Search results for: feature noise
1867 Enhanced Multi-Scale Feature Extraction Using a DCNN by Proposing Dynamic Soft Margin SoftMax for Face Emotion Detection
Authors: Armin Nabaei, M. Omair Ahmad, M. N. S. Swamy
Abstract:
Many facial expression and emotion recognition methods in the traditional approaches of using LDA, PCA, and EBGM have been proposed. In recent years deep learning models have provided a unique platform addressing by automatically extracting the features for the detection of facial expression and emotions. However, deep networks require large training datasets to extract automatic features effectively. In this work, we propose an efficient emotion detection algorithm using face images when only small datasets are available for training. We design a deep network whose feature extraction capability is enhanced by utilizing several parallel modules between the input and output of the network, each focusing on the extraction of different types of coarse features with fined grained details to break the symmetry of produced information. In fact, we leverage long range dependencies, which is one of the main drawback of CNNs. We develop this work by introducing a Dynamic Soft-Margin SoftMax.The conventional SoftMax suffers from reaching to gold labels very soon, which take the model to over-fitting. Because it’s not able to determine adequately discriminant feature vectors for some variant class labels. We reduced the risk of over-fitting by using a dynamic shape of input tensor instead of static in SoftMax layer with specifying a desired Soft- Margin. In fact, it acts as a controller to how hard the model should work to push dissimilar embedding vectors apart. For the proposed Categorical Loss, by the objective of compacting the same class labels and separating different class labels in the normalized log domain.We select penalty for those predictions with high divergence from ground-truth labels.So, we shorten correct feature vectors and enlarge false prediction tensors, it means we assign more weights for those classes with conjunction to each other (namely, “hard labels to learn”). By doing this work, we constrain the model to generate more discriminate feature vectors for variant class labels. Finally, for the proposed optimizer, our focus is on solving weak convergence of Adam optimizer for a non-convex problem. Our noteworthy optimizer is working by an alternative updating gradient procedure with an exponential weighted moving average function for faster convergence and exploiting a weight decay method to help drastically reducing the learning rate near optima to reach the dominant local minimum. We demonstrate the superiority of our proposed work by surpassing the first rank of three widely used Facial Expression Recognition datasets with 93.30% on FER-2013, and 16% improvement compare to the first rank after 10 years, reaching to 90.73% on RAF-DB, and 100% k-fold average accuracy for CK+ dataset, and shown to provide a top performance to that provided by other networks, which require much larger training datasets.Keywords: computer vision, facial expression recognition, machine learning, algorithms, depp learning, neural networks
Procedia PDF Downloads 641866 Robust Heart Rate Estimation from Multiple Cardiovascular and Non-Cardiovascular Physiological Signals Using Signal Quality Indices and Kalman Filter
Authors: Shalini Rankawat, Mansi Rankawat, Rahul Dubey, Mazad Zaveri
Abstract:
Physiological signals such as electrocardiogram (ECG) and arterial blood pressure (ABP) in the intensive care unit (ICU) are often seriously corrupted by noise, artifacts, and missing data, which lead to errors in the estimation of heart rate (HR) and incidences of false alarm from ICU monitors. Clinical support in ICU requires most reliable heart rate estimation. Cardiac activity, because of its relatively high electrical energy, may introduce artifacts in Electroencephalogram (EEG), Electrooculogram (EOG), and Electromyogram (EMG) recordings. This paper presents a robust heart rate estimation method by detection of R-peaks of ECG artifacts in EEG, EMG & EOG signals, using energy-based function and a novel Signal Quality Index (SQI) assessment technique. SQIs of physiological signals (EEG, EMG, & EOG) were obtained by correlation of nonlinear energy operator (teager energy) of these signals with either ECG or ABP signal. HR is estimated from ECG, ABP, EEG, EMG, and EOG signals from separate Kalman filter based upon individual SQIs. Data fusion of each HR estimate was then performed by weighing each estimate by the Kalman filters’ SQI modified innovations. The fused signal HR estimate is more accurate and robust than any of the individual HR estimate. This method was evaluated on MIMIC II data base of PhysioNet from bedside monitors of ICU patients. The method provides an accurate HR estimate even in the presence of noise and artifacts.Keywords: ECG, ABP, EEG, EMG, EOG, ECG artifacts, Teager-Kaiser energy, heart rate, signal quality index, Kalman filter, data fusion
Procedia PDF Downloads 6891865 An Advanced Automated Brain Tumor Diagnostics Approach
Authors: Berkan Ural, Arif Eser, Sinan Apaydin
Abstract:
Medical image processing is generally become a challenging task nowadays. Indeed, processing of brain MRI images is one of the difficult parts of this area. This study proposes a hybrid well-defined approach which is consisted from tumor detection, extraction and analyzing steps. This approach is mainly consisted from a computer aided diagnostics system for identifying and detecting the tumor formation in any region of the brain and this system is commonly used for early prediction of brain tumor using advanced image processing and probabilistic neural network methods, respectively. For this approach, generally, some advanced noise removal functions, image processing methods such as automatic segmentation and morphological operations are used to detect the brain tumor boundaries and to obtain the important feature parameters of the tumor region. All stages of the approach are done specifically with using MATLAB software. Generally, for this approach, firstly tumor is successfully detected and the tumor area is contoured with a specific colored circle by the computer aided diagnostics program. Then, the tumor is segmented and some morphological processes are achieved to increase the visibility of the tumor area. Moreover, while this process continues, the tumor area and important shape based features are also calculated. Finally, with using the probabilistic neural network method and with using some advanced classification steps, tumor area and the type of the tumor are clearly obtained. Also, the future aim of this study is to detect the severity of lesions through classes of brain tumor which is achieved through advanced multi classification and neural network stages and creating a user friendly environment using GUI in MATLAB. In the experimental part of the study, generally, 100 images are used to train the diagnostics system and 100 out of sample images are also used to test and to check the whole results. The preliminary results demonstrate the high classification accuracy for the neural network structure. Finally, according to the results, this situation also motivates us to extend this framework to detect and localize the tumors in the other organs.Keywords: image processing algorithms, magnetic resonance imaging, neural network, pattern recognition
Procedia PDF Downloads 4101864 Predicting Subsurface Abnormalities Growth Using Physics-Informed Neural Networks
Authors: Mehrdad Shafiei Dizaji, Hoda Azari
Abstract:
