Search  English (United States) Hrvatski (Hrvatska)

innovative promotional partnershipArtificial Intelligence towards EU Multilingualism

Technical co-sponsorship

 
Hybrid Event
Event program
Thursday, 5/25/2023 10:00 AM - 1:00 PM,
Leut II, Hotel Admiral, Opatija
10:00 AM - 1:00 PMDistributed Computing, Information Systems, and Software 
1.M. Saari, J. Harjamäki, M. Nurminen, P. Rantanen (Tampere University, Pori, Finland)
Comparing the Use of Custom-built and Commercial Off-the-shelf Data Gathering Devices in IoT Systems 
IoT refers to a wide range of devices that have Internet access, and collect and transmit information. Data have to be gathered from the most diverse environments. Custom data gathering devices are often necessary in situations where the data being collected are unique or need to be collected in a certain way. This is because commercial off-the-shelf (COTS) products may not have the necessary capabilities or features to meet the specific requirements of the data collection. This study focuses on COTS and selfbuilt data gathering devices by comparing two data gathering prototype systems: The first designed and implemented from scratch and the second built with COTS components. Both systems are compared in terms of various features such as cost-effectiveness, time saving, and the usefulness of the output. The end goal of the study is to provide a proposal for when it is appropriate to use COTS components and when it is necessary to design and build devices yourself.
2.J. Grönman, P. Rantanen, P. Sillberg (Tampere University, Pori, Finland), T. Pohjola (Turku School of Economics at the University of Turku, Pori, Finland), T. Jönkkäri (Tampere University, Pori, Finland)
Challenges of Combining Open and Commercial Data Sources in Visitor Mobility Estimations 
This paper explores the opportunities and challenges of open data. Open and free interfaces can provide useful data, but sometimes there are challenges with data sufficiency and quality. Open data can be enriched by acquiring commercial data, and by merging multiple data sources, a more comprehensive understanding of a situation can be obtained. The focus of this paper is on nature parks, but the technical solutions can be applied to other venues. In the case presented by this paper, the parks are equipped with on-location visitor counters, and with the help of commercial data provided by a mobile network operator, and open sources, such as weather and road traffic services, we are attempting to gain a better view of visitor movements. The motivation for this research originated from the cities located in the Satakunta region of Finland. For the purpose of developing recreational and business opportunities in the region, the cities require a better understanding of movement of people - which locations and services are popular, and which locations need improvement. To help achieve this goal, this paper presents a technical solution of how analysis and visualization of combined data can be used to estimate visitor amounts within a geographical area.
3.H. Čeović, M. Šilić, G. Delač, K. Vladimir (University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia)
An Overview of Diffusion Models for Text Generation 
Given the great success that diffusion models have achieved in generating various types of continuous data, including image, video and audio, there has been a growing interest in the application of these models to text generation. However, the discrete nature of text presents a challenge for diffusion models initially designed for application in a continuous feature space. The two main lines of work that aim to bring together diffusion models and natural language processing are focused on either defining the diffusion process in continuous space by converting discrete tokens to embeddings or defining the diffusion process in discrete space. These recent works attempt to combine diffusion models with leading sequence-to-sequence generation Transformer architecture as well as with existing pre-trained language models. In this work, we give a detailed overview of the approaches developed to date. We present and analyze the benefits and limitations that each model introduces, along with how they compare to the autoregressive models that dominate this field.
4.M. Džida, D. Vukadin, M. Šilić, G. Delač, K. Vladimir (University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia)
An Overview of State-of-the-Art Solutions for Scene Text Detection 
Scene text detection is a task of identifying text regions and labeling them with bounding boxes in a complex background. It has received a lot of attention recently and has become far from unsolvable due to progress of deep learning for computer vision and also due to rapid development of computer hardware which is able to process complex neural networks. Some of the most common challenges that make this task difficult are irregular text shapes, text interferences, very complex background, different text sizes and low image quality. This paper presents an overview of state-of-the-art solutions for scene text detection where ICDAR 2015 was used as a benchmark dataset. We compare solutions with respect to precision, recall and F-score.
5.E. Šmuc, G. Delač, M. Šilić, K. Vladimir (Sveučilište u Zagrebu Fakultet elektrotehnike i računarstva, Zagreb, Croatia)
A Transfer Learning Method for Hate Speech Detection 
In this work we explore the possibilities of using transfer learning techniques to enhance performance of hate speech detection models by relying on similar linguistic problems (e.g. toxic language detection). Multiple algorithms are trained for similar linguistic tasks on larger datasets, and the obtained models are used for getting predictions on the ETHOS dataset, which we chose as the target dataset of our work. The obtained predictions are used as sole or additional features in the subsequently performed experiments. Multiple algorithms are evaluated, including Logistic Regression, SVM, RidgeClassifier, Decision Tree, Random Forest, AdaBoost, GradBoost, Bagging. Furthermore, multiple textual representations are taken into account including Tf-Idf, Bert embeddings and BERT embeddings combined with the aforementioned additional features. Transformer-based models BERT and DistilBERT are introduced and fine-tuned on ETHOS dataset. All the obtained models are evaluated and the resulting performance metrics are compared to results obtained by the authors of the ETHOS dataset. In order to explore the remaining underlying issues, model-agnostic method LIME is used to obtain explanations for incorrect predictions.
