Novel hybrid firefly algorithm: an application to enhance XGBoost tuning for intrusion detection classification
PeerJ, Apr 29, 2022
The research proposed in this article presents a novel improved version of the widely adopted fir... more The research proposed in this article presents a novel improved version of the widely adopted firefly algorithm and its application for tuning and optimising XGBoost classifier hyper-parameters for network intrusion detection. One of the greatest issues in the domain of network intrusion detection systems are relatively high false positives and false negatives rates. In the proposed study, by using XGBoost classifier optimised with improved firefly algorithm, this challenge is addressed. Based on the established practice from the modern literature, the proposed improved firefly algorithm was first validated on 28 well-known CEC2013 benchmark instances a comparative analysis with the original firefly algorithm and other state-of-the-art metaheuristics was conducted. Afterwards, the devised method was adopted and tested for XGBoost hyper-parameters optimisation and the tuned classifier was tested on the widely used benchmarking NSL-KDD dataset and more recent USNW-NB15 dataset for network intrusion detection. Obtained experimental results prove that the proposed metaheuristics has significant potential in tackling machine learning hyper-parameters optimisation challenge and that it can be used for improving classification accuracy and average precision of network intrusion detection systems.
This paper describes an own implementation of a regular expression preprocessor written in PHP. I... more This paper describes an own implementation of a regular expression preprocessor written in PHP. It extends the regular expression functionality by allowing users to define named segments. These segments include custom character classes, matching groups etc. The pre-processor allows for writing complex regular expressions that are simpler to maintain. In addition, this paper presents a use case of the practical utilisation of the pre-processor. Furthermore, it includes a comparison of expressions written with and without user-defined segments.
Proceedings of the International Scientific Conference - Sinteza 2017, 2017
This paper discusses the application of modern information technologies to complement communicati... more This paper discusses the application of modern information technologies to complement communication between educational institutions and students. Communication is important for development and improvement of the quality of the education system. In place of direct communication, educational institutions sometimes need to make additional educational materials available to students. Information technologies play an important role in distribution of these materials. Students have access to personalised and relevant information at any time, as well as many ways and opportunities to clarify any doubt in a quick, easy and efficient way. This paper explains multiple applications of the educational content distribution. It includes reviews of different web-based services, such as a video distribution platform, a portal designed for scientific research and an application for the Android operating system for delivery of different kinds of content of interest to higher education students.
There is a growing interest in the study development of artificial intelligence and machine learn... more There is a growing interest in the study development of artificial intelligence and machine learning, especially regarding the support vector machine pattern classification method. This study proposes an enhanced implementation of the well-known whale optimisation algorithm, which combines chaotic and opposition-based learning strategies, which is adopted for hyper-parameter optimisation and feature selection machine learning challenges. The whale optimisation algorithm is a relatively recent addition to the group of swarm intelligence algorithms commonly used for optimisation. The Proposed improved whale optimisation algorithm was first tested for standard unconstrained CEC2017 benchmark suite and it was later adapted for simultaneous feature selection and support vector machine hyper-parameter tuning and validated for medical diagnostics by using breast cancer, diabetes, and erythemato-squamous dataset. The performance of the proposed model is compared with multiple competitive support vector machine models boosted with other metaheuristics, including another improved whale optimisation approach, particle swarm optimisation algorithm, bacterial foraging optimisation algorithms, and genetic algorithms. Results of the simulation show that the proposed model outperforms other competitors concerning the performance of classification and the selected subset feature size.
The evaluation of the appeal of website designs to a target group, with the aid of artificial neural network software
In this paper, we present a method of using an artificial neural network to evaluate the appeal o... more In this paper, we present a method of using an artificial neural network to evaluate the appeal of website design and layout to members of a target group, where the design is defined by predefined parameters that describe section properties. We have trained the neural network with a training set based on survey participants' website design evaluations. For this project’s implementation, we have utilized an open source JavaScript library for the neural network simulation and an own implementation of a dynamic website preview generator. The paper presents a derivative application that uses the trained neural network to generate suggested designs and their feature parameters that are likely to be appealing to members of the target group matching surveyed participants.
