and understanding characteristics of massive data sets by retrieving useful geometric and statistical patterns. We propose a supervised parallel fuzzy rough support vector machine (PFRSVM) for data classification in cloud environment. The classification is performed by PFRSVM using hyperbolic tangent kernel. The fuzzy rough set model takes care of sensitiveness of noisy samples and handles impreciseness in training samples bringing robustness to results. The membership function is function of center and radius of each class in feature space and is represented with kernel. It plays an important role towards sampling the decision surface. The success of PFRSVM is governed by choosing appropriate parameter values. The training samples are either linear or nonlinear separable. The different input points make unique contributions to decision surface. The algorithm is parallelized with a view to reduce training times. The system is built on support vector machine library using Hadoop implementation of MapReduce. The algorithm is tested on large data sets to check its feasibility and convergence. The performance of classifier is also assessed in terms of number of support vectors. The challenges encountered towards implementing big data classification in machine learning frameworks are also discussed. The experiments are done on the cloud environment available at University of Technology and Management, India. The results are illustrated for Gaussian RBF and Bayesian kernels. The effect of variability in prediction and generalization of PFRSVM is examined with respect to values of parameter C. It effectively resolves outliers’ effects, imbalance and overlapping class problems, normalizes to unseen data and relaxes dependency between features and labels. The average classification accuracy for PFRSVM is better than other classifiers for both Gaussian RBF and Bayesian kernels. The experimental results on both synthetic and real data sets clearly demonstrate the superiority of the proposed technique.

Keywords: FRSVM, Hadoop, MapReduce, PFRSVM

Procedia PDF Downloads 460

Authors: Dangut Maren David, Skaf Zakwan

Abstract:

Adequate monitoring of vehicle component in other to obtain high uptime is the goal of predictive maintenance, the major challenge faced by businesses in industries is the significant cost associated with a delay in service delivery due to system downtime. Most of those businesses are interested in predicting those problems and proactively prevent them in advance before it occurs, which is the core advantage of Prognostic Health Management (PHM) application. The recent emergence of industry 4.0 or industrial internet of things (IIoT) has led to the need for monitoring systems activities and enhancing system-to-system or component-to- component interactions, this has resulted to a large generation of data known as big data. Analysis of big data represents an increasingly important, however, due to complexity inherently in the dataset such as imbalance classification problems, it becomes extremely difficult to build a model with accurate high precision. Data-driven predictive modeling for condition-based maintenance (CBM) has recently drowned research interest with growing attention to both academics and industries. The large data generated from industrial process inherently comes with a different degree of complexity which posed a challenge for analytics. Thus, imbalance classification problem exists perversely in industrial datasets which can affect the performance of learning algorithms yielding to poor classifier accuracy in model development. Misclassification of faults can result in unplanned breakdown leading economic loss. In this paper, an advanced approach for handling imbalance classification problem is proposed and then a prognostic model for predicting aircraft component replacement is developed to predict component replacement in advanced by exploring aircraft historical data, the approached is based on hybrid ensemble-based method which improves the prediction of the minority class during learning, we also investigate the impact of our approach on multiclass imbalance problem. We validate the feasibility and effectiveness in terms of the performance of our approach using real-world aircraft operation and maintenance datasets, which spans over 7 years. Our approach shows better performance compared to other similar approaches. We also validate our approach strength for handling multiclass imbalanced dataset, our results also show good performance compared to other based classifiers.

Keywords: prognostics, data-driven, imbalance classification, deep learning

Procedia PDF Downloads 149

Authors: Mohsen Bahrami, Banafsheh Nikmehr, Yueqiang Song, Anuradha Koduru, Ayse K. Vuruskan, Hongkun Lu, Tamer M. Yalcinkaya

Abstract:

In recent years, we have witnessed an explosion of studies aimed at using a combination of artificial intelligence (AI) and time-lapse imaging data on embryos to improve IVF outcomes. However, despite promising results, no study has used a matched cohort of transferred embryos which only differ in pregnancy outcome, i.e., embryos from a single clinic which are similar in parameters, such as: morphokinetic condition, patient age, and overall clinic and lab performance. Here, we used time-lapse data on embryos with known pregnancy outcomes to see if the rich spatiotemporal information embedded in this data would allow the prediction of the pregnancy outcome regardless of such critical parameters. Methodology—We did a retrospective analysis of time-lapse data from our IVF clinic utilizing Embryoscope 100% of the time for embryo culture to blastocyst stage with known clinical outcomes, including live birth vs nonpregnant (embryos with spontaneous abortion outcomes were excluded). We used time-lapse data from 200 elective single transfer embryos randomly selected from January 2019 to June 2021. Our sample included 100 embryos in each group with no significant difference in patient age (P=0.9550) and morphokinetic scores (P=0.4032). Data from all patients were combined to make a 4th order tensor, and feature extraction were subsequently carried out by a tensor decomposition methodology. The features were then used in a machine learning classifier to classify the two groups. Major Findings—The performance of the model was evaluated using 100 random subsampling cross validation (train (80%) - test (20%)). The prediction accuracy, averaged across 100 permutations, exceeded 80%. We also did a random grouping analysis, in which labels (live birth, nonpregnant) were randomly assigned to embryos, which yielded 50% accuracy. Conclusion—The high accuracy in the main analysis and the low accuracy in random grouping analysis suggest a consistent spatiotemporal pattern which is associated with pregnancy outcomes, regardless of patient age and embryo morphokinetic condition, and beyond already known parameters, such as: early cleavage or early blastulation. Despite small samples size, this ongoing analysis is the first to show the potential of AI methods in capturing the complex morphokinetic changes embedded in embryo time-lapse data, which contribute to successful pregnancy outcomes, regardless of already known parameters. The results on a larger sample size with complementary analysis on prediction of other key outcomes, such as: euploidy and aneuploidy of embryos will be presented at the meeting.

Keywords: IVF, embryo, machine learning, time-lapse imaging data

Procedia PDF Downloads 64