Search results for: input performance

Authors: Tatiana A. Dolenko, Sergey A. Burikov, Alexander O. Efitorov, Sergey A. Dolenko

Abstract:

In this study, a comparative analysis of the approaches associated with the use of neural network algorithms for effective solution of a complex inverse problem – the problem of identifying and determining the individual concentrations of inorganic salts in multicomponent aqueous solutions by the spectra of Raman scattering of light – is performed. It is shown that application of artificial neural networks provides the average accuracy of determination of concentration of each salt no worse than 0.025 M. The results of comparative analysis of input data compression methods are presented. It is demonstrated that use of uniform aggregation of input features allows decreasing the error of determination of individual concentrations of components by 16-18% on the average.

Keywords: inverse problems, multi-component solutions, neural networks, Raman spectroscopy

Procedia PDF Downloads 524

Authors: Jiahao Cui

Abstract:

In the past centuries, due to climate change and intense whaling, the global whale population has dramatically declined. Among the various whale species, the fin whale experienced the most drastic drop in number due to its popularity in whaling. Under this background, identifying fin whale calls could be immensely beneficial to the preservation of the species. This paper uses feature extraction to process the input audio signal, then a network based on AlexNet and three networks based on the ResNet model was constructed to classify fin whale calls. A mixture of the DOSITS database and the Watkins database was used during training. The results demonstrate that a modified ResNet network has the best performance considering precision and network complexity.

Keywords: convolutional neural network, ResNet, AlexNet, fin whale preservation, feature extraction

Procedia PDF Downloads 116

Authors: Teklay Tesfay, Gebreyesus Brhane Tesfahunegn, Abadi Berhane, Selemawit Girmay

Abstract:

Adjusting the proper sowing date of a crop at a particular location with a changing climate is an essential management option to maximize crop yield. However, determining the optimum sowing date for rainfed maize production through field experimentation requires repeated trials for many years in different weather conditions and crop management. To avoid such long-term experimentation to determine the optimum sowing date, crop models such as AquaCrop are useful. Therefore, the overall objective of this study was to evaluate the performance of AquaCrop model in simulating maize productivity under varying sowing dates. A field experiment was conducted for two consecutive cropping seasons by deploying four maize seed sowing dates in a randomized complete block design with three replications. Input data required to run this model are stored as climate, crop, soil, and management files in the AquaCrop database and adjusted through the user interface. Observed data from separate field experiments was used to calibrate and validate the model. AquaCrop model was validated for its performance in simulating the green canopy and aboveground biomass of maize for the varying sowing dates based on the calibrated parameters. Results of the present study showed that there was a good agreement (an overall R2 =, Ef= d= RMSE =) between measured and simulated values of the canopy cover and biomass yields. Considering the overall values of the statistical test indicators, the performance of the model to predict maize growth and biomass yield was successful, and so this is a valuable tool help for decision-making. Hence, this calibrated and validated model is suggested to use for determining optimum maize crop sowing date for similar climate and soil conditions to the study area, instead of conducting long-term experimentation.

Keywords: AquaCrop model, calibration, validation, simulation

Procedia PDF Downloads 58

Authors: Chidozie C. Nwobi-Okoye, Basil Q. Ochieze, Stanley Okiy

Abstract:

This research reports on the modeling of age hardening process using adaptive neuro-fuzzy inference system (ANFIS). The age hardening output (Hardness) was predicted using ANFIS. The input parameters were ageing time, temperature and percentage composition of cow horn particles (CHp%). The results show the correlation coefficient (R) of the predicted hardness values versus the measured values was of 0.9985. Subsequently, values outside the experimental data points were predicted. When the temperature was kept constant, and other input parameters were varied, the average relative error of the predicted values was 0.0931%. When the temperature was varied, and other input parameters kept constant, the average relative error of the hardness values predictions was 80%. The results show that ANFIS with coarse experimental data points for learning is not very effective in predicting process outputs in the age hardening operation of A356 alloy/CHp particulate composite. The fine experimental data requirements by ANFIS make it more expensive in modeling and optimization of age hardening operations of A356 alloy/CHp particulate composite.