This work analyses the impact of the server operating system on the overall performance of hosted... more This work analyses the impact of the server operating system on the overall performance of hosted web sites. In the paper, we are presenting the results of comparing the influence of the two most common server operating system families, Windows and the GNU/Linux distributions. Through experiments, we have collected data based on which we have concluded the impact of these operating systems on the performance of a web site developed using the most commonly used web platform WordPress. The experiment was set up in such a way as to eliminate the impact of hardware on our measurements, which is done by using the same hardware, network and matching conditions. The experiment was conducted with a large number of repetitions to reduce relative errors in measurements. The performance analysis covers files of different sizes and types, including text and multimedia content typical for modern web sites.
Proceedings of the International Scientific Conference - Sinteza 2017, 2017
This paper describes an implementation of a method of sending file license information for files ... more This paper describes an implementation of a method of sending file license information for files downloaded from a server. There are many files available for download on the Internet and many of them were created by authors who have published them under a certain license. In most cases, the license information is lost or is unknown by the person downloading the file. Those files may be images, video content, text documents, audio recordings, executables, compressed archive files with software source code, educational materials of different kinds etc. Currently, the only way to specify the license information is to embed the license information in a file (overlay or stamp on an image, footer text in a document, a separate file in a compressed archive of files, etc.) or to show license information on the web page shown just prior to the step where the download link is available. This proposed method does not provide for a way to embed the license information into the original file, but instead, it allows for the license information to be sent with the file from the server to the user in the same HTTP response and vice-versa, from the server to the client. The license is stored using the extended file attributes mechanism.
Proceedings of the International Scientific Conference - Synthesis 2015, 2015
This paper presents an implementation of a location-based service developed for smartphone device... more This paper presents an implementation of a location-based service developed for smartphone devices. It also covers a review of the need for location-based services and methods of location acquisition, which include GPS, cell-stations, wireless access point and IP address based Geo-location lookup positioning. The paper covers a review of contemporary technologies for development of software systems based on the location-based services and their use on smartphone devices. The paper presents the implementation of a GIS designed to provide venue information based on the location information of smartphones with Android operating system. Apstrakt: U radu je predstavljena primena usluge bazirane na lokaciji, kreirane za pametne telefone. U njemu se analizira potreba za uslugama baziranim na lokaciji, kao i metode utvrđivanja lokacije, među kojima su GPS, lociranje pomoću baznih stanica, bežičnih mreža i utvrđivanje lokacije pomoću IP adrese. Rad sadrži i pregled savremenih tehnologija za izradu softverskih sistema baziranih na lokaciji i njihovu upotrebu na pametnim telefonima. Predstavljena je implementacija jednog geografskog informacionog sistema dizajniranog da pruži informacije o ugostiteljskim objektima, koje su u blizini pametnih uređaja sa Android operativnim sistemom, bazirane na njihovoj lokaciji.
Proceedings of the International Scientific Conference - Sinteza 2016, 2016
In this paper, we present a proposed multi-layered image format for use on the web. The format im... more In this paper, we present a proposed multi-layered image format for use on the web. The format implements an algorithm for selecting for rendering one of two or more layers adaptively. The selection process depends on the weighted average brightness of the underlying web page background within the image bounds. The proposed image format supports multiple image layers with adjoined thresholds. Depending on these thresholds and the underlying background, a layer will either be shown or hidden. The selection algorithm takes into account the background brightness and each layer's adjoined threshold values.
