Browsing by Author "Guzelis, Cuneyt"

Now showing 1 - 20 of 46

Citation - WoS: 1
Citation - Scopus: 1
24-hour Electricity Consumption Forecasting for Day Ahead Market with Long Short Term Memory Deep Learning Model
(IEEE, 2020) Nalan Ozkurt; Hacer Sekerci Oztura; Cuneyt Guzelis; Guzelis, Cuneyt; Oztura, Hacer Sekerci; Ozkurt, Nalan
In 2015 with the foundation of Energy Market Management Inc. AS (EPIAS) the production and pricing of electrical energy began to be made according to consumption estimates. In this study twenty-four hours energy consumption forecasting was made by using long short-term memory method and data was downloded from EPIAS's official web page for the Day Ahead Market. The data set used covers 1500 days between June 2016 and July 2020. The results obtained have been compared with EPIAS's own estimates and actual consumption data.
Citation - WoS: 1
Citation - Scopus: 1
A 2-dimensional model of polynomial type for oscillatory ATM-Wipl dynamics in p53 network
(Institute of Electrical and Electronics Engineers Inc., 2017) Gökhan Demirkıran; Güleser Kalayci Demir; Cüneyt Güzeliş; Guzelis, Cuneyt; Demirkiran, Gokhan; Demir, Guleser Kalayc
Under gamma irradiation p53 gene regulatory network is able to exhibit three different modes namely low state oscillations and high state. There are experimental studies demonstrating that oscillatory behaviour of p53 is due to the interaction between upstream mediator of p53 i.e. ATM and a negative feedback loop formed by Wipl with that upstream. By proposing a canonical model based on ordinary differential equations made up of polynomial type birth and death terms we show mathematically that the simple interaction between ATM and Wipl is indeed able to exhibit three different behaviours relevant to DNA damage response of p53 network. We further carry out bifurcation analysis on the model with the aim of investigating the mutations such as Wip1 overexpression and ATM deficiency. Based on the proposed canonical model we show that Wipl is an important target for curing these types of mutations. © 2023 Elsevier B.V. All rights reserved.
Citation - WoS: 1
Citation - Scopus: 1
A 2-dimensional reduced oscillator model with rational nonlinearities for p53 dynamics
(Institute of Electrical and Electronics Engineers Inc., 2017) Gökhan Demirkıran; Güleser Kalayci Demir; Cüneyt Güzeliş; Guzelis, Cuneyt; Demirkiran, Gokhan; Demir, Guleser Kalayc
p53 tumour suppressor network plays the key role in DNA damage response of the cell. We present a comprehensive 2-dimensional oscillator model for p53 network dynamics. The model is reduced from the known 17-dimensional two-phase model of p53 network which shows temporary oscillatory behaviour under DNA damage type of Double Strand Breaks. The introduced oscillator model shows the same qualitative dynamics of two-phase model of p53 network such as bistability and oscillations when the parameters are adjusted accordingly. With the help of the identified oscillator in p53 network we introduce a new oscillator perspective: p53 network has an oscillator in the centre which other systems in the cell manipulate this oscillator to contribute to cell fate. The introduction of such low dimensional oscillator model will make it possible to study p53 network and the effects of other biological systems in the context of oscillations. © 2023 Elsevier B.V. All rights reserved.
Citation - WoS: 2
Citation - Scopus: 4
A comparison of feature selection algorithms for cancer classification through gene expression data: Leukemia case
(Institute of Electrical and Electronics Engineers Inc., 2017) Asli Tasci; Türker Ince; Cüneyt Güzeliş; Guzelis, Cuneyt; Tasci, Asli; Ince, Turker
In this study three different feature selection algorithms are compared using Support Vector Machines as classifier for cancer classification through gene expression data. The ability of feature selection algorithms to select an optimal gene subset for a cancer type is evaluated by the classification ability of selected genes. A publicly available micro array dataset is employed for gene expression values. Selected gene subsets were able to classify subtypes of the considered cancer type with high accuracies and showed that these feature selection methods were applicable for bio-marker gene selection. © 2023 Elsevier B.V. All rights reserved.