Keywords: adaptive neuro-fuzzy inference system (ANFIS), age hardening, aluminum alloy, metal matrix composite

Procedia PDF Downloads 149

Authors: Amir Ahmad Dehghani, Morteza Nabizadeh

Abstract:

This paper presents the application of an Adaptive Neuro-Fuzzy Inference System (ANFIS) to flood hydrograph modeling of Shahid Rajaee reservoir dam located in Iran. This was carried out using 11 flood hydrographs recorded in Tajan river gauging station. From this dataset, 9 flood hydrographs were chosen to train the model and 2 flood hydrographs to test the model. The different architectures of neuro-fuzzy model according to the membership function and learning algorithm were designed and trained with different epochs. The results were evaluated in comparison with the observed hydrographs and the best structure of model was chosen according the least RMSE in each performance. To evaluate the efficiency of neuro-fuzzy model, various statistical indices such as Nash-Sutcliff and flood peak discharge error criteria were calculated. In this simulation, the coordinates of a flood hydrograph including peak discharge were estimated using the discharge values occurred in the earlier time steps as input values to the neuro-fuzzy model. These results indicate the satisfactory efficiency of neuro-fuzzy model for flood simulating. This performance of the model demonstrates the suitability of the implemented approach to flood management projects.

Keywords: adaptive neuro-fuzzy inference system, flood hydrograph, hybrid learning algorithm, Shahid Rajaee reservoir dam

Procedia PDF Downloads 475

Authors: Reza Safari, Ali Rajabzadeh Ghatari, Raheleh Hossseinzadeh Mahabadi

Abstract:

In the present world, there are different pressures on firms such as competition, legislations, social etc. these pressures force the firms to follow “survival” as their primary goal and then growth. One of the main factors that helps firms to reach their goals is proper financial performance. To find out about the financial performance, a firm should monitors its financial performance. Financial performance affected by many factors. This research seeks to clear which financial performance indicators are most important according to Environmental situation of a firm and what are their priorities. To do so, environmental indicators specified as presented on OECD Key Environmental Indicators 2008 and so the financial performance indicators such as Profitability, Liquidity, Gearing, Investor ratios, and etc. At this stage, the affections questioned through questionnaires. After gaining the results, data analyzed using Promethee technique. By using decision matrixes extracted from those techniques an expert system designed. This expert system suggests the suitable financial performance indicators and their ranking by receiving the environment situation given environment indicators weight.

Keywords: environment indicators, financial performance indicators, promethee, expert system

Procedia PDF Downloads 434

Authors: Z. Gharaee

Abstract:

Recognizing actions performed by others is important in our daily lives since it is necessary for communicating with others in a proper way. We perceive an action by observing the kinematics of motions involved in the performance. We use our experience and concepts to make a correct recognition of the actions. Although building the action concepts is a life-long process, which is repeated throughout life, we are very efficient in applying our learned concepts in analyzing motions and recognizing actions. Experiments on the subjects observing the actions performed by an actor show that an action is recognized after only about two hundred milliseconds of observation. In this study, hierarchical action recognition architecture is proposed by using growing grid layers. The first-layer growing grid receives the pre-processed data of consecutive 3D postures of joint positions and applies some heuristics during the growth phase to allocate areas of the map by inserting new neurons. As a result of training the first-layer growing grid, action pattern vectors are generated by connecting the elicited activations of the learned map. The ordered vector representation layer receives action pattern vectors to create time-invariant vectors of key elicited activations. Time-invariant vectors are sent to second-layer growing grid for categorization. This grid creates the clusters representing the actions. Finally, one-layer neural network developed by a delta rule labels the action categories in the last layer. System performance has been evaluated in an experiment with the publicly available MSR-Action3D dataset. There are actions performed by using different parts of human body: Hand Clap, Two Hands Wave, Side Boxing, Bend, Forward Kick, Side Kick, Jogging, Tennis Serve, Golf Swing, Pick Up and Throw. The growing grid architecture was trained by applying several random selections of generalization test data fed to the system during on average 100 epochs for each training of the first-layer growing grid and around 75 epochs for each training of the second-layer growing grid. The average generalization test accuracy is 92.6%. A comparison analysis between the performance of growing grid architecture and self-organizing map (SOM) architecture in terms of accuracy and learning speed show that the growing grid architecture is superior to the SOM architecture in action recognition task. The SOM architecture completes learning the same dataset of actions in around 150 epochs for each training of the first-layer SOM while it takes 1200 epochs for each training of the second-layer SOM and it achieves the average recognition accuracy of 90% for generalization test data. In summary, using the growing grid network preserves the fundamental features of SOMs, such as topographic organization of neurons, lateral interactions, the abilities of unsupervised learning and representing high dimensional input space in the lower dimensional maps. The architecture also benefits from an automatic size setting mechanism resulting in higher flexibility and robustness. Moreover, by utilizing growing grids the system automatically obtains a prior knowledge of input space during the growth phase and applies this information to expand the map by inserting new neurons wherever there is high representational demand.