6.V. Grozdanić, K. Vladimir, G. Delač, M. Šilić (University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia)
Detection of Shilling Attacks on Collaborative Filtering Recommender Systems by Combining Multiple Random Forest Models 
Collaborative filtering recommender systems are one of the essential recommender systems and are still widely used in a combination with other algorithms to make predictions for the users. However, they are vulnerable to shilling attacks, and if there isn't any detection system to prevent those attacks, original recommendations can be heavily influenced to benefit the attackers. Designing attack-resistant recommendation systems is not an easy task, and many researchers have tried to tackle that problem. In this paper, a new approach which combines multiple random forest models is proposed. Each of the random forest models is specialized in detecting one group of the shilling attacks, and then all the models are combined into an ensemble model. Experimental results show that the proposed ensemble is capable of detecting attack profiles at high rate without causing significant bias in the original recommendation system.
7.D. Mileski (Innovation Dooel, Skopje, Macedonia), M. Gusev (Ss. Cyril and Methodius University/Faculty of Computer Science and Engineering, Skopje, Macedonia)
High-Performance Serverless Request Generator: Capable of Generating a Hundred Thousand Requests Per Second 
Load testing of scalable systems is performed with systems that send thousands of requests per second. Depending on the scope of testing, load testing systems are distributed and scalable systems that can use many distributed instances that generate workload. This paper presents a Distributed Serverless Workload Generator, which will be used as a Virtual Patient Generator for testing the ECG Streaming system for monitoring 10.000 patients. For a fully functional workload generator of thousands of requests, the system architecture comprises Serverless services, Pub/Sub, Cloud Scheduler, and Cloud Storage. The paper also presents a price model of this workload generator and its actual cost when it is deployed in a public cloud. A comparison between the load testing system presented in this paper and other workload generators available as SaaS and distributed load testing systems based on servers has been made.
8.A. Periola (Cape Peninsula University of Technology , Cape Town , South Africa), A. Alonge , K. Ogudo (University of Johannesburg, Johannesburg, South Africa)
Future Dynamic Computing and User Profile Execution Considering User Migration in Social Media Contexts 
Abstract— Asymmetricity in user subscription aboard social media platforms leads to a case where emerging social media platforms have more users than existing and established social media platforms. This leads to a scenario where the computing capacity aboard the data centers of established social media platforms are unutilized or under – utilized. In addition, emerging social media platforms are often capital constrained and require access to computing resources to host and execute and user profiles. The presented research proposes a network architecture that enables emerging social media platforms to make use of the idle computing resources of established social media platforms. Analysis shows that the proposed network architecture enhances the power usage effectiveness and computing resource utilization for computing platforms of established social media by an average of 31.2% and (32.8 – 50.5) %, respectively.
Thursday, 5/25/2023 3:00 PM - 5:30 PM,
Leut II, Hotel Admiral, Opatija
3:00 PM - 5:00 PMControl and Automation 
1.D. Miljković (Hrvatska elektroprivreda d.d., Zagreb, Croatia)
Brief Introduction to Active Noise Control 
This paper is a very brief introduction to the field of active noise control. The concept of active noise control based on destructive interference of sounds is presented. The need for high amplitude and phase accuracy of canceling sound for efficient cancellation is shown. Considered is the possibility of noise cancellation in open and closed space. Quiet zone formed in the local noise cancellation related to the noise wavelength is illustrated. Most important approaches used for active noise control such as the feedforward, feedback and hybrid method are laid out. Adaptive feedforward and feedback structures based on adaptive filtering are outlined considering the influence of secondary path transfer function and acoustic feedback. Suitability of particular methods for different sound characteristics like broadband and narrowband noise and availability or absence of the reference signal is considered. Synchronized waveform synthesis is also presented as a solution for active control of periodic noise commonly produced by rotating machinery. General directions for digital implementations such as choice of sampling rate, adaptive filters length, computational cost, circuits for pre-filtering and post-filtering of reference, error, and secondary signal are given. Some typical applications are listed.
2.D. Miljković (Hrvatska elektroprivreda d.d., Zagreb, Croatia)
Active Noise Control Using Waveform Synthesis with Improved Convergence and Tracking 
Active noise control of periodic sounds, as commonly produced by rotating machinery, can be efficiently implemented using a synchronized waveform generator that produces a secondary signal necessary for noise cancellation. Synchronization of synthesized waveform is achieved by the use of reference tacho signal picked up by the magnetic or optical sensor. Waveform adaptation is performed in accordance with the residual noise signal after noise cancelation. Such approach for active noise control works well when a spectrum of offending noise is not changing much. However, during initial machine spool up, varying load and final spool down significant adaptability from waveform synthesizer is required to counteract changes in machine noise. In this paper, methods for improving convergence and tracking properties of waveform generator are presented and analyzed. An adaptive coefficient of waveform generator is modified in accordance to elapsed time during startup, varying frequency of the reference tacho signal and residual noise level. Results of simulations with different noise spectrum corresponding to various machinery rotational speeds are included.
