Search results for: hyper-parameter tuning

Authors: Bharatendra Rai

Abstract:

The sequence of words in text data has long-term dependencies and is known to suffer from vanishing gradient problems when developing deep learning models. Although recurrent networks such as long short-term memory networks help to overcome this problem, achieving high text classification performance is a challenging problem. Convolutional recurrent networks that combine the advantages of long short-term memory networks and convolutional neural networks can be useful for text classification performance improvements. However, arriving at suitable hyperparameter values for convolutional recurrent networks is still a challenging task where fitting a model requires significant computing resources. This paper illustrates the advantages of using convolutional recurrent networks for text classification with the help of statistically planned computer experiments for hyperparameter tuning.

Keywords: long short-term memory networks, convolutional recurrent networks, text classification, hyperparameter tuning, Tukey honest significant differences

Procedia PDF Downloads 128

Authors: Farzad Hosseini Hossein Abadi, Cristina Prieto Sierra, Cesar Álvarez Díaz

Abstract:

Predicting streamflow accurately in flashy catchments prone to floods is a major research and operational challenge in hydrological modeling. Recent advancements in deep learning, particularly Long Short-Term Memory (LSTM) networks, have shown to be promising in achieving accurate hydrological predictions at daily and hourly time scales. In this work, a multi-timescale LSTM (MTS-LSTM) network was applied to the context of regional hydrological predictions at an hourly time scale in flashy catchments. The case study includes 40 catchments allocated in the Basque Country, north of Spain. We explore the impact of hyperparameters on the performance of streamflow predictions given by regional deep learning models through systematic hyperparameter tuning - where optimal regional values for different catchments are identified. The results show that predictions are highly accurate, with Nash-Sutcliffe (NSE) and Kling-Gupta (KGE) metrics values as high as 0.98 and 0.97, respectively. A principal component analysis reveals that a hyperparameter related to the length of the input sequence contributes most significantly to the prediction performance. The findings suggest that input sequence lengths have a crucial impact on the model prediction performance. Moreover, employing catchment-scale analysis reveals distinct sequence lengths for individual basins, highlighting the necessity of customizing this hyperparameter based on each catchment’s characteristics. This aligns with well known “uniqueness of the place” paradigm. In prior research, tuning the length of the input sequence of LSTMs has received limited focus in the field of streamflow prediction. Initially it was set to 365 days to capture a full annual water cycle. Later, performing limited systematic hyper-tuning using grid search, revealed a modification to 270 days. However, despite the significance of this hyperparameter in hydrological predictions, usually studies have overlooked its tuning and fixed it to 365 days. This study, employing a simultaneous systematic hyperparameter tuning approach, emphasizes the critical role of input sequence length as an influential hyperparameter in configuring LSTMs for regional streamflow prediction. Proper tuning of this hyperparameter is essential for achieving accurate hourly predictions using deep learning models.

Keywords: LSTMs, streamflow, hyperparameters, hydrology

Procedia PDF Downloads 69

Authors: Donya Ashtiani Haghighi, Amirali Baniasadi

Abstract:

Coronavirus (COVID-19) disease has spread abruptly all over the world since the end of 2019. Computed tomography (CT) scans and X-ray images are used to detect this disease. Different Deep Neural Network (DNN)-based diagnosis solutions have been developed, mainly based on Convolutional Neural Networks (CNNs), to accelerate the identification of COVID-19 cases. However, CNNs lose important information in intermediate layers and require large datasets. In this paper, Capsule Network (CapsNet) is used. Capsule Network performs better than CNNs for small datasets. Accuracy of 0.9885, f1-score of 0.9883, precision of 0.9859, recall of 0.9908, and Area Under the Curve (AUC) of 0.9948 are achieved on the Capsule-based framework with hyperparameter tuning. Moreover, different dropout rates are investigated to decrease overfitting. Accordingly, a dropout rate of 0.1 shows the best results. Finally, we remove one convolution layer and decrease the number of trainable parameters to 146,752, which is a promising result.