Keywords: action recognition, growing grid, hierarchical architecture, neural networks, system performance

Procedia PDF Downloads 155

Authors: Ahmed Najah Ahmed Al-Mahfoodh, Ali Najah Ahmed Al-Mahfoodh, Ahmed Al-Shafie

Abstract:

In this study, genetic programming (GP) technique has been investigated in prediction of set of rainfall-runoff data. To assess the effect of input parameters on the model, the sensitivity analysis was adopted. To evaluate the performance of the proposed model, three statistical indexes were used, namely; Correlation Coefficient (CC), Mean Square Error (MSE) and Correlation of Efficiency (CE). The principle aim of this study is to develop a computationally efficient and robust approach for predict of rainfall-runoff which could reduce the cost and labour for measuring these parameters. This research concentrates on the Johor River in Johor State, Malaysia.

Keywords: genetic programming, prediction, rainfall-runoff, Malaysia

Procedia PDF Downloads 474

Authors: Alfred L. Guiffrida

Abstract:

A review of the literature on supply chain delivery models which use delivery windows to measure delivery performance is presented. The review herein serves to meet the following objectives: (i) provide a synthesis of previously published literature on supply chain delivery performance models, (ii) provide in one paper a consolidation of research that can serve as a single source to keep researchers up to date with the research developments in supply chain delivery models, and (iii) identify gaps in the modeling of supply chain delivery performance which could stimulate new research agendas.

Keywords: delivery performance, delivery window, supply chain delivery models, supply chain performance

Procedia PDF Downloads 416

Authors: Huayuan Lu, Chunfang Yang, Ma Zhu, Baojun Qi, Yaqiong Qiao, Jiangqian Xu

Abstract:

Image geolocalization has great application prospects in fields such as autonomous driving and virtual/augmented reality. In practical application scenarios, the size of the image to be located is not fixed; it is impractical to train different networks for all possible sizes. When its size does not match the size of the input of the descriptor extraction model, existing image geolocalization methods usually directly scale or crop the image in some common ways. This will result in the loss of some information important to the geolocalization task, thus affecting the performance of the image geolocalization method. For example, excessive down-sampling can lead to blurred building contour, and inappropriate cropping can lead to the loss of key semantic elements, resulting in incorrect geolocation results. To address this problem, this paper designs a learnable image resizer and proposes an arbitrary-sized image geolocation method. (1) The designed learnable image resizer employs the self-attention mechanism to enhance the geometric features of the resized image. Firstly, it applies bilinear interpolation to the input image and its feature maps to obtain the initial resized image and the resized feature maps. Then, SKNet (selective kernel net) is used to approximate the best receptive field, thus keeping the geometric shapes as the original image. And SENet (squeeze and extraction net) is used to automatically select the feature maps with strong contour information, enhancing the geometric features. Finally, the enhanced geometric features are fused with the initial resized image, to obtain the final resized images. (2) The proposed image geolocalization method embeds the above image resizer as a fronting layer of the descriptor extraction network. It not only enables the network to be compatible with arbitrary-sized input images but also enhances the geometric features that are crucial to the image geolocalization task. Moreover, the triplet attention mechanism is added after the first convolutional layer of the backbone network to optimize the utilization of geometric elements extracted by the first convolutional layer. Finally, the local features extracted by the backbone network are aggregated to form image descriptors for image geolocalization. The proposed method was evaluated on several mainstream datasets, such as Pittsburgh30K, Tokyo24/7, and Places365. The results show that the proposed method has excellent size compatibility and compares favorably to recently mainstream geolocalization methods.

Keywords: image geolocalization, self-attention mechanism, image resizer, geometric feature

Procedia PDF Downloads 211

Authors: Aaditya U. Jhamb

Abstract:

Due to the current energy crisis that many countries are battling, energy-efficient buildings are the subject of extensive research in the modern technological era because of growing worries about energy consumption and its effects on the environment. The paper explores 8 factors that help determine energy efficiency for a building: (relative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, and glazing area distribution), with Tsanas and Xifara providing a dataset. The data set employed 768 different residential building models to anticipate heating and cooling loads with a low mean squared error. By optimizing these characteristics, machine learning algorithms may assess and properly forecast a building's heating and cooling loads, lowering energy usage while increasing the quality of people's lives. As a result, the paper studied the magnitude of the correlation between these input factors and the two output variables using various statistical methods of analysis after determining which input variable was most closely associated with the output loads. The most conclusive model was the Decision Tree Regressor, which had a mean squared error of 0.258, whilst the least definitive model was the Isotonic Regressor, which had a mean squared error of 21.68. This paper also investigated the KNN Regressor and the Linear Regression, which had to mean squared errors of 3.349 and 18.141, respectively. In conclusion, the model, given the 8 input variables, was able to predict the heating and cooling loads of a residential building accurately and precisely.