3.T. Špoljarić, B. Vuletić Komljen, L. Cvitaš, G. Malčić (Zagreb University of Applied Sciences, Zagreb, Croatia), J. Matuško (University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia), M. Miletić (Zagreb University of Applied Sciences, Zagreb, Croatia)
Higher-Level Experimental Prototype of a Control Device for Dynamic Beverage-Cooling Process 
In this paper development and testing of an electronic controller for operation of a beverage cooling device is described. This development is a part of a project research that includes a cooling device with a water submerged evaporator and its adequate regulation system. Authors have developed a controller device, mathematical models, and algorithms “from scratch” and in this paper an optimization of a regulation device is described: selection of a smaller and more effective microprocessor, necessary reduction in number of inputs and outputs, new ECB development and an optimized algorithm for the dynamic cooling process.
4.J. Kir Hromatko, Š. Ileš (University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, Croatia), R. Huljev (Genuyn, London, United Kingdom), V. Vučković (Popravak Brodskih Motora d.o.o., Čavle, Croatia)
Artificial Intelligence-based Predictive Fuel Blending Control for Flare Gas Mitigation 
This paper describes a fuel blending algorithm based on artificial intelligence and model predictive control. A gas-fired power plant was modeled using physical laws and on-site measurements. A neural network is used to calculate the methane number and determine the fuel blending ratio limits so that the methane number is within the limits specified by the engine manufacturer. A model predictive controller adjusts the final blending ratio to meet safety requirements and minimize operating costs. The algorithm was tested in both simulations and hardware-in-the-loop experiments.
5.T. Špoljarić, D. Gadže, I. Šulekić (Univerity of Applied Sciences, Zagreb, Zagreb, Croatia)
Program Solution for Regulation and Monitoring of Evaporator Section in Individual Quick-Freezing Process 
In this paper a program solution for control and monitoring of evaporator section in cooling tunnel is proposed. Control strategy for freezing process is designed to work according to temperature probes used for measurement of temperature of frozen goods. Defrost control is designed to work with active probes used for ice thickness measurement thus providing more efficient defrosting of pipeline section in evaporator. Monitoring of certain values of significance is maintained by using pressure observation in condenser circuit and temperature observation expanded to room/tunnel temperature along with probes used in control (goods' temperature). Also, significant actuator states are also monitored: evaporator fan speed and defrosting heater switching state.
6.I. Petrović (Parpar d.o.o., Bjelovar, Croatia), D. Koprivanac, E. Ciriković (Veleučilište u Virovitici, Virovitica, Croatia)
Electrical Profile of User with Photovoltaic System 
Energy and fuel-distribution management performed by suppliers is usually based on daily predictions created by distributors. Those predictions are based on data measured the day before; weather predictions such as ambient temperature, wind, precipitation along with some general information such as day of the week, whether it is weekend or not, are there any preannounced events within the distribution area etc. There's been a rapidly growing trend in demand for micro photovoltaic systems in single user installation for the last few years, which implies that classic user profiles are becoming obsolete for more users, both new and existing ones. This research focuses on a single example representing a consumer with an integrated micro photovoltaic system. In order to keep electrical energy management up-to-date with latest developments, it is essential to create a user profile for consumers with micro photovoltaic systems. This paper focuses on a classic user profile combined with estimated energy production prediction for a micro photovoltaic system, which results in a new profile of a user with a micro photovoltaic system. Measured data will express current user energy consumption as a consumer, compared to the same user energy consumption/production when the consumer is upgraded with a photovoltaic system. With the new profile, it is possible to analyse main characteristics of transformation of the user profile.
5:00 PM - 5:15 PMCroatian Section 
D. Koprivanac (Veleučilište u Virovitici, Virovitica, Croatia), I. Petrović (Parpar d.o.o., Bjelovar, Croatia), I. Heđi (Veleučilište u Virovitici, Virovitica, Croatia)
Analiza rada fotonaponske elektrane radi optimizacije potrošnje energenata 
Klimatske promjene, geopolitička situacija u zadnjih 2 godine, te time izazvana poskupljenja prvenstveno energenata, koncentrirali su planiranje potrošnje energije i energenata u svim područjima planiranja. Tako su se i obnovljivi izvori energije istaknuli kao jedna od alternativa klasičnim skupim izvorima energije. Ovaj rad analizira fotonaponsku elektranu kao jedan primjer obnovljivog izvora energije. Prikazani su mjereni podaci potrošnje i proizvodnje električne energije građevine čija je električna instalacija opremljena fotonaponskom elektranom snage 100 kW u paralelnom radu sa distribucijskom mrežom. Također, prikazani su rezultati indirektno izračunatih parametara i veličina za dva slična dana u uzastopnim godinama čiji su hidrometeorološki podaci slični, te analizirana energetska slika građevine koja se može koristiti u predviđanju proizvodnje fotonaponske elektrane. Zaključci ovog istraživanja omogućavaju lakše planiranje i ekonomično korištenje energenata, ali i planiranje nadogradnje i kombiniranje unutar sustava koji troše energiju i energente.