Citation - WoS: 16
Citation - Scopus: 26
A Multiscale Algorithm for Joint Forecasting-Scheduling to Solve the Massive Access Problem of IoT
(Institute of Electrical and Electronics Engineers Inc., 2020) Volkan Rodoplu; Mert Nakıp; D. T. Eliiyi; Cüneyt Güzeliş; Rodoplu, Volkan; Guzelis, Cuneyt; Nakip, Mert; Eliiyi, Deniz Tursel
The massive access problem of the Internet of Things (IoT) is the problem of enabling the wireless access of a massive number of IoT devices to the wired infrastructure. In this article we describe a multiscale algorithm (MSA) for joint forecasting-scheduling at a dedicated IoT gateway to solve the massive access problem at the medium access control (MAC) layer. Our algorithm operates at multiple time scales that are determined by the delay constraints of IoT applications as well as the minimum traffic generation periods of IoT devices. In contrast with the current approaches to the massive access problem that assume random arrivals for IoT data our algorithm forecasts the upcoming traffic of IoT devices using a multilayer perceptron architecture and preallocates the uplink wireless channel based on these forecasts. The multiscale nature of our algorithm ensures scalable time and space complexity to support up to 6650 IoT devices in our simulations. We compare the throughput and energy consumption of MSA with those of reservation-based access barring (RAB) priority based on average load (PAL) and enhanced predictive version burst-oriented (E-PRV-BO) protocols and show that MSA significantly outperforms these beyond 3000 devices. Furthermore we show that the percentage control overhead of MSA remains less than 1.5%. Our results pave the way to building scalable joint forecasting-scheduling engines to handle a massive number of IoT devices at IoT gateways. © 2020 Elsevier B.V. All rights reserved.
Citation - WoS: 16
Citation - Scopus: 21
Advancements in Deep Reinforcement Learning and Inverse Reinforcement Learning for Robotic Manipulation: Toward Trustworthy Interpretable and Explainable Artificial Intelligence
(IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC, 2024) Recep Ozalp; Aysegul Ucar; Cuneyt Guzelis; Guzelis, Cuneyt; Ucar, Aysegul; Ozalp, Recep
This article presents a literature review of the past five years of studies using Deep Reinforcement Learning (DRL) and Inverse Reinforcement Learning (IRL) in robotic manipulation tasks. The reviewed articles are examined in various categories including DRL and IRL for perception assembly manipulation with uncertain rewards multitasking transfer learning multimodal and Human-Robot Interaction (HRI). The articles are summarized in terms of the main contributions methods challenges and highlights of the latest and relevant studies using DRL and IRL for robotic manipulation. Additionally summary tables regarding the problem and solution are presented. The literature review then focuses on the concepts of trustworthy AI interpretable AI and explainable AI (XAI) in the context of robotic manipulation. Moreover this review provides a resource for future research on DRL/IRL in trustworthy robotic manipulation.
Citation - WoS: 1
Citation - Scopus: 2
Aggregation for Computing Multi-Modal Stationary Distributions in 1-D Gene Regulatory Networks
(IEEE COMPUTER SOC, 2018) Neslihan Avcu; Nihal Pekergin; Ferhan Pekergin; Cuneyt Guzelis; Pekergin, Nihal; Pekergin, Ferhan; Avcu, Neslihan; Guzelis, Cuneyt
This paper proposes aggregation-based three-stage algorithms to overcome the numerical problems encountered in computing stationary distributions and mean first passage times for multi-modal birth-death processes of large state space sizes. The considered birth-death processes which are defined by Chemical Master Equations are used in modeling stochastic behavior of gene regulatory networks. Computing stationary probabilities for a multi-modal distribution from Chemical Master Equations is subject to have numerical problems due to the probability values running out of the representation range of the standard programming languages with the increasing size of the state space. The aggregation is shown to provide a solution to this problem by analyzing first reduced size subsystems in isolation and then considering the transitions between these subsystems. The proposed algorithms are applied to study the bimodal behavior of the lac operon of E. coli described with a one-dimensional birth-death model. Thus the determination of the entire parameter range of bimodality for the stochastic model of lac operon is achieved.
