Wyniki wyszukiwania - BazTech

1

When to Trust AI: Advances and Challenges for Certification of Neural Networks

Kwiatkowska Marta, Zhang Xiyue

Annals of Computer Science and Information Systems

|

2023

|

Vol. 35

25--37

EN

Artificial intelligence (AI) has been advancing at a fast pace and it is now poised for deployment in a wide range of applications, such as autonomous systems, medical diagnosis and natural language processing. Early adoption of AI technology for real-world applications has not been without problems, particularly for neural networks, which may be unstable and susceptible to adversarial examples. In the longer term, appropriate safety assurance techniques need to be developed to reduce potential harm due to avoidable system failures and ensure trustworthiness. Focusing on certification and explainability, this paper provides an overview of techniques that have been developed to ensure safety of AI decisions and discusses future challenges.

2

Useful energy prediction model of a Lithium-ion cell operating on various duty cycles

Burzyński Damian

Eksploatacja i Niezawodność

|

2022

|

Vol. 24, no. 2

317--329

EN

The paper deals with the subject of the prediction of useful energy during the cycling of a lithium-ion cell (LIC), using machine learning-based techniques. It was demonstrated that depending on the combination of cycling parameters, the useful energy (RUEc) that can be transferred during a full cycle is variable, and also three different types of evolution of changes in RUEc were identified. The paper presents a new non-parametric RUEc prediction model based on Gaussian process regression. It was proven that the proposed methodology enables the RUEc prediction for LICs discharged, above the depth of discharge, at a level of 70% with an acceptable error, which is confirmed for new load profiles. Furthermore, techniques associated with explainable artificial intelligence were applied to determine the significance of model input parameters – the variable importance method – and to determine the quantitative effect of individual model parameters (their reciprocal interaction) on RUEc – the accumulated local effects model of the first and second order.

3

Superiority of Simplicity: A Lightweight Model for Network Device Workload Prediction

Acker Alexander, Wittkopp Thorsten, Nedelkoski Sasho, Bogatinovski Jasmin, Kao Odej

Annals of Computer Science and Information Systems

|

2020

|

Vol. 21

7--10

EN

The rapid growth and distribution of IT systems increases their complexity and aggravates operation and maintenance. To sustain control over large sets of hosts and the connecting networks, monitoring solutions are employed and constantly enhanced. They collect diverse key performance indicators (KPIs) (e.g. CPU utilization, allocated memory, etc.) and provide detailed information about the system state. Storing such metrics over a period of time naturally raises the motivation of predicting future KPI progress based on past observations. This allows different ahead of time optimizations like anomaly detection or predictive maintenance. Predicting the future progress of KPIs can be defined as a time series forecasting problem. Although, a variety of time series forecasting methods exist, forecasting the progress of IT system KPIs is very hard. First, KPI types like CPU utilization or allocated memory are very different and hard to be modelled by the same model. Second, system components are interconnected and constantly changing due to soft- or firmware updates and hardware modernization. Thus a frequent model retraining or fine-tuning must be expected. Therefore, we propose a lightweight solution for KPI series prediction based on historic observations. It consists of a weighted heterogeneous ensemble method composed of two models - a neural network and a mean predictor. As ensemble method a weighted summation is used, whereby a heuristic is employed to set the weights. The lightweight nature allows to train models individually on each KPI series and makes model retraining feasible when system changes occur. The modelling approach is evaluated on the available FedCSIS 2020 challenge dataset and achieves an overall R^2 score of 0.10 on the preliminary 10\% test data and 0.15 on the complete test data. We publish our code on the following github repository: https://github.com/citlab/fed\_challenge.