Keywords: capsule network, dropout, hyperparameter tuning, classification

Procedia PDF Downloads 76

Authors: Songul Cinaroglu

Abstract:

Machine learning regression methods are recommended as an alternative to classical regression methods in the existence of variables which are difficult to model. Data for health expenditure is typically non-normal and have a heavily skewed distribution. This study aims to compare machine learning regression methods by hyperparameter tuning to predict health expenditure per capita. A multiple regression model was conducted and performance results of Lasso Regression, Random Forest Regression and Support Vector Machine Regression recorded when different hyperparameters are assigned. Lambda (λ) value for Lasso Regression, number of trees for Random Forest Regression, epsilon (ε) value for Support Vector Regression was determined as hyperparameters. Study results performed by using 'k' fold cross validation changed from 5 to 50, indicate the difference between machine learning regression results in terms of R², RMSE and MAE values that are statistically significant (p < 0.001). Study results reveal that Random Forest Regression (R² ˃ 0.7500, RMSE ≤ 0.6000 ve MAE ≤ 0.4000) outperforms other machine learning regression methods. It is highly advisable to use machine learning regression methods for modelling health expenditures.

Keywords: machine learning, lasso regression, random forest regression, support vector regression, hyperparameter tuning, health expenditure

Procedia PDF Downloads 225

Authors: Shengbiao Wu, Weihua Cao, Min Wu, Can Liu

Abstract:

This paper presents an intelligent tuning method of microwave filter based on complex neural network and improved space mapping. The tuning process consists of two stages: the initial tuning and the fine tuning. At the beginning of the tuning, the return loss of the filter is transferred to the passband via the error of phase. During the fine tuning, the phase shift caused by the transmission line and the higher order mode is removed by the curve fitting. Then, an Cauchy method based on the admittance parameter (Y-parameter) is used to extract the coupling matrix. The influence of the resonant cavity loss is eliminated during the parameter extraction process. By using processed data pairs (the amount of screw variation and the variation of the coupling matrix), a tuning model is established by the complex neural network. In view of the improved space mapping algorithm, the mapping relationship between the actual model and the ideal model is established, and the amplitude and direction of the tuning is constantly updated. Finally, the tuning experiment of the eight order coaxial cavity filter shows that the proposed method has a good effect in tuning time and tuning precision.

Keywords: microwave filter, scattering parameter, coupling matrix, intelligent tuning

Procedia PDF Downloads 310

Authors: Tai-Hsing Lee, Zhe-Wei Lin

Abstract:

This paper demonstrates an ultra-wide tuning VCO implemented by CMOS 0.18μm process technology. By employing the proposed technique of magnetic energy adjustment in the oscillator tank, our proposed VCO achieves a wide frequency tuning range of 69.46% from 0.9 GHz to 1.86 GHz. The phase noise at an operating frequency of 1.86 GHz is -110 dBc/Hz (Offset frequency=1MHz). Furthermore, it achieves an excellent FOMT of 190.03 dBc/Hz.

Keywords: VCO, Ultra-wide tuning, Frequency tuning range, phase noise, Magnetic energy adjustment

Procedia PDF Downloads 38

Authors: Mehdi Shahbazian, Alireza Aarabi, Mohsen Hadiyan

Abstract:

Proportional-Integral-Derivative (PID) controllers are the most widely used controllers in industry because of its simplicity and robustness. Different values of PID parameters make different step response, so an increasing amount of literature is devoted to proper tuning of PID controllers. The problem merits further investigation as traditional tuning methods make large control signal that can damages the system but using evolutionary algorithms based tuning methods improve the control signal and closed loop performance. In this paper three tuning methods for PID controllers have been studied namely Ziegler and Nichols, which is traditional tuning method and evolutionary algorithms based tuning methods, that are, Genetic algorithm and particle swarm optimization. To examine the validity of PSO and GA tuning methods a comparative analysis of DC motor plant is studied. Simulation results reveal that evolutionary algorithms based tuning method have improved control signal amplitude and quality factors of the closed loop system such as rise time, integral absolute error (IAE) and maximum overshoot.