Citation - WoS: 8
Citation - Scopus: 15
An End-to-End Trainable Feature Selection-Forecasting Architecture Targeted at the Internet of Things
(IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC, 2021) Mert Nakip; Kubilay Karakayali; Cuneyt Guzelis; Volkan Rodoplu; Karakayali, Kubilay; Guzelis, Cuneyt; Rodoplu, Volkan; Nakip, Mert
We develop a novel end-to-end trainable feature selection-forecasting (FSF) architecture for predictive networks targeted at the Internet of Things (IoT). In contrast with the existing filter-based wrapper-based and embedded feature selection methods our architecture enables the automatic selection of features dynamically based on feature importance score calculation and gamma-gated feature selection units that are trained jointly and end-to-end with the forecaster. We compare the performance of our FSF architecture on the problem of forecasting IoT device traffic against the following existing (feature selection forecasting) technique pairs: Autocorrelation Function (ACF) Analysis of Variance (ANOVA) Recurrent Feature Elimination (RFE) and Ridge Regression methods for feature selection and Linear Regression Multi-Layer Perceptron (MLP) Long Short Term Memory (LSTM) 1 Dimensional Convolutional Neural Network (1D CNN) Autoregressive Integrated Moving Average (ARIMA) and Logistic Regression for forecasting. We show that our FSF architecture achieves either the best or close to the best performance among all of the competing techniques by virtue of its dynamic automatic feature selection capability. In addition we demonstrate that both the training time and the execution time of FSF are reasonable for IoT applications. This work represents a milestone for the development of predictive networks for IoT in smart cities of the near future.
Citation - Scopus: 16
An Implementation of Vision Based Deep Reinforcement Learning for Humanoid Robot Locomotion
(Institute of Electrical and Electronics Engineers Inc., 2019) Recep Ozalp; Çaǧri Kaymak; Özal Yildirim; Ayşegül Uçar; Yakup Demir; Cüneyt Güzeliş; Ozaln, Recen; Yildirum, Ozal; Guzelis, Cuneyt; Kaymak, Cagri; Ucar, Ayscgul; Demir, Yakup; P. Koprinkova-Hristova , T. Yildirim , V. Piuri , L. Iliadis , D. Camacho
Deep reinforcement learning (DRL) exhibits a promising approach for controlling humanoid robot locomotion. However only values relating sensors such as IMU gyroscope and GPS are not sufficient robots to learn their locomotion skills. In this article we aim to show the success of vision based DRL. We propose a new vision based deep reinforcement learning algorithm for the locomotion of the Robotis-op2 humanoid robot for the first time. In experimental setup we construct the locomotion of humanoid robot in a specific environment in the Webots software. We use Double Dueling Q Networks (D3QN) and Deep Q Networks (DQN) that are a kind of reinforcement learning algorithm. We present the performance of vision based DRL algorithm on a locomotion experiment. The experimental results show that D3QN is better than DQN in that stable locomotion and fast training and the vision based DRL algorithms will be successfully able to use at the other complex environments and applications. © 2020 Elsevier B.V. All rights reserved.
Citation - WoS: 6
Citation - Scopus: 7
Bifurcation analysis of bistable and oscillatory dynamics in biological networks using the root-locus method
(INST ENGINEERING TECHNOLOGY-IET, 2019) Neslihan Avcu; Cuneyt Guzelis; Guzelis, Cuneyt; Avcu, Neslihan
Most of the biological systems including gene regulatory networks can be described well by ordinary differential equation models with rational non-linearities. These models are derived either based on the reaction kinetics or by curve fitting to experimental data. This study demonstrates the applicability of the root-locus-based bifurcation analysis method for studying the complex dynamics of such models. The effectiveness of the bifurcation analysis in determining the exact parameter regions in each of which the system shows a certain dynamical behaviour such as bistability oscillation and asymptotically equilibrium dynamics is shown by considering two mostly studied gene regulatory networks namely Gardner's genetic toggle switch and p53 gene network possessing two-phase (mono-stable/oscillation) dynamics.