Keywords: energy efficient buildings, heating load, cooling load, machine learning models

Procedia PDF Downloads 90

Authors: Cherifi Mehdi, Lahdir Mourad, Ameur Soltane

Abstract:

Image compression is used to reduce the number of bits required to represent an image. The Meteosat Second Generation satellite (MSG) allows the acquisition of 12 image files every 15 minutes. Which results a large databases sizes. The transform selected in the images compression should contribute to reduce the data representing the images. The Radon transform retrieves the Radon points that represent the sum of the pixels in a given angle for each direction. Linear predictive coding (LPC) with filtering provides a good decorrelation of Radon points using a Predictor constitute by the Symmetric Nearest Neighbor filter (SNN) coefficients, which result losses during decompression. Finally, Run Length Coding (RLC) gives us a high and fixed compression ratio regardless of the input image. In this paper, a novel image compression method based on the Radon transform and linear predictive coding (LPC) for MSG images is proposed. MSG image compression based on the Radon transform and the LPC provides a good compromise between compression and quality of reconstruction. A comparison of our method with other whose two based on DCT and one on DWT bi-orthogonal filtering is evaluated to show the power of the Radon transform in its resistibility against the quantization noise and to evaluate the performance of our method. Evaluation criteria like PSNR and the compression ratio allows showing the efficiency of our method of compression.

Keywords: image compression, radon transform, linear predictive coding (LPC), run lengthcoding (RLC), meteosat second generation (MSG)

Procedia PDF Downloads 416

Authors: Justine Ugochukwu Osuagwu, Sazali Abd Wahab

Abstract:

The performance and operation of NNPC have faced series of attacks by all stakeholders as many have observed lots of inefficiency not only on the part of the management but the staff. This has raised questions of whether their operations and performance are being seriously affected by lack of transformational leadership, and the political competition prevalent in the country. The author has applied the administrative leadership theory and institutional theory as a guide to this study and empirically relates such theories to the study. The study also has utilized the quantitative approach where questionnaires were distributed to 370 participants, and the correctly filled and returned questionnaires were used for the analysis using structural equation modeling. The path coefficient of transformational leadership to performance is strong and positive with β = 0.672; t-value = 14.245; p-value = 0.000. Also, the result found that political competition does not mediate the relationship between transformational leadership and performance of NNPC. (β = -0.008; t-value = -0.600; p- value > 0.05). However, the indirect path is all insignificant, meaning that transformational leadership has relationship with corporate performance.The study found that,while political competition does not serve as a mediator in the relationship between transformational leadership and corporate performance, these styles of leadership have a direct and positive impact on corporate performance. The direct relationship between transformational leadership and political competition was not discovered, despite the fact that political competition has a direct and significant impact, both positive and negative, on corporate performance. As a result, both political competition and transformational leadership have the potential to significantly alter corporate performance.

Keywords: performance, transformational leadership, political competition, corporation performance, Nigeria national petroleum corporation

Procedia PDF Downloads 104

Authors: Adel Salem Bahakeem, Ahmad Jamal, Mir Md. Maruf Morshed, Elwaleed Awad Khidir

Abstract:

Direct Current (DC) servo motors, or simply DC motors, play an important role in many industrial applications such as manufacturing of plastics, precise positioning of the equipment, and operating computer-controlled systems where speed of feed control, maintaining the position, and ensuring to have a constantly desired output is very critical. These parameters can be controlled with the help of control systems such as the Proportional Integral Derivative (PID) controller. The aim of the current work is to investigate the effects of Proportional (P) and Integral (I) controllers on the steady state and transient response of the DC motor. The controller gains are varied to observe their effects on the error, damping, and stability of the steady and transient motor response. The current investigation is conducted experimentally on a servo trainer CE 110 using analog PI controller CE 120 and theoretically using Simulink in MATLAB. Both experimental and theoretical work involves varying integral controller gain to obtain the response to a steady-state input, varying, individually, the proportional and integral controller gains to obtain the response to a step input function at a certain frequency, and theoretically obtaining the proportional and integral controller gains for desired values of damping ratio and response frequency. Results reveal that a proportional controller helps reduce the steady-state and transient error between the input signal and output response and makes the system more stable. In addition, it also speeds up the response of the system. On the other hand, the integral controller eliminates the error but tends to make the system unstable with induced oscillations and slow response to eliminate the error. From the current work, it is desired to achieve a stable response of the servo motor in terms of its angular velocity subjected to steady-state and transient input signals by utilizing the strengths of both P and I controllers.