Keywords: evolutionary algorithm, genetic algorithm, particle swarm optimization, PID controller

Procedia PDF Downloads 480

Authors: Rabia Korkmaz Tan, Şebnem Bora

Abstract:

The major problem encountered when modeling complex systems with agent-based modeling and simulation techniques is the existence of large parameter spaces. A complex system model cannot be expected to reflect the whole of the real system, but by specifying the most appropriate parameters, the actual system can be represented by the model under certain conditions. When the studies conducted in recent years were reviewed, it has been observed that there are few studies for parameter tuning problem in agent based simulations, and these studies have focused on tuning parameters of a single model. In this study, an approach of parameter tuning is proposed by using metaheuristic algorithms such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Artificial Bee Colonies (ABC), Firefly (FA) algorithms. With this hybrid structured study, the parameter tuning problems of the models in the different fields were solved. The new approach offered was tested in two different models, and its achievements in different problems were compared. The simulations and the results reveal that this proposed study is better than the existing parameter tuning studies.

Keywords: parameter tuning, agent based modeling and simulation, metaheuristic algorithms, complex systems

Procedia PDF Downloads 225

Authors: Hooman Kaabi, Raziyeh Karkoub

Abstract:

This work proposes a structure of AMOS-varactors. A 5GHz LC-VCO designed in TSMC 0.18μm CMOS to improve phase noise and tuning range performance. The tuning range is from 5.05GHZ to 5.88GHz.The phase noise is -154.9dBc/Hz at 1MHz offset from the carrier. It meets the requirements for IEEE 802.11a WLAN standard.

Keywords: CMOS LC VCO, spiral inductor, varactor, phase noise, tuning range

Procedia PDF Downloads 534

Authors: Deepak M. Sajnekar, S. B. Deshpande, R. M. Mohril

Abstract:

At present many rotary excitation control system are using analog type of Automatic Voltage Regulator which now started to replace with the digital automatic voltage regulator which is provided with PID controller and tuning of PID controller is a challenging task. The cases where digital excitation control system is used tuning of PID controller are still carried out by pole placement method. Tuning of PID controller used for static excitation control system is not challenging because it does not involve exciter time constant. This paper discusses two methods of tuning PID controller i.e. Pole placement method and pole zero cancellation method. GUI prepared for both the methods on the platform of MATLAB. Using this GUI, performance results and time required for tuning for both the methods are compared. Sensitivity of the methods is also presented with parameter variation like loop gain ‘K’ and exciter time constant ‘te’.

Keywords: digital excitation system, automatic voltage regulator, pole placement method, pole zero cancellation method

Procedia PDF Downloads 677

Authors: Arnaud Nougues

Abstract:

This paper describes a two-stage methodology derived from internal model control (IMC) for tuning a proportional-integral-derivative (PID) controller for levels or other integrating processes in an industrial environment. Focus is the ease of use and implementation speed which are critical for an industrial application. Tuning can be done with minimum effort and without the need for time-consuming open-loop step tests on the plant. The first stage of the method applies to levels only: the vessel residence time is calculated from equipment dimensions and used to derive a set of preliminary proportional-integral (PI) settings with IMC. The second stage, re-tuning in closed-loop, applies to levels as well as other integrating processes: a tuning correction mechanism has been developed based on a series of closed-loop simulations with model errors. The tuning correction is done from a simple closed-loop step test and the application of a generic correlation between observed overshoot and integral time correction. A spin-off of the method is that an estimate of the vessel residence time (levels) or open-loop process gain (other integrating process) is obtained from the closed-loop data.

Keywords: closed-loop model identification, IMC-PID tuning method, integrating process control, on-line PID tuning adaptation

Procedia PDF Downloads 220

Authors: Shoeb Ahmed Adeel, Vivek Paul, K. Prajwal, Michael Fenelon

Abstract:

Techniques have been used to tether and stabilize a multi-rotor MAV but carrying out the same process to a fixed wing MAV is a novel method which can be utilized in order to reduce damage occurring to the fixed wing MAVs while conducting flight test trials and PID tuning. A few sensors and on board controller is required to carry out this experiment in horizontal and vertical plane of the vehicle. Here we will be discussing issues such as sensitivity of the air vehicle, endurance and external load of the string acting on the vehicle.