Comparative Study of Forecasting Models for COVID-19 Outbreak in Turkey
(Institute of Electrical and Electronics Engineers Inc., 2021) Mert Nakıp; Onur Çopur; Cüneyt Güzeliş; Guzelis, Cuneyt; Nakip, Mert; Copur, Onur
This paper gives an explanation for the failure of machine learning models for the prediction of the cases and the other future trends of Covid-19 pandemic. The paper shows that simple Linear Regression models provide high prediction accuracy values reliably but only for a 2-weeks period and that relatively complex machine learning models which have the potential of learning long-term predictions with low errors cannot achieve to obtain good predictions with possessing a high generalization ability. It is suggested in the paper that the lack of a sufficient number of samples is the source of the low prediction performance of the forecasting models. To exploit the information which is of most relevant with the active cases we perform feature selection over a variety of variables such as the numbers of active cases deaths recoveries and population. Furthermore we compare Linear Regression Multi-Layer Perceptron and Long-Short Term Memory models each of which is used for prediction of active cases together with various feature selection methods. Our results show that the accurate forecasting of the active cases with high generalization ability is possible up to 3 days because of the small sample size of COVID-19 data. We observe that the Linear Regression model has much better prediction performance with high generalization ability as compared to the complex models but as expected its performance decays sharply for more than 14-days prediction horizons. © 2022 Elsevier B.V. All rights reserved.
Citation - WoS: 15
Citation - Scopus: 21
Comparative Study of Forecasting Schemes for IoT Device Traffic in Machine-to-Machine Communication
(ASSOC COMPUTING MACHINERY, 2019) Mert Nakip; Baran Can Gul; Volkan Rodoplu; Cuneyt Guzelis; Gul, Baran Can; Rodoplu, Volkan; Guzelis, Cuneyt; Nakip, Mert
We present a comparative study of Autoregressive Integrated Moving Average (ARIMA) Multi-Layer Perceptron (MLP) 1-Dimensional Convolutional Neural Network (1-D CNN) and Long-Short Term Memory (LSTM) models on the problem of forecasting the traffic generation patterns of individual Internet of Things (IoT) devices in Machine-to-Machine (M2M) communication. We classify IoT traffic into four classes: Fixed-Bit Periodic (FBP) Variable-Bit Periodic (VBP) Fixed-Bit Aperiodic (FBA) and Variable-Bit Aperiodic (VBA). We show that LSTM outperforms all of the other models significantly in the symmetric Mean Absolute Percentage Error (sMAPE) measure for devices in the VBP class in our simulations. Furthermore we show that LSTM has almost the same performance in this metric for the FBA class as MLP and 1-D CNN. While the training time per IoT device is the highest for LSTM all of the forecasting models have reasonable training times for practical implementation. Our results suggest an architecture in which an IoT Gateway predicts the future traffic of IoT devices in the FBP VBP and FBA classes and pre-allocates the uplink wireless channel for these classes in advance in order to alleviate the Massive Access Problem of M2M communication.
Citation - WoS: 27
Citation - Scopus: 34
Conditional Weighted Ensemble of Transferred Models for Camera Based Onboard Pedestrian Detection in Railway Driver Support Systems
(IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC, 2020) Tugce Toprak; Burak Belenlioglu; Burak Aydin; Cuneyt Guzelis; M. Alper Selver; Toprak, Tugce; Guzelis, Cuneyt; Selver, M. Alper; Belenlioglu, Burak; Aydin, Burak
Pedestrian Detection (PD) is one of the most studied issues of driver assistance systems. Although a tremendous effort is already given to create datasets and to develop classifiers for cars studies about railway systems remain very limited. This article shows that direct application of neither existing advanced object detectors (such as AlexNet VGG YOLOetc.) nor specifically created systems for PD(such as Caltech/INRIA trained classifiers) can provide enough performance to overcome railway specific challenges. Fortunately it is also shown that without waiting the collection of a mature dataset for railways as comprehensively diverse and annotated as the existing ones for cars a Transfer Learning (TL) approach to fine-tune various successful deep models (pre-trained using both extensive image and pedestrian datasets) to railway PD tasks provides an effective solution. To achieve TL a new RAil-Way PEdestrian Dataset (RAWPED) is collected and annotated. Then a novel three-stage system is designed. At its first stage a feature-classifier fusion is created to overcome the localization and adaptation limitations of deep models. At the second stage the complementarity of the transferred models and diversity of their results are exploited by conducted measurements and analyses. Based on the findings at the third stage a novel learning strategy is developed to create an ensemble which conditionally weights the outputs of individual models and performs consistently better than its components. The proposed system is shown to achieve a log average miss rate of 0.34 and average precision of 0.93 which are significantly better than the performance of compared well-established models.