Keywords: DC servo motor, proportional controller, integral controller, controller gain optimization, Simulink

Procedia PDF Downloads 103

Authors: Jiahe Liu, Hongyang Yu, Miao Luo, Feng Qian

Abstract:

The widespread application of 3D human body reconstruction spans various fields. Smplify-X, an algorithm reliant on single-image input, employs three distinct body parameter templates, necessitating gender classification of individuals within the input image. Researchers employed a ResNet18 network to train a gender classifier within the Smplify-X framework, setting the threshold at 0.9, designating images falling below this threshold as having neutral gender. This model achieved 62.38% accurate predictions and 7.54% incorrect predictions. Our improvement involved refining the MobileNet network, resulting in a raised threshold of 0.97. Consequently, we attained 78.89% accurate predictions and a mere 0.2% incorrect predictions, markedly enhancing prediction precision and enabling more precise 3D human body reconstruction.

Keywords: SMPLX, mobileNet, gender classification, 3D human reconstruction

Procedia PDF Downloads 88

Authors: Tugrul Oktay, Mehmet Konar, Mustafa Soylak, Firat Sal, Murat Onay, Orhan Kizilkaya

Abstract:

In this paper, autonomous performance of a small manufactured unmanned helicopter is tried to be increased. For this purpose, a small unmanned helicopter is manufactured in Erciyes University, Faculty of Aeronautics and Astronautics. It is called as ZANKA-Heli-I. For performance maximization, autopilot parameters are determined via minimizing a cost function consisting of flight performance parameters such as settling time, rise time, overshoot during trajectory tracking. For this purpose, a stochastic optimization method named as simultaneous perturbation stochastic approximation is benefited. Using this approach, considerable autonomous performance increase (around %23) is obtained.

Keywords: small helicopters, hierarchical control, stochastic optimization, autonomous performance maximization, autopilots

Procedia PDF Downloads 577

Authors: Zheming Cao

Abstract:

Control loops are the most important components in the wind turbine system. Product quality, operation safety, and the economic performance are directly or indirectly connected to the performance of control systems. This paper proposed a performance evaluation method based on minimum-variance for wind turbine control system. This method can be applied on PID controller for pitch control system in the wind turbine. The good performance result demonstrated in the paper was achieved by retuning and optimizing the controller settings based on the evaluation result. The concepts presented in this paper are illustrated with the actual data of the industrial wind farm.

Keywords: control performance, evaluation, minimum-variance, wind turbine

Procedia PDF Downloads 365

Authors: Maryam Sadeghi, Claude Bazin, Daniel Hodouin, Laura Perez Barnuevo

Abstract:

The performance of a separation process is usually evaluated using performance indices calculated from elemental assays readily available from the chemical analysis laboratory. However, the separation process performance is essentially related to the properties of the minerals that carry the elements and not those of the elements. Since elements or metals can be carried by valuable and gangue minerals in the ore and that each mineral responds differently to a mineral processing method, the use of only elemental assays could lead to erroneous or uncertain conclusions on the process performance. This paper discusses the advantages of using performance indices calculated from minerals content, such as minerals recovery, for process performance assessments. A method is presented that uses elemental assays to estimate the minerals content of the solids in various process streams. The method combines the stoichiometric composition of the minerals and constraints of mass conservation for the minerals through the concentration process to estimate the minerals content from elemental assays. The advantage of assessing a concentration process using mineral based performance indices is illustrated for an iron ore concentration circuit.

Keywords: data reconciliation, iron ore concentration, mineral composition, process performance assessment

Procedia PDF Downloads 211

Authors: Jessica Lopez Espejel, Mahaman Sanoussi Yahaya Alassan, Walid Dahhane, El Hassane Ettifouri

Abstract:

Pretrained transformer-based models have shown high performance in natural language generation tasks. However, a new wave of interest has surged: automatic programming language code generation. This task consists of translating natural language instructions to a source code. Despite the fact that well-known pre-trained models on language generation have achieved good performance in learning programming languages, effort is still needed in automatic code generation. In this paper, we introduce JaCoText, a model based on Transformer neural network. It aims to generate java source code from natural language text. JaCoText leverages the advantages of both natural language and code generation models. More specifically, we study some findings from state of the art and use them to (1) initialize our model from powerful pre-trained models, (2) explore additional pretraining on our java dataset, (3) lead experiments combining the unimodal and bimodal data in training, and (4) scale the input and output length during the fine-tuning of the model. Conducted experiments on CONCODE dataset show that JaCoText achieves new state-of-the-art results.