The research explores the pioneering integration of Physics-Informed Neural Networks (PINNs) into the domain of Ground-Penetrating Radar (GPR) data prediction, akin to advancements in medical imaging for tracking tumor progression in the human body. This research presents a detailed development framework for a specialized PINN model proficient at interpreting and forecasting GPR data, much like how medical imaging models predict tumor behavior. By harnessing the synergy between deep learning algorithms and the physical laws governing subsurface structures—or, in medical terms, human tissues—the model effectively embeds the physics of electromagnetic wave propagation into its architecture. This ensures that predictions not only align with fundamental physical principles but also mirror the precision needed in medical diagnostics for detecting and monitoring tumors. The suggested deep learning structure comprises three components: a CNN, a spatial feature channel attention (SFCA) mechanism, and ConvLSTM, along with temporal feature frame attention (TFFA) modules. The attention mechanism computes channel attention and temporal attention weights using self-adaptation, thereby fine-tuning the visual and temporal feature responses to extract the most pertinent and significant visual and temporal features. By integrating physics directly into the neural network, our model has shown enhanced accuracy in forecasting GPR data. This improvement is vital for conducting effective assessments of bridge deck conditions and other evaluations related to civil infrastructure. The use of Physics-Informed Neural Networks (PINNs) has demonstrated the potential to transform the field of Non-Destructive Evaluation (NDE) by enhancing the precision of infrastructure deterioration predictions. Moreover, it offers a deeper insight into the fundamental mechanisms of deterioration, viewed through the prism of physics-based models.Keywords: physics-informed neural networks, deep learning, ground-penetrating radar (GPR), NDE, ConvLSTM, physics, data driven
Procedia PDF Downloads 191863 Design and Performance Improvement of Three-Dimensional Optical Code Division Multiple Access Networks with NAND Detection Technique
Authors: Satyasen Panda, Urmila Bhanja
Abstract:
In this paper, we have presented and analyzed three-dimensional (3-D) matrices of wavelength/time/space code for optical code division multiple access (OCDMA) networks with NAND subtraction detection technique. The 3-D codes are constructed by integrating a two-dimensional modified quadratic congruence (MQC) code with one-dimensional modified prime (MP) code. The respective encoders and decoders were designed using fiber Bragg gratings and optical delay lines to minimize the bit error rate (BER). The performance analysis of the 3D-OCDMA system is based on measurement of signal to noise ratio (SNR), BER and eye diagram for a different number of simultaneous users. Also, in the analysis, various types of noises and multiple access interference (MAI) effects were considered. The results obtained with NAND detection technique were compared with those obtained with OR and AND subtraction techniques. The comparison results proved that the NAND detection technique with 3-D MQC\MP code can accommodate more number of simultaneous users for longer distances of fiber with minimum BER as compared to OR and AND subtraction techniques. The received optical power is also measured at various levels of BER to analyze the effect of attenuation.Keywords: Cross Correlation (CC), Three dimensional Optical Code Division Multiple Access (3-D OCDMA), Spectral Amplitude Coding Optical Code Division Multiple Access (SAC-OCDMA), Multiple Access Interference (MAI), Phase Induced Intensity Noise (PIIN), Three Dimensional Modified Quadratic Congruence/Modified Prime (3-D MQC/MP) code
Procedia PDF Downloads 4041862 Information-Controlled Laryngeal Feature Variations in Korean Consonants
Authors: Ponghyung Lee
Abstract:
This study seeks to investigate the variations occurring to Korean consonantal variations center around laryngeal features of the concerned sounds, to the exclusion of others. Our fundamental premise is that the weak contrast associated with concerned segments might be held accountable for the oscillation of the status quo of the concerned consonants. What is more, we assume that an array of notions as a measure of communicative efficiency of linguistic units would be significantly influential on triggering those variations. To this end, we have tried to compute the surprisal, entropic contribution, and relative contrastiveness associated with Korean obstruent consonants. What we found therein is that the Information-theoretic perspective is compelling enough to lend support our approach to a considerable extent. That is, the variant realizations, chronologically and stylistically, prove to be profoundly affected by a set of Information-theoretic factors enumerated above. When it comes to the biblical proper names, we use Georgetown University CQP Web-Bible corpora. From the 8 texts (4 from Old Testament and 4 from New Testament) among the total 64 texts, we extracted 199 samples. We address the issue of laryngeal feature variations associated with Korean obstruent consonants under the presumption that the variations stem from the weak contrast among the triad manifestations of laryngeal features. The variants emerge from diverse sources in chronological and stylistic senses: Christianity biblical texts, ordinary casual speech, the shift of loanword adaptation over time, and ideophones. For the purpose of discussing what they are really like from the perspective of Information Theory, it is necessary to closely look at the data. Among them, the massive changes occurring to loanword adaptation of proper nouns during the centennial history of Korean Christianity draw our special attention. We searched 199 types of initially capitalized words among 45,528-word tokens, which account for around 5% of total 901,701-word tokens (12,786-word types) from Georgetown University CQP Web-Bible corpora. We focus on the shift of the laryngeal features incorporated into word-initial consonants, which are available through the two distinct versions of Korean Bible: one came out in the 1960s for the Protestants, and the other was published in the 1990s for the Catholic Church. Of these proper names, we have closely traced the adaptation of plain obstruents, e. g. /b, d, g, s, ʤ/ in the sources. The results show that as much as 41% of the extracted proper names show variations; 37% in terms of aspiration, and 4% in terms of tensing. This study set out in an effort to shed light on the question: to what extent can we attribute the variations occurring to the laryngeal features associated with Korean obstruent consonants to the communicative aspects of linguistic activities? In this vein, the concerted effects of the triad, of surprisal, entropic contribution, and relative contrastiveness can be credited with the ups and downs in the feature specification, despite being contentiousness on the role of surprisal to some extent.Keywords: entropic contribution, laryngeal feature variation, relative contrastiveness, surprisal
Procedia PDF Downloads 1201861 A Machine Learning Approach for Assessment of Tremor: A Neurological Movement Disorder
Authors: Rajesh Ranjan, Marimuthu Palaniswami, A. A. Hashmi
Abstract:
With the changing lifestyle and environment around us, the prevalence of the critical and incurable disease has proliferated. One such condition is the neurological disorder which is rampant among the old age population and is increasing at an unstoppable rate. Most of the neurological disorder patients suffer from some movement disorder affecting the movement of their body parts. Tremor is the most common movement disorder which is prevalent in such patients that infect the upper or lower limbs or both extremities. The tremor symptoms are commonly visible in Parkinson’s disease patient, and it can also be a pure tremor (essential tremor). The patients suffering from tremor face enormous trouble in performing the daily activity, and they always need a caretaker for assistance. In the clinics, the assessment of tremor is done through a manual clinical rating task such as Unified Parkinson’s disease rating scale which is time taking and cumbersome. Neurologists have also affirmed a challenge in differentiating a Parkinsonian tremor with the pure tremor which is essential in providing an accurate diagnosis. Therefore, there is a need to develop a monitoring and assistive tool for the tremor patient that keep on checking their health condition by coordinating them with the clinicians and caretakers for early diagnosis and assistance in performing the daily activity. In our research, we focus on developing a system for automatic classification of tremor which can accurately differentiate the pure tremor from the Parkinsonian tremor using a wearable accelerometer-based device, so that adequate diagnosis can be provided to the correct patient. In this research, a study was conducted in the neuro-clinic to assess the upper wrist movement of the patient suffering from Pure (Essential) tremor and Parkinsonian tremor using a wearable accelerometer-based device. Four tasks were designed in accordance with Unified Parkinson’s disease motor rating scale which is used to assess the rest, postural, intentional and action tremor in such patient. Various features such as time-frequency domain, wavelet-based and fast-Fourier transform based cross-correlation were extracted from the tri-axial signal which was used as input feature vector space for the different supervised and unsupervised learning tools for quantification of severity of tremor. A minimum covariance maximum correlation energy comparison index was also developed which was used as the input feature for various classification tools for distinguishing the PT and ET tremor types. An automatic system for efficient classification of tremor was developed using feature extraction methods, and superior performance was achieved using K-nearest neighbors and Support Vector Machine classifiers respectively.Keywords: machine learning approach for neurological disorder assessment, automatic classification of tremor types, feature extraction method for tremor classification, neurological movement disorder, parkinsonian tremor, essential tremor
Procedia PDF Downloads 1481860 Enhancement Method of Network Traffic Anomaly Detection Model Based on Adversarial Training With Category Tags
Authors: Zhang Shuqi, Liu Dan
Abstract:
For the problems in intelligent network anomaly traffic detection models, such as low detection accuracy caused by the lack of training samples, poor effect with small sample attack detection, a classification model enhancement method, F-ACGAN(Flow Auxiliary Classifier Generative Adversarial Network) which introduces generative adversarial network and adversarial training, is proposed to solve these problems. Generating adversarial data with category labels could enhance the training effect and improve classification accuracy and model robustness. FACGAN consists of three steps: feature preprocess, which includes data type conversion, dimensionality reduction and normalization, etc.; A generative adversarial network model with feature learning ability is designed, and the sample generation effect of the model is improved through adversarial iterations between generator and discriminator. The adversarial disturbance factor of the gradient direction of the classification model is added to improve the diversity and antagonism of generated data and to promote the model to learn from adversarial classification features. The experiment of constructing a classification model with the UNSW-NB15 dataset shows that with the enhancement of FACGAN on the basic model, the classification accuracy has improved by 8.09%, and the score of F1 has improved by 6.94%.Keywords: data imbalance, GAN, ACGAN, anomaly detection, adversarial training, data augmentation
Procedia PDF Downloads 931859 Computer-Aided Classification of Liver Lesions Using Contrasting Features Difference
Authors: Hussein Alahmer, Amr Ahmed
Abstract:
Liver cancer is one of the common diseases that cause the death. Early detection is important to diagnose and reduce the incidence of death. Improvements in medical imaging and image processing techniques have significantly enhanced interpretation of medical images. Computer-Aided Diagnosis (CAD) systems based on these techniques play a vital role in the early detection of liver disease and hence reduce liver cancer death rate. This paper presents an automated CAD system consists of three stages; firstly, automatic liver segmentation and lesion’s detection. Secondly, extracting features. Finally, classifying liver lesions into benign and malignant by using the novel contrasting feature-difference approach. Several types of intensity, texture features are extracted from both; the lesion area and its surrounding normal liver tissue. The difference between the features of both areas is then used as the new lesion descriptors. Machine learning classifiers are then trained on the new descriptors to automatically classify liver lesions into benign or malignant. The experimental results show promising improvements. Moreover, the proposed approach can overcome the problems of varying ranges of intensity and textures between patients, demographics, and imaging devices and settings.Keywords: CAD system, difference of feature, fuzzy c means, lesion detection, liver segmentation
Procedia PDF Downloads 3131858 Investigation of Complexity Dynamics in a DC Glow Discharge Magnetized Plasma Using Recurrence Quantification Analysis
Authors: Vramori Mitra, Bornali Sarma, Arun K. Sarma
Abstract:
Recurrence is a ubiquitous feature of any real dynamical system. The states in phase space trajectory of a system have an inherent tendency to return to the same state or its close state after certain time laps. Recurrence quantification analysis technique, based on this fundamental feature of a dynamical system, detects evaluation of state under variation of control parameter of the system. The paper presents the investigation of nonlinear dynamical behavior of plasma floating potential fluctuations obtained by using a Langmuir probe in different magnetic field under the variation of discharge voltages. The main measures of recurrence quantification analysis are considered as determinism, linemax and entropy. The increment of the DET and linemax variables asserts that the predictability and periodicity of the system is increasing. The variable linemax indicates that the chaoticity is being diminished with the slump of magnetic field while increase of magnetic field enhancing the chaotic behavior. Fractal property of the plasma time series estimated by DFA technique (Detrended fluctuation analysis) reflects that long-range correlation of plasma fluctuations is decreasing while fractal dimension is increasing with the enhancement of magnetic field which corroborates the RQA analysis.Keywords: detrended fluctuation analysis, chaos, phase space, recurrence
Procedia PDF Downloads 3131857 Roughness Discrimination Using Bioinspired Tactile Sensors
Authors: Zhengkun Yi
Abstract:
Surface texture discrimination using artificial tactile sensors has attracted increasing attentions in the past decade as it can endow technical and robot systems with a key missing ability. However, as a major component of texture, roughness has rarely been explored. This paper presents an approach for tactile surface roughness discrimination, which includes two parts: (1) design and fabrication of a bioinspired artificial fingertip, and (2) tactile signal processing for tactile surface roughness discrimination. The bioinspired fingertip is comprised of two polydimethylsiloxane (PDMS) layers, a polymethyl methacrylate (PMMA) bar, and two perpendicular polyvinylidene difluoride (PVDF) film sensors. This artificial fingertip mimics human fingertips in three aspects: (1) Elastic properties of epidermis and dermis in human skin are replicated by the two PDMS layers with different stiffness, (2) The PMMA bar serves the role analogous to that of a bone, and (3) PVDF film sensors emulate Meissner’s corpuscles in terms of both location and response to the vibratory stimuli. Various extracted features and classification algorithms including support vector machines (SVM) and k-nearest neighbors (kNN) are examined for tactile surface roughness discrimination. Eight standard rough surfaces with roughness values (Ra) of 50 μm, 25 μm, 12.5 μm, 6.3 μm 3.2 μm, 1.6 μm, 0.8 μm, and 0.4 μm are explored. The highest classification accuracy of (82.6 ± 10.8) % can be achieved using solely one PVDF film sensor with kNN (k = 9) classifier and the standard deviation feature.Keywords: bioinspired fingertip, classifier, feature extraction, roughness discrimination
Procedia PDF Downloads 3021856 An Improved Adaptive Dot-Shape Beamforming Algorithm Research on Frequency Diverse Array
Authors: Yanping Liao, Zenan Wu, Ruigang Zhao
Abstract:
Frequency diverse array (FDA) beamforming is a technology developed in recent years, and its antenna pattern has a unique angle-distance-dependent characteristic. However, the beam is always required to have strong concentration, high resolution and low sidelobe level to form the point-to-point interference in the concentrated set. In order to eliminate the angle-distance coupling of the traditional FDA and to make the beam energy more concentrated, this paper adopts a multi-carrier FDA structure based on proposed power exponential frequency offset to improve the array structure and frequency offset of the traditional FDA. The simulation results show that the beam pattern of the array can form a dot-shape beam with more concentrated energy, and its resolution and sidelobe level performance are improved. However, the covariance matrix of the signal in the traditional adaptive beamforming algorithm is estimated by the finite-time snapshot data. When the number of snapshots is limited, the algorithm has an underestimation problem, which leads to the estimation error of the covariance matrix to cause beam distortion, so that the output pattern cannot form a dot-shape beam. And it also has main lobe deviation and high sidelobe level problems in the case of limited snapshot. Aiming at these problems, an adaptive beamforming technique based on exponential correction for multi-carrier FDA is proposed to improve beamforming robustness. The steps are as follows: first, the beamforming of the multi-carrier FDA is formed under linear constrained minimum variance (LCMV) criteria. Then the eigenvalue decomposition of the covariance matrix is performed to obtain the diagonal matrix composed of the interference subspace, the noise subspace and the corresponding eigenvalues. Finally, the correction index is introduced to exponentially correct the small eigenvalues of the noise subspace, improve the divergence of small eigenvalues in the noise subspace, and improve the performance of beamforming. The theoretical analysis and simulation results show that the proposed algorithm can make the multi-carrier FDA form a dot-shape beam at limited snapshots, reduce the sidelobe level, improve the robustness of beamforming, and have better performance.Keywords: adaptive beamforming, correction index, limited snapshot, multi-carrier frequency diverse array, robust
Procedia PDF Downloads 1211855 Robust Design of a Ball Joint Considering Uncertainties
Authors: Bong-Su Sin, Jong-Kyu Kim, Se-Il Song, Kwon-Hee Lee
Abstract:
An automobile ball joint is a pivoting element used to allow rotational motion between the parts of the steering and suspension system. And it plays a role in smooth transmission of steering movement, also reduction in impact from the road surface. A ball joint is under various repeated loadings that may cause cracks and abrasion. This damages lead to safety problems of a car, as well as reducing the comfort of the driver's ride, and raise questions about the ball joint procedure and the whole durability of the suspension system. Accordingly, it is necessary to ensure the high durability and reliability of a ball joint. The structural responses of stiffness and pull-out strength were then calculated to check if the design satisfies the related requirements. The analysis was sequentially performed, following the caulking process. In this process, the deformation and stress results obtained from the analysis were saved. Sequential analysis has a strong advantage, in that it can be analyzed by considering the deformed shape and residual stress. The pull-out strength means the required force to pull the ball stud out from the ball joint assembly. The low pull-out strength can deteriorate the structural stability and safety performances. In this study, two design variables and two noise factors were set up. Two design variables were the diameter of a stud and the angle of a socket. And two noise factors were defined as the uncertainties of Young's modulus and yield stress of a seat. The DOE comprises 81 cases using these conditions. Robust design of a ball joint was performed using the DOE. The pull-out strength was generated from the uncertainties in the design variables and the design parameters. The purpose of robust design is to find the design with target response and smallest variation.Keywords: ball joint, pull-out strength, robust design, design of experiments
Procedia PDF Downloads 4101854 Evolving Convolutional Filter Using Genetic Algorithm for Image Classification
Authors: Rujia Chen, Ajit Narayanan
Abstract:
Convolutional neural networks (CNN), as typically applied in deep learning, use layer-wise backpropagation (BP) to construct filters and kernels for feature extraction. Such filters are 2D or 3D groups of weights for constructing feature maps at subsequent layers of the CNN and are shared across the entire input. BP as a gradient descent algorithm has well-known problems of getting stuck at local optima. The use of genetic algorithms (GAs) for evolving weights between layers of standard artificial neural networks (ANNs) is a well-established area of neuroevolution. In particular, the use of crossover techniques when optimizing weights can help to overcome problems of local optima. However, the application of GAs for evolving the weights of filters and kernels in CNNs is not yet an established area of neuroevolution. In this paper, a GA-based filter development algorithm is proposed. The results of the proof-of-concept experiments described in this paper show the proposed GA algorithm can find filter weights through evolutionary techniques rather than BP learning. For some simple classification tasks like geometric shape recognition, the proposed algorithm can achieve 100% accuracy. The results for MNIST classification, while not as good as possible through standard filter learning through BP, show that filter and kernel evolution warrants further investigation as a new subarea of neuroevolution for deep architectures.Keywords: neuroevolution, convolutional neural network, genetic algorithm, filters, kernels
Procedia PDF Downloads 1781853 Random Forest Classification for Population Segmentation
Authors: Regina Chua
Abstract:
To reduce the costs of re-fielding a large survey, a Random Forest classifier was applied to measure the accuracy of classifying individuals into their assigned segments with the fewest possible questions. Given a long survey, one needed to determine the most predictive ten or fewer questions that would accurately assign new individuals to custom segments. Furthermore, the solution needed to be quick in its classification and usable in non-Python environments. In this paper, a supervised Random Forest classifier was modeled on a dataset with 7,000 individuals, 60 questions, and 254 features. The Random Forest consisted of an iterative collection of individual decision trees that result in a predicted segment with robust precision and recall scores compared to a single tree. A random 70-30 stratified sampling for training the algorithm was used, and accuracy trade-offs at different depths for each segment were identified. Ultimately, the Random Forest classifier performed at 87% accuracy at a depth of 10 with 20 instead of 254 features and 10 instead of 60 questions. With an acceptable accuracy in prioritizing feature selection, new tools were developed for non-Python environments: a worksheet with a formulaic version of the algorithm and an embedded function to predict the segment of an individual in real-time. Random Forest was determined to be an optimal classification model by its feature selection, performance, processing speed, and flexible application in other environments.Keywords: machine learning, supervised learning, data science, random forest, classification, prediction, predictive modeling
Procedia PDF Downloads 861852 Application of Seasonal Autoregressive Integrated Moving Average Model for Forecasting Monthly Flows in Waterval River, South Africa
Authors: Kassahun Birhanu Tadesse, Megersa Olumana Dinka
Abstract:
Reliable future river flow information is basic for planning and management of any river systems. For data scarce river system having only a river flow records like the Waterval River, a univariate time series models are appropriate for river flow forecasting. In this study, a univariate Seasonal Autoregressive Integrated Moving Average (SARIMA) model was applied for forecasting Waterval River flow using GRETL statistical software. Mean monthly river flows from 1960 to 2016 were used for modeling. Different unit root tests and Mann-Kendall trend analysis were performed to test the stationarity of the observed flow time series. The time series was differenced to remove the seasonality. Using the correlogram of seasonally differenced time series, different SARIMA models were identified, their parameters were estimated, and diagnostic check-up of model forecasts was performed using white noise and heteroscedasticity tests. Finally, based on minimum Akaike Information (AIc) and Hannan-Quinn (HQc) criteria, SARIMA (3, 0, 2) x (3, 1, 3)12 was selected as the best model for Waterval River flow forecasting. Therefore, this model can be used to generate future river information for water resources development and management in Waterval River system. SARIMA model can also be used for forecasting other similar univariate time series with seasonality characteristics.Keywords: heteroscedasticity, stationarity test, trend analysis, validation, white noise
Procedia PDF Downloads 1971851 Emerging Film Makers in Tamil Cinema Liberated by Digital Media
Authors: Valarmathi Subramaniam
Abstract:
Ever since the first Indian feature film was produced and released by Shri Dada Saheb Phalke in the year 1931, the Indian Film Industry has grown leaps and bounds. The Indian Film Industry stands as the largest film industry in the world, and it produces more than a thousand films every year with investments and revenues worth several billion rupees. As per the official report published by UNESCO in the year 2017 on their website, it states that in the year 2015, India has produced one thousand nine hundred and seven feature films using digital technology. Not only is the cinema adapted to digital technologies, but the digital technologies also opened up avenues for talents to enter the cinema industry. This paper explores such talents who have emerged in the film industry without any background, neither academic nor from their family background, but holding digital media as their weapon. The research involves two variants of filmmaking technology – Celluloid and Digital. The study used a selective sampling of films that were released from the year 2020-to 2022. The sample has been organized, resulting in popular and fresh talents in the editing phase of filmmaking. There were 48 editors, of which 12 editors were not popular and 6 of them were fresh into the film without any background. Interview methods were used to collect data on what helped them to get into the industry straight. The study found that the digital medium and the digital technology enabled them to get into the film industry.Keywords: digital media, digital in cinema, digital era talents, emerging new talents
Procedia PDF Downloads 1081850 The Impact of Temperamental Traits of Candidates for Aviation School on Their Strategies for Coping with Stress during Selection Exams in Physical Education
Authors: Robert Jedrys, Zdzislaw Kobos, Justyna Skrzynska, Zbigniew Wochynski
Abstract:
Professions connected to aviation require an assessment of the suitability of health, psychological and psychomotor skills and overall physical fitness of the organism, who applies. Assessment of the physical condition is conducted by the committees consisting of aero-medical specialists in clinical medicine and aviation. In addition, psychological predispositions should be evaluated by specialized psychologists familiar with the specifics of the tasks and requirements for the various positions in aviation. Both, physical abilities and general physical fitness of candidates for aviation shall be assessed during the selection exams, which also test the ability to deal with stress what is very important in aviation. Hence, the mentioned exams in physical education not only help to judge on the ranking in candidates in terms of their efficiency and performance, but also allows to evaluate the functioning under stress measured using psychological tests. Moreover, before-test stress is a predictors of successfulness in the next stages of education and practical training in the aviation. The aim of the study was to evaluate the influence of temperamental traits on strategies used for coping with stress during selection exams in physical education, deciding on admission to aviation school. The study involved 30 candidates for fighter pilot training in aviation school . To evaluate the temperament 'The Formal Characteristics of Behavior-Temperament Inventory' (FCB-TI) by B. Zawadzki and J.Strelau was used. To determine the pattern of coping with stress 'The Coping Inventory for Stressful Situations' (CISS) to N. S. Endler and J. D. A. Parker were engaged. Study of temperament and styles of coping with stress was conducted directly before the exam selection of physical education. The results were analyzed with 'Statistica 9' program. The studies showed that:-There is a negative correlation between such a temperament feature as 'perseverance' and preferred style of coping with stress concentrated on the task (r = -0.590; p < 0.004); -There is a positive correlation between such a feature of temperament as 'emotional reactivity,' and preference to deal with a stressful situation with ‘style centered on emotions’ (r = 0.520; p <0.011); -There is a negative correlation between such a feature of temperament as ‘strength’ and ‘style of coping with stress concentrated on emotions’ (r = -0.580; p < 0.004). Studies indicate that temperament traits determine the perception of stress and preferred coping styles used during the selection, as during the exams in physical education.Keywords: aviation, physical education, stress, temperamental traits
Procedia PDF Downloads 2461849 Surface Flattening Assisted with 3D Mannequin Based on Minimum Energy
Authors: Shih-Wen Hsiao, Rong-Qi Chen, Chien-Yu Lin
Abstract:
The topic of surface flattening plays a vital role in the field of computer aided design and manufacture. Surface flattening enables the production of 2D patterns and it can be used in design and manufacturing for developing a 3D surface to a 2D platform, especially in fashion design. This study describes surface flattening based on minimum energy methods according to the property of different fabrics. Firstly, through the geometric feature of a 3D surface, the less transformed area can be flattened on a 2D platform by geodesic. Then, strain energy that has accumulated in mesh can be stably released by an approximate implicit method and revised error function. In some cases, cutting mesh to further release the energy is a common way to fix the situation and enhance the accuracy of the surface flattening, and this makes the obtained 2D pattern naturally generate significant cracks. When this methodology is applied to a 3D mannequin constructed with feature lines, it enhances the level of computer-aided fashion design. Besides, when different fabrics are applied to fashion design, it is necessary to revise the shape of a 2D pattern according to the properties of the fabric. With this model, the outline of 2D patterns can be revised by distributing the strain energy with different results according to different fabric properties. Finally, this research uses some common design cases to illustrate and verify the feasibility of this methodology.Keywords: surface flattening, strain energy, minimum energy, approximate implicit method, fashion design
Procedia PDF Downloads 3261848 Beam Coding with Orthogonal Complementary Golay Codes for Signal to Noise Ratio Improvement in Ultrasound Mammography
Authors: Y. Kumru, K. Enhos, H. Köymen
Abstract:
In this paper, we report the experimental results on using complementary Golay coded signals at 7.5 MHz to detect breast microcalcifications of 50 µm size. Simulations using complementary Golay coded signals show perfect consistence with the experimental results, confirming the improved signal to noise ratio for complementary Golay coded signals. For improving the success on detecting the microcalcifications, orthogonal complementary Golay sequences having cross-correlation for minimum interference are used as coded signals and compared to tone burst pulse of equal energy in terms of resolution under weak signal conditions. The measurements are conducted using an experimental ultrasound research scanner, Digital Phased Array System (DiPhAS) having 256 channels, a phased array transducer with 7.5 MHz center frequency and the results obtained through experiments are validated by Field-II simulation software. In addition, to investigate the superiority of coded signals in terms of resolution, multipurpose tissue equivalent phantom containing series of monofilament nylon targets, 240 µm in diameter, and cyst-like objects with attenuation of 0.5 dB/[MHz x cm] is used in the experiments. We obtained ultrasound images of monofilament nylon targets for the evaluation of resolution. Simulation and experimental results show that it is possible to differentiate closely positioned small targets with increased success by using coded excitation in very weak signal conditions.Keywords: coded excitation, complementary golay codes, DiPhAS, medical ultrasound
Procedia PDF Downloads 2541847 Fault Detection and Isolation in Sensors and Actuators of Wind Turbines
Authors: Shahrokh Barati, Reza Ramezani
Abstract:
Due to the countries growing attention to the renewable energy producing, the demand for energy from renewable energy has gone up among the renewable energy sources; wind energy is the fastest growth in recent years. In this regard, in order to increase the availability of wind turbines, using of Fault Detection and Isolation (FDI) system is necessary. Wind turbines include of various faults such as sensors fault, actuator faults, network connection fault, mechanical faults and faults in the generator subsystem. Although, sensors and actuators have a large number of faults in wind turbine but have discussed fewer in the literature. Therefore, in this work, we focus our attention to design a sensor and actuator fault detection and isolation algorithm and Fault-tolerant control systems (FTCS) for Wind Turbine. The aim of this research is to propose a comprehensive fault detection and isolation system for sensors and actuators of wind turbine based on data-driven approaches. To achieve this goal, the features of measurable signals in real wind turbine extract in any condition. The next step is the feature selection among the extract in any condition. The next step is the feature selection among the extracted features. Features are selected that led to maximum separation networks that implemented in parallel and results of classifiers fused together. In order to maximize the reliability of decision on fault, the property of fault repeatability is used.Keywords: FDI, wind turbines, sensors and actuators faults, renewable energy
Procedia PDF Downloads 3911846 An Approach for Vocal Register Recognition Based on Spectral Analysis of Singing
Authors: Aleksandra Zysk, Pawel Badura
Abstract:
Recognizing and controlling vocal registers during singing is a difficult task for beginner vocalist. It requires among others identifying which part of natural resonators is being used when a sound propagates through the body. Thus, an application has been designed allowing for sound recording, automatic vocal register recognition (VRR), and a graphical user interface providing real-time visualization of the signal and recognition results. Six spectral features are determined for each time frame and passed to the support vector machine classifier yielding a binary decision on the head or chest register assignment of the segment. The classification training and testing data have been recorded by ten professional female singers (soprano, aged 19-29) performing sounds for both chest and head register. The classification accuracy exceeded 93% in each of various validation schemes. Apart from a hard two-class clustering, the support vector classifier returns also information on the distance between particular feature vector and the discrimination hyperplane in a feature space. Such an information reflects the level of certainty of the vocal register classification in a fuzzy way. Thus, the designed recognition and training application is able to assess and visualize the continuous trend in singing in a user-friendly graphical mode providing an easy way to control the vocal emission.Keywords: classification, singing, spectral analysis, vocal emission, vocal register
Procedia PDF Downloads 2941845 Contrast-to-Noise Ratio Comparison of Different Calcification Types in Dual Energy Breast Imaging
Authors: Vaia N. Koukou, Niki D. Martini, George P. Fountos, Christos M. Michail, Athanasios Bakas, Ioannis S. Kandarakis, George C. Nikiforidis
Abstract:
Various substitute materials of calcifications are used in phantom measurements and simulation studies in mammography. These include calcium carbonate, calcium oxalate, hydroxyapatite and aluminum. The aim of this study is to compare the contrast-to-noise ratio (CNR) values of the different calcification types using the dual energy method. The constructed calcification phantom consisted of three different calcification types and thicknesses: hydroxyapatite, calcite and calcium oxalate of 100, 200, 300 thicknesses. The breast tissue equivalent materials were polyethylene and polymethyl methacrylate slabs simulating adipose tissue and glandular tissue, respectively. The total thickness was 4.2 cm with 50% fixed glandularity. The low- (LE) and high-energy (HE) images were obtained from a tungsten anode using 40 kV filtered with 0.1 mm cadmium and 70 kV filtered with 1 mm copper, respectively. A high resolution complementary metal-oxide-semiconductor (CMOS) active pixel sensor (APS) X-ray detector was used. The total mean glandular dose (MGD) and entrance surface dose (ESD) from the LE and HE images were constrained to typical levels (MGD=1.62 mGy and ESD=1.92 mGy). On average, the CNR of hydroxyapatite calcifications was 1.4 times that of calcite calcifications and 2.5 times that of calcium oxalate calcifications. The higher CNR values of hydroxyapatite are attributed to its attenuation properties compared to the other calcification materials, leading to higher contrast in the dual energy image. This work was supported by Grant Ε.040 from the Research Committee of the University of Patras (Programme K. Karatheodori).Keywords: calcification materials, CNR, dual energy, X-rays