Keywords: MAV, PID tuning, tethered flight, UAV

Procedia PDF Downloads 634

Authors: Mathias Marquardt, Peter Dünow, Sandra Baßler

Abstract:

The paper describes the use of subspace based identification methods for auto tuning of a state space control system. The plant is an unstable but self balancing transport robot. Because of the unstable character of the process it has to be identified from closed loop input-output data. Based on the identified model a state space controller combined with an observer is calculated. The subspace identification algorithm and the controller design procedure is combined to a auto tuning method. The capability of the approach was verified in a simulation experiments under different process conditions.

Keywords: auto tuning, balanced robot, closed loop identification, subspace identification

Procedia PDF Downloads 379

Authors: Isha Rathore, Peeyush Jain, Elangovan Rajasekar

Abstract:

The thermal capacity of the envelope impacts the cooling and heating demand of a building and modulates the peak electricity demand. This paper presents the thermal capacity tuning of a building envelope to minimize peak electricity demand for space cooling. We consider a 40 m² residential testbed located in Hyderabad, India (Composite Climate). An EnergyPlus model is validated using real-time data. A Parametric simulation framework for thermal capacity tuning is created using the Honeybee plugin. Diffusivity, Thickness, layer position, orientation and fenestration size of the exterior envelope are parametrized considering a five-layered wall system. A total of 1824 parametric runs are performed and the optimum wall configuration leading to minimum peak cooling demand is presented.

Keywords: thermal capacity, tuning, peak demand reduction, parametric analysis

Procedia PDF Downloads 183

Authors: Adnan Baig, Ishteeaq Naeem, Saad Mansoor

Abstract:

The Internet of Things (IoT) gadgets have advanced quickly in recent years, and their use is steadily rising daily. However, cyber-attackers can target these gadgets due to their distributed nature. Additionally, many IoT devices have significant security flaws in their implementation and design, making them vulnerable to security threats. Hence, these threats can cause important data security and privacy loss from a single attack on network devices or systems. Botnets are a significant security risk that can harm the IoT network; hence, sophisticated techniques are required to mitigate the risk. This work uses a machine learning-based method to identify IoT orchestrated by botnets. The proposed technique identifies the net attack by distinguishing between legitimate and malicious traffic. This article proposes a hyperparameter tuning model to improvise the method to improve the accuracy of existing processes. The results demonstrated an improved and more accurate indication of botnet-based cyber-attacks.

Keywords: Internet of Things, Botnet, BoT-IoT dataset, ML techniques

Procedia PDF Downloads 10

Authors: Saeid Jalilzadeh

Abstract:

PID controllers are widely used to control the industrial plants because of their robustness and simple structures. Tuning of the controller's parameters to get a desired response is difficult and time consuming. With the development of computer technology and artificial intelligence in automatic control field, all kinds of parameters tuning methods of PID controller have emerged in endlessly, which bring much energy for the study of PID controller, but many advanced tuning methods behave not so perfect as to be expected. Honey Bee algorithm (HBA) and genetic algorithm (GA) are extensively used for real parameter optimization in diverse fields of study. This paper describes an application of HBA and GA to the problem of designing a PID controller whose parameters comprise proportionality constant, integral constant and derivative constant. Presence of three parameters to optimize makes the task of designing a PID controller more challenging than conventional P, PI, and PD controllers design. The suitability of the proposed approach has been demonstrated through computer simulation using MATLAB/SIMULINK.

Keywords: controller, GA, optimization, PID, PSO

Procedia PDF Downloads 543

Authors: Yuanjin Liu

Abstract:

Retirement withdrawal strategies are very important to minimize the probability of ruin in retirement. The ruin probability is modeled as a function of initial withdrawal age, gender, asset allocation, inflation rate, and initial withdrawal rate. The ruin probability is obtained based on the 2019 period life table for the Social Security, IRS Required Minimum Distribution (RMD) Worksheets, US historical bond and equity returns, and inflation rates using simulation. Several popular machine learning algorithms of the generalized additive model, random forest, support vector machine, extreme gradient boosting, and artificial neural network are built. The model validation and selection are based on the test errors using hyperparameter tuning and train-test split. The optimal model is recommended for retirees to monitor the ruin probability. The optimal withdrawal strategy can be obtained based on the optimal predictive model.