Keywords: java code generation, natural language processing, sequence-to-sequence models, transformer neural networks

Procedia PDF Downloads 276

Authors: Bipasha Sen, Aditya Agarwal

Abstract:

Multilingual automatic speech recognition (ASR) system is a single entity capable of transcribing multiple languages sharing a common phone space. Performance of such a system is highly dependent on the compatibility of the languages. State of the art speech recognition systems are built using sequential architectures based on recurrent neural networks (RNN) limiting the computational parallelization in training. This poses a significant challenge in terms of time taken to bootstrap and validate the compatibility of multiple languages for building a robust multilingual system. Complex architectural choices based on self-attention networks are made to improve the parallelization thereby reducing the training time. In this work, we propose Reed, a simple system based on 1D convolutions which uses very short context to improve the training time. To improve the performance of our system, we use raw time-domain speech signals directly as input. This enables the convolutional layers to learn feature representations rather than relying on handcrafted features such as MFCC. We report improvement on training and inference times by atleast a factor of 4x and 7.4x respectively with comparable WERs against standard RNN based baseline systems on SpeechOcean's multilingual low resource dataset.

Keywords: convolutional neural networks, language compatibility, low resource languages, multilingual automatic speech recognition

Procedia PDF Downloads 120

Authors: B. R. Campomanes-Alvarez, P. Quiros, B. Fernandez

Abstract:

Automatic Speech Recognition (ASR) is a machine-based process of decoding and transcribing oral speech. A typical ASR system receives acoustic input from a speaker or an audio file, analyzes it using algorithms, and produces an output in the form of a text. Some speech recognition systems use Hidden Markov Models (HMMs) to deal with the temporal variability of speech and Gaussian Mixture Models (GMMs) to determine how well each state of each HMM fits a short window of frames of coefficients that represents the acoustic input. Another way to evaluate the fit is to use a feed-forward neural network that takes several frames of coefficients as input and produces posterior probabilities over HMM states as output. Deep neural networks (DNNs) that have many hidden layers and are trained using new methods have been shown to outperform GMMs on a variety of speech recognition systems. Acoustic models for state-of-the-art ASR systems are usually training on massive amounts of data. However, audio files with their corresponding transcriptions can be difficult to obtain, especially in the Spanish language. Hence, in the case of these low-resource scenarios, building an ASR model is considered as a complex task due to the lack of labeled data, resulting in an under-trained system. Semi-supervised learning approaches arise as necessary tasks given the high cost of transcribing audio data. The main goal of this proposal is to develop a procedure based on acoustic semi-supervised learning for Spanish ASR systems by using DNNs. This semi-supervised learning approach consists of: (a) Training a seed ASR model with a DNN using a set of audios and their respective transcriptions. A DNN with a one-hidden-layer network was initialized; increasing the number of hidden layers in training, to a five. A refinement, which consisted of the weight matrix plus bias term and a Stochastic Gradient Descent (SGD) training were also performed. The objective function was the cross-entropy criterion. (b) Decoding/testing a set of unlabeled data with the obtained seed model. (c) Selecting a suitable subset of the validated data to retrain the seed model, thereby improving its performance on the target test set. To choose the most precise transcriptions, three confidence scores or metrics, regarding the lattice concept (based on the graph cost, the acoustic cost and a combination of both), was performed as selection technique. The performance of the ASR system will be calculated by means of the Word Error Rate (WER). The test dataset was renewed in order to extract the new transcriptions added to the training dataset. Some experiments were carried out in order to select the best ASR results. A comparison between a GMM-based model without retraining and the DNN proposed system was also made under the same conditions. Results showed that the semi-supervised ASR-model based on DNNs outperformed the GMM-model, in terms of WER, in all tested cases. The best result obtained an improvement of 6% relative WER. Hence, these promising results suggest that the proposed technique could be suitable for building ASR models in low-resource environments.