Citation - Scopus: 4
Converting Utility Meters from Analogue to Smart based on Deep Learning Models
(Institute of Electrical and Electronics Engineers Inc., 2020) Humberto J.Cabeza Barreto; Ilker Kurtulan; Suleyman Inci; Mert Nakıp; Cüneyt Güzeliş; Barreto, Humberto J Cabeza; Kurtulan, Ilker; Guzelis, Cuneyt; Inci, Suleyman; Nakip, Mert
In this paper we proposed a system that automatically interprets the data of the utility meters by analyzing the photo of an analogue meter. In addition it sends the meter data to the consumers and the providers. We based the system on Convolutional Neural Networks (CNN) where we compared the You Only Look Once (YOLO) and a LeNet as CNN models. We collected the data for the training of each CNN model from the demonstration set of the project. Our results show that the YOLO model is reliable and fast. The model has a 99% accuracy for the gas meter and 98% accuracy for the water meter. © 2020 Elsevier B.V. All rights reserved.
Citation - WoS: 1
Citation - Scopus: 1
Coupling of cell fate selection model enhances DNA damage response and may underlie BE phenomenon
(INST ENGINEERING TECHNOLOGY-IET, 2020) Goekhan Demirkiran; Guleser Kalayc Demir; Cuneyt Guzelis; Guzelis, Cuneyt; Demirkiran, Goekhan; Demir, Guleser Kalayci
Double-strand break-induced (DSB) cells send signal that induces DSBs in neighbour cells resulting in the interaction among cells sharing the same medium. Since p53 network gives oscillatory response to DSBs such interaction among cells could be modelled as an excitatory coupling of p53 network oscillators. This study proposes a plausible coupling model of three-mode two-dimensional oscillators which models the p53-mediated cell fate selection in globally coupled DSB-induced cells. The coupled model consists of ATM and Wip1 proteins as variables. The coupling mechanism is realised through ATM variable via a mean-field modelling the bystander signal in the intercellular medium. Investigation of the model reveals that the coupling generates more sensitive DNA damage response by affecting cell fate selection. Additionally the authors search for the cause-effect relationship between coupled p53 network oscillators and bystander effect (BE) endpoints. For this they search for the possible values of uncertain parameters that may replicate BE experiments' results. At certain parametric regions there is a correlation between the outcomes of cell fate and endpoints of BE suggesting that the intercellular coupling of p53 network may manifest itself as the form of observed BEs.
Citation - Scopus: 3
Data Dependent Stable Robust Adaptive Controller Design for Altitude Control of Quadrotor Model
(Institute of Electrical and Electronics Engineers Inc., 2018) Mehmet Uğur Soydemir; Ishak Alkus; Parvin Bulucu; Aykut Kocaoǧlu; Cüneyt Güzeliş; Savaş Şahin; Bulucu, Parvin; Kocaoglu, Aykut; Soydemir, Mehmet Ugur; Guzelis, Cuneyt; Alkus, Ishak; Sahin, Savas; D. Maga , A. Stefek , T. Brezina
This paper presents Nonlinear Auto Regressive Moving Average (NARMA) based stable robust adaptive controller design. Both the plant and the closed-loop controller systems are modelled by the proposed NARMA based input-output models. During online supervised learning for the system identification and the controller design phases input-output data obtained from the simulated plant are evaluated in suitable parameter regions providing Schur stability for the overall closed-loop system. At the same time ϵ-insentive loss function and ℓ1 norm are used for providing robustness for proposed system identification and adaptive controller parameters. The proposed controller design method is performed on quadrotor model which is an unmanned air vehicle benchmark plant. The performance results are compared against proportional derivative controller. © 2023 Elsevier B.V. All rights reserved.
Citation - Scopus: 2
Data fusion integrated network forecasting scheme classifier (DFI-NFSC) via multi-layer perceptron decomposition architecture
(Elsevier B.V., 2024) Erdem Çakan; Volkan Rodoplu; Cüneyt Güzeliş; Rodoplu, Volkan; Guzelis, Cuneyt; Cakan, Erdem
The Massive Access Problem of the Internet of Things stands for the access problem of the wireless devices to the Gateway when the device population in the coverage area is excessive. We develop a hybrid model called Data Fusion Integrated Network Forecasting Scheme Classifier (DFI-NFSC) using a Multi-Layer Perceptron (MLP) Decomposition architecture specifically designed to address the Massive Access Problem. We utilize our custom error metric to display throughput and energy consumption results. These results are obtained by emulating the Joint Forecasting–Scheduling (JFS) system on a single IoT Gateway and distinguishing between ARIMA LSTM and MLP forecasters of the JFS system. The outcomes indicate that the DFI-NFCS method plays a notable role in improving performance and mitigating challenges arising from the dynamic fluctuations in the diversity of device types within an IoT gateway's coverage zone. © 2024 Elsevier B.V. All rights reserved.