Keywords: ruin probability, retirement withdrawal strategies, predictive models, optimal model

Procedia PDF Downloads 73

Authors: Muhammad Abdullah Arafat, Nahrin Nowrose

Abstract:

Wireless charging of electric vehicles is becoming more and more attractive as large amount of power can now be transferred to a reasonable distance using magnetic resonance coupling method. However, proper tuning of the compensation network is required to achieve maximum power transmission. Due to the variation of coil distance from the nominal value as a result of change in tire condition, change in weight or uneven road condition, the tuning of the compensation network has become challenging. In this paper, a tuning method has been described to determine the optimum values of the compensation network in order to maximize the average output power. The simulation results show that 5.2 percent increase in average output power is obtained for 10 percent variation in coupling coefficient using the optimum values without the need of additional space and electro-mechanical components. The proposed method is applicable to both static and dynamic charging of electric vehicles.

Keywords: coupling coefficient, electric vehicles, magnetic resonance coupling, tuning capacitor, wireless power transfer

Procedia PDF Downloads 194

Authors: W. T. A. Mohammed, M. Eltaeb, R. Kashani

Abstract:

The first bending modes of tall asymmetric structures in the two lateral X and Y-directions have two different natural frequencies. To add tuned damping to these bending modes, one needs to either a) use two pendulum-tuned mass dampers (PTMDs) with one tuning frequency, each PTMD targeting one of the bending modes, or b) use one PTMD with two tuning frequencies (one in each lateral directions). Option (a), being more massive, requiring more space, and being more expensive, is less attractive than option (b). Considering that the tuning frequency of a pendulum depends mainly on the pendulum length, one way of realizing option (b) is by constraining the swinging length of the pendulum in one direction but not in the other; such PTMD is dubbed passive Bi-PTMD. Alternatively, option (b) can be realized by actively setting the tuning frequencies of the PTMD in the two directions. In this work, accurate physical models of passive Bi-PTMD and active PTMD are developed and incorporated into the numerical model of a tall asymmetric structure. The model of PTMDs plus structure is used for a)synthesizing such PTMDs for particular applications and b)evaluating their damping effectiveness in mitigating the dynamic lateral responses of their target asymmetric structures, perturbed by wind load in X and Y-directions. Depending on how elaborate the control scheme is, the active PTMD can either be made to yield the same damping effectiveness as the passive Bi-PTMD of the same size or the passive Bi-TMD twice as massive as the active PTMD.

Keywords: active tuned mass damper, high-rise building, multi-frequency tuning, vibration control

Procedia PDF Downloads 104

Authors: Truong Nguyen Luan Vu, Le Hieu Giang, Le Linh

Abstract:

The fractional–order proportional integral (FOPI) controller tuning rules based on the fractional calculus for the cascade control system are systematically proposed in this paper. Accordingly, the ideal controller is obtained by using internal model control (IMC) approach for both the inner and outer loops, which gives the desired closed-loop responses. On the basis of the fractional calculus, the analytical tuning rules of FOPI controller for the inner loop can be established in the frequency domain. Besides, the outer loop is tuned by using any integer PI/PID controller tuning rules in the literature. The simulation study is considered for the stable process model and the results demonstrate the simplicity, flexibility, and effectiveness of the proposed method for the cascade control system in compared with the other methods.

Keywords: Bode’s ideal transfer function, fractional calculus, fractional–order proportional integral (FOPI) controller, cascade control system

Procedia PDF Downloads 376

Authors: Jenq-Shyong Chen, Ben-Fong Yu

Abstract:

CNC(computer numerical control) machining centers have been widely used for machining different metal components for various industries. For a specific CNC machine, its everyday job is assigned to cut different products with quite different attributes such as material type, workpiece weight, geometry, tooling, and cutting conditions. Theoretically, the dynamic characteristics of the CNC machine should be properly tuned match each machining job in order to get the optimal machining performance. However, most of the CNC machines are set with only a standard set of CNC parameters. In this study, we have developed an auto-tuning system which can automatically change the CNC parameters and in hence change the machine dynamic characteristics according to the selection of machining modes which are set by the mixed combination of three machine performance indexes: the HO (high surface quality) index, HP (high precision) index and HS (high speed) index. The acceleration, jerk, corner error tolerance, oscillation and dynamic bandwidth of machine’s feed axes have been changed according to the selection of the machine performance indexes. The proposed auto-tuning system of the CNC parameters has been implemented on a PC-based CNC controller and a three-axis machining center. The measured experimental result have shown the promising of our proposed auto-tuning system.