Basic information:
Chairs:

Vlado Sruk (Croatia), Dejan Škvorc (Croatia)

Program Committee:

Željko Hocenski (Croatia), Leonardo Jelenković (Croatia), Hrvoje Mlinarić (Croatia), Jadranko F. Novak (Croatia), Vlado Sruk (Croatia), Dejan Škvorc (Croatia)

 

Registration / Fees:
REGISTRATION / FEES
Price in EUR
EARLY BIRD
Up to 8 May 2023
REGULAR
From 9 May 2023
Members of MIPRO and IEEE 230 260
Students (undergraduate and graduate), primary and secondary school teachers 120 140
Others 250 280

The discount doesn't apply to PhD students.

Contact:

Vlado Sruk
University of Zagreb
Faculty of Electrical Engineering and Computing 
Unska 3 
HR-10000 Zagreb, Croatia

Phone: +385 1 612 99 45 
Fax: +385 1 612 96 53
E-mail: vlado.sruk@fer.hr


Dejan Škvorc

University of Zagreb
Faculty of Electrical Engineering and Computing 
Unska 3 
HR-10000 Zagreb, Croatia

Phone: +385 1 612 99 43
Fax: +385 1 612 96 53
E-mail: dejan.skvorc@fer.hr
 

The best papers will get a special award.
Accepted papers will be published in the ISSN registered conference proceedings. Presented papers in English will be submitted for inclusion in the IEEE Xplore Digital Library.
.............
There is a possibility that the selected scientific papers with some further modification and refinement are being published in the following journals: Journal of Computing and Information Technology (CIT)MDPI Applied ScienceMDPI Information JournalFrontiers and EAI Endorsed Transaction on Scalable Information Systems.



Location:

Opatija is the leading seaside resort of the Eastern Adriatic and one of the most famous tourist destinations on the Mediterranean. With its aristocratic architecture and style, Opatija has been attracting artists, kings, politicians, scientists, sportsmen, as well as business people, bankers and managers for more than 170 years.

The tourist offer in Opatija includes a vast number of hotels, excellent restaurants, entertainment venues, art festivals, superb modern and classical music concerts, beaches and swimming pools – this city satisfies all wishes and demands.

Opatija, the Queen of the Adriatic, is also one of the most prominent congress cities in the Mediterranean, particularly important for its ICT conventions, one of which is MIPRO, which has been held in Opatija since 1979, and has attracted more than a thousand participants from over forty countries. These conventions promote Opatija as one of the most desirable technological, business, educational and scientific centers in South-eastern Europe and the European Union in general.


For more details, please visit www.opatija.hr and visitopatija.com.

Download
 
News about event
Currently there are no news
 
Patrons - random
Končar Elektroindustrija ZagrebENT ZagrebA1 HrvatskaEkonomski fakultet RijekaTehničko veleučilište u Zagrebu