Citation - WoS: 1
Citation - Scopus: 3
Design of an interactive fashion recommendation platform with intelligent systems, Proiectarea unei platforme interactive de recomandare a articolelor de modă cu sisteme inteligente
(Inst. Nat. Cercetare-Dezvoltare Text. Pielarie, 2024) Arzu Vuruşkan; Gökhan Demirkıran; Ender Yazgan Bulgun; Türker Ince; Cüneyt Güzeliş; Vuruskan, Arzu; Demirkiran, Gokhan; Ince, Turker; Bulgun, Ender; Guzelis, Cuneyt
With the increase in customer expectations in online fashion sales greater integration of fashion recommender systems (RSs) allows more personalization. Design decisions rely on personal taste as well as many other external influences such as trends and social media making it challenging to adapt intelligent systems for the fashion industry. Different methods for recommending personalized fashion items have been proposed however the literature still lacks an approach for recommending expert-suggested and personalized items. In this research an interactive web-based platform is developed to support personalized fashion styling focusing on users with diverse body shapes. To merge the user’s taste and the expert’s suggestion the proposed methodology in this research combines genetic algorithms and machine learning techniques allowing the system to access expert knowledge (including external influences) and incremental learning capability by adapting to the user preferences that unfold during interaction with the system. © 2024 Elsevier B.V. All rights reserved.
Citation - WoS: 7
Citation - Scopus: 8
Design of microcontroller-based decentralized controller board to drive chiller systems using PID and fuzzy logic algorithms
(SAGE Publications Ltd, 2020) Yalcin Yalcin Isler; Savaş Şahin; Orhan Ekren; Cüneyt Güzeliş; Isler, Yalcin; Ekren, Orhan; Guzelis, Cuneyt; Sahin, Savas
This study deals with designing a decentralized multi-input multi-output controller board based on a low-cost microcontroller which drives both parts of variable-speed scroll compressor and electronic-type expansion valve simultaneously in a chiller system. This study aims to show the applicability of commercial low-cost microcontroller to increase the efficiency of the chiller system having variable-speed scroll compressor and electronic-type expansion valve with a new electronic card. Moreover the refrigerant system proposed in this study provides the compactness mobility and flexibility and also a decrease in the controller unit’s budget. The study was tested on a chiller system that consists of an air-cooled condenser a variable-speed scroll compressor and a stepper driven electronic-type expansion valve. The R134a was used as a refrigerant fluid and its flow was controlled by electronic-type expansion valve in this setup. Both variable-speed scroll compressor and electronic-type expansion valve were driven by the proposed hardware using either proportional integral derivative or fuzzy logic controller which defines four distinct controller modes. The experimental results show that fuzzy logic controlled electronic-type expansion valve and proportional integral derivative controlled variable-speed scroll compressor mode give more robustness by considering the response time. © 2022 Elsevier B.V. All rights reserved.
Citation - WoS: 3
Citation - Scopus: 5
Development of a Multi-Sensor Fire Detector Based On Machine Learning Models
(IEEE, 2019) Mert Nakip; Cuneyt Guzelis; Guzelis, Cuneyt; Nakip, Mert
This paper proposes a method to reduce false positive fire alarms by fusing data from different sensors using a specific machine learning model. We design an electronic circuit with 6 sensors to detect 7 physical sensory inputs. We experimentally collect dataset for training and testing of machine learning models which are used for the implementation of fusing and classifying sensor data. An algorithm which employs the trained machine learning model for the classification of sensor data and then the thresholding is designed. Machine learning models are selected based on the results of comparisons among multi-layer perceptron support vector machine and radial basis function network. We use classification accuracy percentage false negative error and false positive error as measures for comparison. Multi-layer perceptron is observed as the best model according to its 96.875% classification accuracy.