Keywords: automatic speech recognition, deep neural networks, machine learning, semi-supervised learning

Procedia PDF Downloads 337

Authors: Mathias U. Bonet, Pericles Pilidis, Georgios Doulgeris

Abstract:

An aero-derivative marine Gas Turbine engine model is simulated to be installed as the main propulsion prime mover to power a cruise ship which is designed and routed to transport intending Muslim pilgrims for the annual hajj pilgrimage from Nigeria to the Islamic port city of Jeddah in Saudi Arabia. A performance assessment of the Gas Turbine engine has been conducted by examining the effect of varying aerodynamic and hydrodynamic conditions encountered at various geographical locations along the scheduled transit route during the voyage. The investigation focuses on the overall behavior of the Gas Turbine engine employed to power and propel the ship as it operates under ideal and adverse conditions to be encountered during calm and rough weather according to the different seasons of the year under which the voyage may be undertaken. The variation of engine performance under varying operating conditions has been considered as a very important economic issue by determining the time the speed by which the journey is completed as well as the quantity of fuel required for undertaking the voyage. The assessment also focuses on the increased resistance caused by the fouling of the submerged portion of the ship hull surface with its resultant effect on the power output of the engine as well as the overall performance of the propulsion system. Daily ambient temperature levels were obtained by accessing data from the UK Meteorological Office while the varying degree of turbulence along the transit route and according to the Beaufort scale were also obtained as major input variables of the investigation. By assuming the ship to be navigating the Atlantic Ocean and the Mediterranean Sea during winter, spring and summer seasons, the performance modeling and simulation was accomplished through the use of an integrated Gas Turbine performance simulation code known as ‘Turbomach’ along with a Matlab generated code named ‘Poseidon’, all of which have been developed at the Power and Propulsion Department of Cranfield University. As a case study, the results of the various assumptions have further revealed that the marine Gas Turbine is a reliable and available alternative to the conventional marine propulsion prime movers that have dominated the maritime industry before now. The techno-economic and environmental assessment of this type of propulsion prime mover has enabled the determination of the effect of changes in weather and sea conditions on the ship speed as well as trip time and the quantity of fuel required to be burned throughout the voyage.

Keywords: ambient temperature, hull fouling, marine gas turbine, performance, propulsion, voyage

Procedia PDF Downloads 184

Authors: N. H. Othman, N. Udin, M. Ishak, L. H. Shah

Abstract:

This study focuses on the effect of pin taper tool ratio on friction stir welding of magnesium alloy AZ31. Two pieces of AZ31 alloy with thickness of 6 mm were friction stir welded by using the conventional milling machine. The shoulder diameter used in this experiment is fixed at 18 mm. The taper pin ratio used are varied at 6:6, 6:5, 6:4, 6:3, 6:2 and 6:1. The rotational speeds that were used in this study were 500 rpm, 1000 rpm and 1500 rpm, respectively. The welding speeds used are 150 mm/min, 200 mm/min and 250 mm/min. Microstructure observation of welded area was studied by using optical microscope. Equiaxed grains were observed at the TMAZ and stir zone indicating fully plastic deformation. Tool pin diameter ratio 6/1 causes low heat input to the material because of small contact surface between tool surface and stirred materials compared to other tool pin diameter ratio. The grain size of stir zone increased with increasing of ratio of rotational speed to transverse speed due to higher heat input. It is observed that worm hole is produced when excessive heat input is applied. To evaluate the mechanical properties of this specimen, tensile test was used in this study. Welded specimens using taper pin ratio 6:1 shows higher tensile strength compared to other taper pin ratio up to 204 MPa. Moreover, specimens using taper pin ratio 6:1 showed better tensile strength with 500 rpm of rotational speed and 150mm/min welding speed.

Keywords: friction stir welding, magnesium AZ31, cylindrical taper tool, taper pin ratio

Procedia PDF Downloads 283

Authors: Apollinaire Barme, Simon Tamayo, Arthur Gaudron

Abstract:

This paper presents a statistical approach to identify explanatory variables linearly related to e-commerce sales. The proposed methodology allows specifying a regression model in order to quantify the relevance between openly available data (economic and demographic) and national e-commerce sales. The proposed methodology consists in collecting data, preselecting input variables, performing regressions for choosing variables and models, testing and validating. The usefulness of the proposed approach is twofold: on the one hand, it allows identifying the variables that influence e- commerce sales with an accessible approach. And on the other hand, it can be used to model future sales from the input variables. Results show that e-commerce is linearly dependent on 11 economic and demographic indicators.