Keywords: auto-tuning, CNC parameters, machining mode, high speed, high accuracy, high surface quality

Procedia PDF Downloads 379

Authors: Maitree Thamma, Witchupong Wiboonjaroen, Thanat Suknuan, Karan Homchat

Abstract:

First order plus dead time (FOPDT) is a high dynamic system. Therefore, the controller must be intelligent. This paper presents the development and implementation of self-tuning Fuzzy-PID controller for controlling the FOPDT system. The water level process used represented FOPDT system and the mathematical model of the system was approximated by using System Identification toolbox in Matlab. The control programming and Fuzzy-PID algorithm used Matlab/Simulink and run on Microcontroller STM32.

Keywords: real-time control, self-tuning fuzzy-PID, FOPDT system, the water lever process

Procedia PDF Downloads 291

Authors: Yoonsuh Jung

Abstract:

As DNA microarray data contain relatively small sample size compared to the number of genes, high dimensional models are often employed. In high dimensional models, the selection of tuning parameter (or, penalty parameter) is often one of the crucial parts of the modeling. Cross-validation is one of the most common methods for the tuning parameter selection, which selects a parameter value with the smallest cross-validated score. However, selecting a single value as an "optimal" value for the parameter can be very unstable due to the sampling variation since the sample sizes of microarray data are often small. Our approach is to choose multiple candidates of tuning parameter first, then average the candidates with different weights depending on their performance. The additional step of estimating the weights and averaging the candidates rarely increase the computational cost, while it can considerably improve the traditional cross-validation. We show that the selected value from the suggested methods often lead to stable parameter selection as well as improved detection of significant genetic variables compared to the tradition cross-validation via real data and simulated data sets.

Keywords: cross validation, parameter averaging, parameter selection, regularization parameter search

Procedia PDF Downloads 414

Authors: Khairani Binti Supyan, Fatimah Khalid, Mas Rina Mustaffa, Azreen Bin Azman, Amirul Azuani Romle

Abstract:

Perennial plant classification plays a significant role in various agricultural and environmental applications, assisting in plant identification, disease detection, and biodiversity monitoring. Nevertheless, attaining high accuracy in perennial plant image classification remains challenging due to the complex variations in plant appearance, the diverse range of environmental conditions under which images are captured, and the inherent variability in image quality stemming from various factors such as lighting conditions, camera settings, and focus. This paper proposes an adaptation approach to optimize perennial plant image classification by fine-tuning the pre-trained DNNs model. This paper explores the efficacy of fine-tuning prevalent architectures, namely VGG16, ResNet50, and InceptionV3, leveraging transfer learning to tailor the models to the specific characteristics of perennial plant datasets. A subset of the MYLPHerbs dataset consisted of 6 perennial plant species of 13481 images under various environmental conditions that were used in the experiments. Different strategies for fine-tuning, including adjusting learning rates, training set sizes, data augmentation, and architectural modifications, were investigated. The experimental outcomes underscore the effectiveness of fine-tuning deep neural networks for perennial plant image classification, with ResNet50 showcasing the highest accuracy of 99.78%. Despite ResNet50's superior performance, both VGG16 and InceptionV3 achieved commendable accuracy of 99.67% and 99.37%, respectively. The overall outcomes reaffirm the robustness of the fine-tuning approach across different deep neural network architectures, offering insights into strategies for optimizing model performance in the domain of perennial plant image classification.