The research proposed in this article presents a novel improved version of the widely adopted fir... more The research proposed in this article presents a novel improved version of the widely adopted firefly algorithm and its application for tuning and optimising XGBoost classifier hyper-parameters for network intrusion detection. One of the greatest issues in the domain of network intrusion detection systems are relatively high false positives and false negatives rates. In the proposed study, by using XGBoost classifier optimised with improved firefly algorithm, this challenge is addressed. Based on the established practice from the modern literature, the proposed improved firefly algorithm was first validated on 28 well-known CEC2013 benchmark instances a comparative analysis with the original firefly algorithm and other state-of-the-art metaheuristics was conducted. Afterwards, the devised method was adopted and tested for XGBoost hyper-parameters optimisation and the tuned classifier was tested on the widely used benchmarking NSL-KDD dataset and more recent USNW-NB15 dataset for network intrusion detection. Obtained experimental results prove that the proposed metaheuristics has significant potential in tackling machine learning hyper-parameters optimisation challenge and that it can be used for improving classification accuracy and average precision of network intrusion detection systems.
There is a growing interest in the study development of artificial intelligence and machine learn... more There is a growing interest in the study development of artificial intelligence and machine learning, especially regarding the support vector machine pattern classification method. This study proposes an enhanced implementation of the well-known whale optimisation algorithm, which combines chaotic and opposition-based learning strategies, which is adopted for hyper-parameter optimisation and feature selection machine learning challenges. The whale optimisation algorithm is a relatively recent addition to the group of swarm intelligence algorithms commonly used for optimisation. The Proposed improved whale optimisation algorithm was first tested for standard unconstrained CEC2017 benchmark suite and it was later adapted for simultaneous feature selection and support vector machine hyper-parameter tuning and validated for medical diagnostics by using breast cancer, diabetes, and erythemato-squamous dataset. The performance of the proposed model is compared with multiple competitive support vector machine models boosted with other metaheuristics, including another improved whale optimisation approach, particle swarm optimisation algorithm, bacterial foraging optimisation algorithms, and genetic algorithms. Results of the simulation show that the proposed model outperforms other competitors concerning the performance of classification and the selected subset feature size.
Proceedings of the International Scientific Conference - Sinteza 2016, 2016
This paper describes an own implementation of a regular expression preprocessor written in PHP. I... more This paper describes an own implementation of a regular expression preprocessor written in PHP. It extends the regular expression functionality by allowing users to define named segments. These segments include custom character classes, matching groups etc. The pre-processor allows for writing complex regular expressions that are simpler to maintain. In addition, this paper presents a use case of the practical utilisation of the pre-processor. Furthermore, it includes a comparison of expressions written with and without user-defined segments.
Proceedings of the International Scientific Conference - Sinteza 2021, 2021
In this paper, the authors present an experimental study covering the evaluation of the average e... more In this paper, the authors present an experimental study covering the evaluation of the average encryption and decryption times using different programming languages for the Web. This study covers the use of the most commonly used AES implementations for four major Web programming and scripting languages: Java, Node.js, PHP and Python. The aim of the study is to determine the cost of encrypting and decrypting data on these platforms, expressed as type per byte of data. The experiment covers data encryption and decryption with the AES algorithm in the CBC mode with 128-bit, 192-bit and 256-bit keys. In this paper, we present the results and pros and cons of use of the AES algorithm implementations on these major Web platforms.
There is a growing interest in the study development of artificial intelligence and machine learn... more There is a growing interest in the study development of artificial intelligence and machine learning, especially regarding the support vector machine pattern classification method. This study proposes an enhanced implementation of the well-known whale optimisation algorithm, which combines chaotic and opposition-based learning strategies, which is adopted for hyper-parameter optimisation and feature selection machine learning challenges. The whale optimisation algorithm is a relatively recent addition to the group of swarm intelligence algorithms commonly used for optimisation. The Proposed improved whale optimisation algorithm was first tested for standard unconstrained CEC2017 benchmark suite and it was later adapted for simultaneous feature selection and support vector machine hyper-parameter tuning and validated for medical diagnostics by using breast cancer, diabetes, and erythemato-squamous dataset. The performance of the proposed model is compared with multiple competitive support vector machine models boosted with other metaheuristics, including another improved whale optimisation approach, particle swarm optimisation algorithm, bacterial foraging optimisation algorithms, and genetic algorithms. Results of the simulation show that the proposed model outperforms other competitors concerning the performance of classification and the selected subset feature size.