4

Short-term air pollution forecasting based on environmental factors and deep learning models

Arsov Mirche, Zdravevski Eftim, Lameski Petre, Corizzo Roberto, Koteli Nikola, Mitreski Kosta, Trajkovik Vladimir

Annals of Computer Science and Information Systems

|

2020

|

Vol. 21

15--22

EN

The effects of air pollution on people, the environment, and the global economy are profound - and often under-recognized. Air pollution is becoming a global problem. Urban areas have dense populations and a high concentration of emission sources: vehicles, buildings, industrial activity, waste, and wastewater. Tackling air pollution is an immediate problem in developing countries, such as North Macedonia, especially in larger urban areas. This paper exploits Recurrent Neural Network (RNN) models with Long Short-Term Memory units to predict the level of PM10 particles in the near future (+3 hours), measured with sensors deployed in different locations in the city of Skopje. Historical air quality measurements data were used to train the models. In order to capture the relation of air pollution and seasonal changes in meteorological conditions, we introduced temperature and humidity data to improve the performance. The accuracy of the models is compared to PM10 concentration forecast using an Autoregressive Integrated Moving Average (ARIMA) model. The obtained results show that specific deep learning models consistently outperform the ARIMA model, particularly when combining meteorological and air pollution historical data. The benefit of the proposed models for reliable predictions of only 0.01 MSE could facilitate preemptive actions to reduce air pollution, such as temporarily shutting main polluters, or issuing warnings so the citizens can go to a safer environment and minimize exposure.

5

Network Device Workload Prediction: A Data Mining Challenge at Knowledge Pit

Janusz Andrzej, Przyborowski Mateusz, Biczyk Piotr, Ślęzak Dominik

Annals of Computer Science and Information Systems

|

2020

|

Vol. 21

77--80

EN

FedCSIS 2020 Data Mining Challenge: Network Device Workload Prediction was the seventh edition of the international data mining competition organized at Knowledge Pit, in association with the Conference on Computer Science and Information Systems. The main goal was to answer the question of whether it is possible to reliably predict workload-related characteristics of monitored network devices based on historical readings. We describe the scope and explain the motivation for this challenge. We also analyze solutions uploaded by the most successful participants and investigate prediction errors which had the greatest influence on the results. Finally, we describe our baseline solution to the considered problem, which turned out to be the most reliable in the final evaluation.

6

Data Mining-Based Phishing Detection

Bohacik Jan, Skula Ivan, Zabovsky Michal

Annals of Computer Science and Information Systems

|

2020

|

Vol. 21

27--30

EN

Webpages can be faked easily nowadays and as there are many internet users, it is not hard to find some becoming victims of them. Simultaneously, it is not uncommon these days that more and more activities such as banking and shopping are being moved to the internet, which may lead to huge financial losses. In this paper, a developed Chrome plugin for data mining-based detection of phishing webpages is described. The plugin is written in JavaScript and it uses a C4.5 decision tree model created on the basis of collected data with eight describing attributes. The usability of the model is validated with 10-fold cross-validation and the computation of sensitivity, specificity and overall accuracy. The achieved results of experiments are promising.

7

On graph mining with deep learning: introducing model r for link weight prediction

Hou Yuchen, Holder Lawrence B.

Journal of Artificial Intelligence and Soft Computing Research

|

2019

|

Vol. 9, No. 1

21--40

EN

Deep learning has been successful in various domains including image recognition, speech recognition and natural language processing. However, the research on its application in graph mining is still in an early stage. Here we present Model R, a neural network model created to provide a deep learning approach to the link weight prediction problem. This model uses a node embedding technique that extracts node embeddings (knowledge of nodes) from the known links’ weights (relations between nodes) and uses this knowledge to predict the unknown links’ weights. We demonstrate the power of Model R through experiments and compare it with the stochastic block model and its derivatives. Model R shows that deep learning can be successfully applied to link weight prediction and it outperforms stochastic block model and its derivatives by up to 73% in terms of prediction accuracy. We analyze the node embeddings to confirm that closeness in embedding space correlates with stronger relationships as measured by the link weight. We anticipate this new approach will provide effective solutions to more graph mining tasks

8

A HPC (High Performance Computing) based model for optimisation of the waste management systems in metropolitan regions

Gaska K., Generowicz A., Ciuła J.