Keywords: perennial plants, image classification, deep neural networks, fine-tuning, transfer learning, VGG16, ResNet50, InceptionV3

Procedia PDF Downloads 63

Authors: Kemka C. Ihemelandu, Chukwuemeka U. Ihemelandu

Abstract:

The incidence of melanoma has been increasing rapidly over the past two decades, making melanoma a current public health crisis. Unfortunately, even as screening efforts continue to expand in an effort to ameliorate the death rate from melanoma, there is a need to improve diagnostic accuracy to decrease misdiagnosis. Artificial intelligence (AI) a new frontier in patient care has the ability to improve the accuracy of melanoma diagnosis. Convolutional neural network (CNN) a form of deep neural network, most commonly applied to analyze visual imagery, has been shown to outperform the human brain in pattern recognition. However, there are noted limitations with the accuracy of the CNN models. Our aim in this study was the optimization of convolutional neural network algorithms for the automated diagnosis of melanoma. We hypothesized that Optimal selection of the momentum and batch hyperparameter increases model accuracy. Our most successful model developed during this study, showed that optimal selection of momentum of 0.25, batch size of 2, led to a superior performance and a faster model training time, with an accuracy of ~ 83% after nine hours of training. We did notice a lack of diversity in the dataset used, with a noted class imbalance favoring lighter vs. darker skin tone. Training set image transformations did not result in a superior model performance in our study.

Keywords: melanoma, convolutional neural network, momentum, batch hyperparameter

Procedia PDF Downloads 100

Authors: Krishang Surapaneni

Abstract:

The goal of this project is to determine how to predict important aspects of options, including the ask price. We want to compare different machine learning models to learn the best model and the best hyperparameters for that model for this purpose and data set. Option pricing is a relatively new field, and it can be very complicated and intimidating, especially to inexperienced people, so we want to create a machine learning model that can predict important aspects of an option stock, which can aid in future research. We tested multiple different models and experimented with hyperparameter tuning, trying to find some of the best parameters for a machine-learning model. We tested three different models: a Random Forest Regressor, a linear regressor, and an MLP (multi-layer perceptron) regressor. The most important feature in this experiment is the ask price; this is what we were trying to predict. In the field of stock pricing prediction, there is a large potential for error, so we are unable to determine the accuracy of the models based on if they predict the pricing perfectly. Due to this factor, we determined the accuracy of the model by finding the average percentage difference between the predicted and actual values. We tested the accuracy of the machine learning models by comparing the actual results in the testing data and the predictions made by the models. The linear regression model performed worst, with an average percentage error of 17.46%. The MLP regressor had an average percentage error of 11.45%, and the random forest regressor had an average percentage error of 7.42%

Keywords: finance, linear regression model, machine learning model, neural network, stock price

Procedia PDF Downloads 74

Authors: Salman Hameed

Abstract:

In this paper, a self-tuning fuzzy PI controller (STFPIC) is proposed for thyristor controlled series capacitor (TCSC) to improve power system dynamic performance. In a STFPIC controller, the output scaling factor is adjusted on-line by an updating factor (α). The value of α is determined from a fuzzy rule-base defined on error (e) and change of error (Δe) of the controlled variable. The proposed self-tuning controller is designed using a very simple control rule-base and the most natural and unbiased membership functions (MFs) (symmetric triangles with equal base and 50% overlap with neighboring MFs). The comparative performances of the proposed STFPIC and the standard fuzzy PI controller (FPIC) have been investigated on a multi-machine power system (namely, 4 machine two area system) through detailed non-linear simulation studies using MATLAB/SIMULINK. From the simulation studies it has been found out that for damping oscillations, the performance of the proposed STFPIC is better than that obtained by the standard FPIC. Moreover, the proposed STFPIC as well as the FPIC have been found to be quite effective in damping oscillations over a wide range of operating conditions and are quite effective in enhancing the power carrying capability of the power system significantly.