Proceedings of the International Scientific Conference - Sinteza 2019, 2019
In this paper, authors present the result of an experimental study which covers the technical eva... more In this paper, authors present the result of an experimental study which covers the technical evaluation of 200 websites. Of the total, 100 are with content in the Arabic language and 100 are with content in the Serbian language. The technical evaluation covers the most common web development practices which are usually neglected. These include the application of the responsive page design, page download speed, the analysis of the most frequent HTML code errors and warnings, the use of CSS and JavaScript compression and the use of HTTPS certificates for secure connections. The test data is obtained with a script which collects information from a list of domains. The script processes the acquired information and performs code validation using the W3C HTML Validator tool. This paper presents a comparison of the evaluation results of web sites in Arabic and web sites in Serbian. The paper points out the most common technical problems and inadequacies of web sites in Arabic and Serbian languages. This research aims to help developers to maintain and improve their web page development processes and to avoid making common errors, which lead to increased incompatibility of web page with different browsers.
Sinteza 2021 - International Scientific Conference on Information Technology and Data Related Research, 2021
This work analyses the impact of the server operating system on the overall performance of hosted... more This work analyses the impact of the server operating system on the overall performance of hosted web sites. In the paper, we are presenting the results of comparing the influence of the two most common server operating system families, Windows and the GNU/Linux distributions. Through experiments, we have collected data based on which we have concluded the impact of these operating systems on the performance of a web site developed using the most commonly used web platform - WordPress. The experiment was set up in such a way as to eliminate the impact of hardware on our measurements, which is done by using the same hardware, network and matching conditions. The experiment was conducted with a large number of repetitions to reduce relative errors in measurements. The performance analysis covers files of different sizes and types, including text and multimedia content typical for modern web sites.
Sinteza 2021 - International Scientific Conference on Information Technology and Data Related Research , 2021
In this paper, the authors present an experimental study covering the evaluation of the average e... more In this paper, the authors present an experimental study covering the evaluation of the average encryption and decryption times using different programming languages for the Web. This study covers the use of the most commonly used AES implementations for four major Web programming and scripting languages: Java, Node.js, PHP and Python. The aim of the study is to determine the cost of encrypting and decrypting data on these platforms, expressed as type per byte of data. The experiment covers data encryption and decryption with the AES algorithm in the CBC mode with 128-bit, 192-bit and 256-bit keys. In this paper, we present the results and pros and cons of use of the AES algorithm implementations on these major Web platforms.
Sažetak: U ovom diplomskom radu je opisan princip kompjuterske emulacije sa primerom kompjutersko... more Sažetak: U ovom diplomskom radu je opisan princip kompjuterske emulacije sa primerom kompjuterskog programa za emuliranje rada računarskog hardvera arhitekture računara Milantex Tx16. Emulator je realizovan sa procesorom u vidu interpretera programskog koda i sa virtualnim periferijama za ulaz i izlaz, kao i sa virtualnim uređajima za trajno skladištenje podataka.
Proceedings of the International Scientific Conference - Sinteza 2016, 2016
The aim of this paper is to clarify the significance of using responsive logos on web pages. It a... more The aim of this paper is to clarify the significance of using responsive logos on web pages. It aims to bring a more flexible and contextual system of brand identification. This can be done by moving away from strict and inflexible design guidelines. The authors shall attempt to explain why logos in all its variations can be unambiguous memory hooks. Also, we emphasise that they can behave like many other elements in responsive web pages. Contemporary businesses demand adaptation of their content to many sophisticated communication platforms. Among these are the devices with different varying screen dimensions. A responsive logo can be used to build better user experiences across a wide range of devices, thus elevating brand freshness. Having in mind that logos are the key elements of corporate identity and brand positioning, it is mandatory to optimise it for every scenario. It is an element on par with colours or slogans. It can be simplified by applying responsive design principles to individual elements that make out the logo. By removing unnecessary parts in relation to the varying screen size, a clearer and much more readable logo can be arranged. We have built a model for applying responsive design to logos and performed experimental testing in the real environment.
Uploads
Papers by Milan Tair