Procedia PDF Downloads 3411844 Using Machine Learning to Classify Human Fetal Health and Analyze Feature Importance
Authors: Yash Bingi, Yiqiao Yin
Abstract:
Reduction of child mortality is an ongoing struggle and a commonly used factor in determining progress in the medical field. The under-5 mortality number is around 5 million around the world, with many of the deaths being preventable. In light of this issue, Cardiotocograms (CTGs) have emerged as a leading tool to determine fetal health. By using ultrasound pulses and reading the responses, CTGs help healthcare professionals assess the overall health of the fetus to determine the risk of child mortality. However, interpreting the results of the CTGs is time-consuming and inefficient, especially in underdeveloped areas where an expert obstetrician is hard to come by. Using a support vector machine (SVM) and oversampling, this paper proposed a model that classifies fetal health with an accuracy of 99.59%. To further explain the CTG measurements, an algorithm based on Randomized Input Sampling for Explanation ((RISE) of Black-box Models was created, called Feature Alteration for explanation of Black Box Models (FAB), and compared the findings to Shapley Additive Explanations (SHAP) and Local Interpretable Model Agnostic Explanations (LIME). This allows doctors and medical professionals to classify fetal health with high accuracy and determine which features were most influential in the process.Keywords: machine learning, fetal health, gradient boosting, support vector machine, Shapley values, local interpretable model agnostic explanations
Procedia PDF Downloads 1351843 Verbal Prefix Selection in Old Japanese: A Corpus-Based Study
Authors: Zixi You
Abstract:
There are a number of verbal prefixes in Old Japanese. However, the selection or the compatibility of verbs and verbal prefixes is among the least investigated topics on Old Japanese language. Unlike other types of prefixes, verbal prefixes in dictionaries are more often than not listed with very brief information such as ‘unknown meaning’ or ‘rhythmic function only’. To fill in a part of this knowledge gap, this paper presents an exhaustive investigation based on the newly developed ‘Oxford Corpus of Old Japanese’ (OCOJ), which included nearly all existing resource of Old Japanese language, with detailed linguistics information in TEI-XML tags. In this paper, we propose the possibility that the following three prefixes, i-, sa-, ta- (with ta- being considered as a variation of sa-), are relevant to split intransitivity in Old Japanese, with evidence that unergative verbs favor i- and that unergative verbs favor sa-(ta-). This might be undermined by the fact that transitives are also found to follow i-. However, with several manifestations of split intransitivity in Old Japanese discussed, the behavior of transitives in verbal prefix selection is no longer as surprising as it may seem to be when one look at the selection of verbal prefix in isolation. It is possible that there are one or more features that played essential roles in determining the selection of i-, and the attested transitive verbs happen to have these features. The data suggest that this feature is a sense of ‘change’ of location or state involved in the event donated by the verb, which is a feature of typical unaccusatives. This is further discussed in the ‘affectedness’ hierarchy. The presentation of this paper, which includes a brief demonstration of the OCOJ, is expected to be of the interest of both specialists and general audiences.Keywords: old Japanese, split intransitivity, unaccusatives, unergatives, verbal prefix selection
Procedia PDF Downloads 4031842 Methods of Livable Goal-Oriented Master Urban Design: A Case Study on Zibo City
Authors: Xiaoping Zhang, Fengying Yan
Abstract:
The implementation of the 'Urban Design Management Measures' requires that the master urban design should aim at creating a livable urban space. However, to our best knowledge, the existing researches and practices of master urban design not only focus less on the livable space but also face a number of problems such as paying more attention to the image of the city, ignoring the people-oriented and lacking dynamic continuity. In order to make the master urban design can better guide the construction of city. Firstly, the paper proposes the livable city hierarchy system to meet the needs of different groups of people and then constructs the framework of livable goal-oriented master urban design based on the theory of livable content and the ideological origin of people-oriented. Secondly, the paper takes the master urban design practice of Zibo as a sample and puts forward the design strategy of strengthening the pattern, improve the quality of space, shape the feature, and establish a series of action plans based on the strategy of urban space development. Finally, the paper explores the method system of livable goal-oriented master urban design from the aspects of safety pattern, morphology pattern, neighborhood scale, open space, street space, public interface, style feature, public participation and action plans.Keywords: livable, master urban design, public participation, zibo city
Procedia PDF Downloads 3031841 From Shallow Semantic Representation to Deeper One: Verb Decomposition Approach
Authors: Aliaksandr Huminski
Abstract:
Semantic Role Labeling (SRL) as shallow semantic parsing approach includes recognition and labeling arguments of a verb in a sentence. Verb participants are linked with specific semantic roles (Agent, Patient, Instrument, Location, etc.). Thus, SRL can answer on key questions such as ‘Who’, ‘When’, ‘What’, ‘Where’ in a text and it is widely applied in dialog systems, question-answering, named entity recognition, information retrieval, and other fields of NLP. However, SRL has the following flaw: Two sentences with identical (or almost identical) meaning can have different semantic role structures. Let consider 2 sentences: (1) John put butter on the bread. (2) John buttered the bread. SRL for (1) and (2) will be significantly different. For the verb put in (1) it is [Agent + Patient + Goal], but for the verb butter in (2) it is [Agent + Goal]. It happens because of one of the most interesting and intriguing features of a verb: Its ability to capture participants as in the case of the verb butter, or their features as, say, in the case of the verb drink where the participant’s feature being liquid is shared with the verb. This capture looks like a total fusion of meaning and cannot be decomposed in direct way (in comparison with compound verbs like babysit or breastfeed). From this perspective, SRL looks really shallow to represent semantic structure. If the key point in semantic representation is an opportunity to use it for making inferences and finding hidden reasons, it assumes by default that two different but semantically identical sentences must have the same semantic structure. Otherwise we will have different inferences from the same meaning. To overcome the above-mentioned flaw, the following approach is suggested. Assume