Keywords: genetic algorithm, power system stability, self-tuning fuzzy controller, thyristor controlled series capacitor

Procedia PDF Downloads 422

Authors: Nikolai Bertelsen, Robert A. Alphinas, Klaus B. Orskov

Abstract:

The shortest possible tool stickout has been the traditional go-to approach with expectations of increased stability and productivity. However, experimental studies at Danish Advanced Manufacturing Research Center (DAMRC) have proven that for some tool stickout lengths, there exist local productivity optimums when utilizing the Stability Lobe Diagrams for chatter avoidance. This contradicts with traditional logic and the best practices taught to machinists. This paper explores the vibrational characteristics and behaviour of a milling system over the tool stickout length. The experimental investigation has been conducted by tap testing multiple endmills where the tool stickout length has been varied. For each length, the modal parameters have been recorded and mapped to visualize behavioural tendencies. Furthermore, the paper explores the correlation between the modal parameters and the Stability Lobe Diagram to outline the influence and importance of each parameter in a multi-mode system. The insights are conceptualized into a tool tuning approximation solution. It builds on an almost linear change in the natural frequencies when amending tool stickout, which results in changed positions of the Chatter-free Stability Lobes. Furthermore, if the natural frequency of two modes become too close, it will onset of the dynamic absorber effect phenomenon. This phenomenon increases the critical stable depth of cut, allowing for a more stable milling process. Validation tests on the tool tuning approximation solution have shown varying success of the solution. This outlines the need for further research on the boundary conditions of the solution to understand at which conditions the tool tuning approximation solution is applicable. If the conditions get defined, the conceptualized tool tuning approximation solution outlines an approach for quick and roughly approximating tool stickouts with the potential for increased stiffness and optimized productivity.

Keywords: milling, modal parameters, stability lobes, tap testing, tool tuning

Procedia PDF Downloads 156

Authors: H. Mansor, S.B. Mohd-Noor, N. I. Othman, N. Tazali, R. I. Boby

Abstract:

This paper presents an improved robust Proportional Derivative controller for a 3-Degree-of-Freedom (3-DOF) bench-top helicopter by using adaptive methodology. Bench-top helicopter is a laboratory scale helicopter used for experimental purposes which is widely used in teaching laboratory and research. Proportional Derivative controller has been developed for a 3-DOF bench-top helicopter by Quanser. Experiments showed that the transient response of designed PD controller has very large steady state error i.e., 50%, which is very serious. The objective of this research is to improve the performance of existing pitch angle control of PD controller on the bench-top helicopter by integration of PD controller with adaptive controller. Usually standard adaptive controller will produce zero steady state error; however response time to reach desired set point is large. Therefore, this paper proposed an adaptive with deadbeat algorithm to overcome the limitations. The output response that is fast, robust and updated online is expected. Performance comparisons have been performed between the proposed self-tuning deadbeat PD controller and standard PD controller. The efficiency of the self-tuning dead beat controller has been proven from the tests results in terms of faster settling time, zero steady state error and capability of the controller to be updated online.

Keywords: adaptive control, deadbeat control, bench-top helicopter, self-tuning control

Procedia PDF Downloads 322

Authors: Ming Liu, Chong Wu, Bingquan Liu, Lei Chen

Abstract:

Internet users have to face the massive amount of textual data every day. Organizing texts into categories can help users dig the useful information from large-scale text collection. Clustering, in fact, is one of the most promising tools for categorizing texts due to its unsupervised characteristic. Unfortunately, most of traditional clustering algorithms lose their high qualities on large-scale text collection. This situation mainly attributes to the high- dimensional vectors generated from texts. To effectively and efficiently cluster large-scale text collection, this paper proposes a vector reconstruction based clustering algorithm. Only the features that can represent the cluster are preserved in cluster’s representative vector. This algorithm alternately repeats two sub-processes until it converges. One process is partial tuning sub-process, where feature’s weight is fine-tuned by iterative process. To accelerate clustering velocity, an intersection based similarity measurement and its corresponding neuron adjustment function are proposed and implemented in this sub-process. The other process is overall tuning sub-process, where the features are reallocated among different clusters. In this sub-process, the features useless to represent the cluster are removed from cluster’s representative vector. Experimental results on the three text collections (including two small-scale and one large-scale text collections) demonstrate that our algorithm obtains high quality on both small-scale and large-scale text collections.

Keywords: vector reconstruction, large-scale text clustering, partial tuning sub-process, overall tuning sub-process

Procedia PDF Downloads 434