Keywords: e-commerce, statistical modeling, regression, empirical research

Procedia PDF Downloads 221

Authors: Hassan Masoumi, Ahad Salimi, Nazanin Barhemmat, Babak Gholami

Abstract:

Artificial neural networks have gained a lot of interest as empirical models for their powerful representational capacity, multi input and output mapping characteristics. In fact, most feed-forward networks with nonlinear nodal functions have been proved to be universal approximates. In this paper, we propose a new supervised method for color image classification based on self organizing feature maps (SOFM). This algorithm is based on competitive learning. The method partitions the input space using self-organizing feature maps to introduce the concept of local neighborhoods. Our image classification system entered into RGB image. Experiments with simulated data showed that separability of classes increased when increasing training time. In additional, the result shows proposed algorithms are effective for color image classification.

Keywords: classification, SOFM algorithm, neural network, neighborhood, RGB image

Procedia PDF Downloads 473

Authors: Olomon Wu, Hung Lu, Nick Wilkins, Daniel Kerr, Zekeriya Aliyazicioglu, H. K. Hwang

Abstract:

Multiple Input Multiple Output (MIMO) radar has received increasing attention in recent years. MIMO radar has many advantages over conventional phased array radar such as target detection, resolution enhancement, and interference suppression. In this paper, the results are presented from a simulation study of MIMO Uniformly-Spaced Linear Array (ULA) antennas. The performance is investigated under varied parameters, including varied array size, Pseudo Random (PN) sequence length, number of snapshots, and Signal to Noise Ratio (SNR). The results of MIMO are compared to a traditional array antenna.

Keywords: MIMO radar, phased array antenna, target detection, radar signal processing

Procedia PDF Downloads 538

Authors: Branci Sarra, Branci Saadia

Abstract:

Because of the multiplied data growth, many computer science tools have been developed to process and analyze these Big Data. High-performance computing architectures have been designed to meet the treatment needs of Big Data (view transaction processing standpoint, strategic, and tactical analytics). The purpose of this article is to provide a historical and global perspective on the recent trend of high-performance computing architectures especially what has a relation with Analytics and Data Mining.

Keywords: high performance computing, HPC, big data, data analysis

Procedia PDF Downloads 514

Authors: M. Jebali, M. Jemni

Abstract:

This work is the part of vision based hand gesture recognition system for Natural Human Computer Interface. Hand tracking and segmentation are the primary steps for any hand gesture recognition system. The aim of this paper is to develop robust and efficient hand segmentation algorithm such as an input to another system which attempt to bring the HCI performance nearby the human-human interaction, by modeling an intelligent sign language recognition system based on prediction in the context of dialogue between the system (avatar) and the interlocutor. For the purpose of hand segmentation, an overcoming occlusion approach has been proposed for superior results for detection of hand from an image.

Keywords: HCI, sign language recognition, object tracking, hand segmentation

Procedia PDF Downloads 405

Authors: Emmanuel Barrantes Chaves, Ernesto Rivera Alvarado

Abstract:

This paper presents a method to enable agents for the Terraria game to evaluate algorithms commonly used in general video game artificial intelligence competitions. The usage of the ‘You Only Look Once’ model in the first layer of the process obtains information from the screen, translating this information into a video game description language known as “Video Game Description Language”; the agents take that as input to make decisions. For this, the state-of-the-art algorithms were tested and compared; Monte Carlo Tree Search and Rolling Horizon Evolutionary; in this case, Rolling Horizon Evolutionary shows a better performance. This approach’s main advantage is that a VGDL beforehand is unnecessary. It will be built on the fly and opens the road for using more games as a framework for AI.

Keywords: AI, MCTS, RHEA, Terraria, VGDL, YOLOv5

Procedia PDF Downloads 92

Authors: Seigo Matsuno, Yasuo Uchida, Shozo Tokinaga

Abstract:

This paper proposes and empirically tests a model of the relationships between green supply chain management (GSCM) activities and corporate performance. From the literature review, we identified five constructs, namely, environmental commitment, supplier collaboration, supplier assessment, information sharing among suppliers, and business process improvement. These explanatory variables are used to form a structural model explaining the environmental and economic performance. The model was analyzed using the data from a survey of a sample of manufacturing firms in Japan. The results suggest that the degree of supplier collaboration has an influence on the environmental performance directly. While, the impact of supplier assessment on the environmental performance is mediated by the information sharing and/or business process improvement. And the environmental performance has a positive relationship on the economic performance. Academic and managerial implications of our findings are discussed.

Keywords: corporate performance, empirical study, green supply chain management, path modeling

Procedia PDF Downloads 389