that: P is a participant of relation; F is a feature of a participant; Vcp is a verb that captures a participant; Vcf is a verb that captures a feature of a participant; Vpr is a primitive verb or a verb that does not capture any participant and represents only a relation. In another word, a primitive verb is a verb whose meaning does not include meanings from its surroundings. Then Vcp and Vcf can be decomposed as: Vcp = Vpr +P; Vcf = Vpr +F. If all Vcp and Vcf will be represented this way, then primitive verbs Vpr can be considered as a canonical form for SRL. As a result of that, there will be no hidden participants caught by a verb since all participants will be explicitly unfolded. An obvious example of Vpr is the verb go, which represents pure movement. In this case the verb drink can be represented as man-made movement of liquid into specific direction. Extraction and using primitive verbs for SRL create a canonical representation unique for semantically identical sentences. It leads to the unification of semantic representation. In this case, the critical flaw related to SRL will be resolved.Keywords: decomposition, labeling, primitive verbs, semantic roles
Procedia PDF Downloads 3551840 The Effect of the Archeological and Architectural Nature of the Cities on the Design of Public Transportation Vehicles
Authors: Mohamed Moheyeldin Mahmoud
Abstract:
Various Islamic, Coptic and Jewish archeological places are located in many Egyptian neighborhoods such as Alsayeda zainab, Aldarb Alahmar, Algammaleya and many other in which they are daily exposed to a great traffic intensity causing vibrations. Vibrations could be stated as one of the most important challenges that face the archeological buildings and threaten their survival. The impact of vibrations varies according to the nature of the soil, nature and building conditions, how far the source of vibration is and the period of exposure. Traffic vibrations could be also stated as one of the most common types of vibrations having the greatest impact on buildings and archaeological installations. These vibrations result from the way that the vehicles act with different types of roads which vary according to the shape, nature, and type of obstacles. Other elements concerning the vehicle itself such as speed, weight, and load have a direct impact on the vibrations resulting from the vehicle movement that couldn't be neglected. The research aims to determine some of the requirements that must be observed when designing the public means of transport operating in the archaeological areas, in order to preserve the archaeological nature of the place. The research concludes that light weight slow motion vehicles should be used (25-50 km/h at maximum) having a multi-leaf steel spring suspension system instead of having an air-bag one should be used in order to reduce generated vibrations that could destroy the archeological buildings. Isolation layers could be used in the engine chamber in order to reduce the resulting noise causing vibrations. Electrically operated engines that use solar photovoltaic cells as a source of electricity could be used instead of gas ones in order to reduce the resulting engine noise.Keywords: archeological, design, isolation layers, suspension, vibrations
Procedia PDF Downloads 1861839 An Algebraic Geometric Imaging Approach for Automatic Dairy Cow Body Condition Scoring System
Authors: Thi Thi Zin, Pyke Tin, Ikuo Kobayashi, Yoichiro Horii
Abstract:
Today dairy farm experts and farmers have well recognized the importance of dairy cow Body Condition Score (BCS) since these scores can be used to optimize milk production, managing feeding system and as an indicator for abnormality in health even can be utilized to manage for having healthy calving times and process. In tradition, BCS measures are done by animal experts or trained technicians based on visual observations focusing on pin bones, pin, thurl and hook area, tail heads shapes, hook angles and short and long ribs. Since the traditional technique is very manual and subjective, the results can lead to different scores as well as not cost effective. Thus this paper proposes an algebraic geometric imaging approach for an automatic dairy cow BCS system. The proposed system consists of three functional modules. In the first module, significant landmarks or anatomical points from the cow image region are automatically extracted by using image processing techniques. To be specific, there are 23 anatomical points in the regions of ribs, hook bones, pin bone, thurl and tail head. These points are extracted by using block region based vertical and horizontal histogram methods. According to animal experts, the body condition scores depend mainly on the shape structure these regions. Therefore the second module will investigate some algebraic and geometric properties of the extracted anatomical points. Specifically, the second order polynomial regression is employed to a subset of anatomical points to produce the regression coefficients which are to be utilized as a part of feature vector in scoring process. In addition, the angles at thurl, pin, tail head and hook bone area are computed to extend the feature vector. Finally, in the third module, the extracted feature vectors are trained by using Markov Classification process to assign BCS for individual cows. Then the assigned BCS are revised by using multiple regression method to produce the final BCS score for dairy cows. In order to confirm the validity of proposed method, a monitoring video camera is set up at the milk rotary parlor to take top view images of cows. The proposed method extracts the key anatomical points and the corresponding feature vectors for each individual cows. Then the multiple regression calculator and Markov Chain Classification process are utilized to produce the estimated body condition score for each cow. The experimental results tested on 100 dairy cows from self-collected dataset and public bench mark dataset show very promising with accuracy of 98%.Keywords: algebraic geometric imaging approach, body condition score, Markov classification, polynomial regression
Procedia PDF Downloads 1481838 From Electroencephalogram to Epileptic Seizures Detection by Using Artificial Neural Networks
Authors: Gaetano Zazzaro, Angelo Martone, Roberto V. Montaquila, Luigi Pavone
Abstract:
Seizure is the main factor that affects the quality of life of epileptic patients. The diagnosis of epilepsy, and hence the identification of epileptogenic zone, is commonly made by using continuous Electroencephalogram (EEG) signal monitoring. Seizure identification on EEG signals is made manually by epileptologists and this process is usually very long and error prone. The aim of this paper is to describe an automated method able to detect seizures in EEG signals, using knowledge discovery in database process and data mining methods and algorithms, which can support physicians during the seizure detection process. Our detection method is based on Artificial Neural Network classifier, trained by applying the multilayer perceptron algorithm, and by using a software application, called Training Builder that has been developed for the massive extraction of features from EEG signals. This tool is able to cover all the data preparation steps ranging from signal processing to data analysis techniques, including the sliding window paradigm, the dimensionality reduction algorithms, information theory, and feature selection measures. The final model shows excellent performances, reaching an accuracy of over 99% during tests on data of a single patient retrieved from a publicly available EEG dataset.Keywords: artificial neural network, data mining, electroencephalogram, epilepsy, feature extraction, seizure detection, signal processing
Procedia PDF Downloads 176