Technical Transactions

|

2018

|

Vol. 115, iss. 3

153--163

EN

The article presents the conception of an intelligent system for monitoring and managing the municipal waste disposal in metropolises. Applying advanced IT solutions using intelligent computational techniques enables the passage from the passive position of selfgovernment units (JST) in managing the waste disposal to the active position, especially in decision making during the problem solving of planning systems associated with the organisation management of the complex infrastructure of the waste disposal. The aim of using ICT systems is an increase in the reliability of the economy of systemic waste, monitoring in real time, the stabilization of the work of the system and the optimization of logistic and technological processes in the context of the raw material, energy application and simultaneously limiting the influence on all components of the environment.

PL

W artykule przedstawiono koncepcję inteligentnego systemu monitorowania i zarządzania gospodarką odpadami komunalnymi w metropoliach. Zastosowanie zaawansowanych rozwiązań informatycznych wykorzystujących inteligentne techniki obliczeniowe umożliwia na przejście z biernej pozycji jednostek samorządowych (JST) w zarządzaniu gospodarką odpadami do aktywnego działania, w tym szczególnie, podejmowanie decyzji podczas rozwiązywania problemów planistycznych związanych z organizacją systemu zbiórki i systemu transportu odpadów i kompleksowym zarządzaniem złożoną infrastrukturą systemów gospodarki odpadami. Celem wykorzystania systemów ICT jest zwiększenie niezawodności systemów gospodarki odpadami, monitoring w czasie rzeczywistym, stabilizacja pracy systemu oraz optymalizacja procesów logistycznych i technologicznych w kontekście wykorzystania surowcowego, energetycznego przy jednoczesnym ograniczeniu wpływu na wszystkie komponenty środowiska (woda, powietrze, gleba).

9

Model adaptacyjnego algorytmu wspomagania decyzji w systemie utrzymania ruchu

Kosicka E., Mazurkiewicz D., Gola A.

Zarządzanie Przedsiębiorstwem

|

2016

|

Vol. 19, nr 4

29--35

EN

Currently used predictive maintenance systems predict future events by monitoring residual processes using the enforced predictive model. Despite the benefits resulting from their implementation in companies (e.g. savings resulting from preventing failure), it is necessary to draw attention to the fact that such models lack flexibility in adapting to the dynamically changing values of observation vectors due to real-time readout which can provide more accurate predictions. The paper proposes a model of adaptive algorithm for maintenance decision support system which - depending on the changing parameters of residual processes - selects an adequate mathematical model based on predictive and in-formative criteria. Moreover, to produce more accurate predictions this model uses additional input data for prediction including values of residual processes as well as technical or quality-related aspects due to the extended range of observed factors that affect failure occurrence. The proposed model additionally contains a maintenance decision-related part which - based on the information about actions taken by maintenance services - generates a constrained optimal time interval for performing the necessary maintenance work.

10

Wpływ wstępnego przetwarzania danych wejściowych na jakość modeli predykcyjnych budowanych technikami Data Mining

Nęcka K.

Śląskie Wiadomości Elektryczne

|

2014

|

Nr 2 (113)

11--17

PL

Artykuł porusza problematykę wstępnego przetwarzania danych wejściowych wykorzystywanych do prognozowania godzinowego zapotrzebowania na energię elektryczną. Analizy zostały przeprowadzone na danych uzyskanych w wyniku badań własnych wykonanych w ubojni drobiu. Zaprezentowane w pracy wyniki dotyczą przykładowych prognoz godzinowego zapotrzebowania na energię elektryczną wykonanych technikami data mining z wykorzystaniem zmiennych wejściowych poddanych różnym przekształceniom..

EN

This article contains issue of data pre-processing used in prediction of hourly energy consumption. All analyses and studies were done based on own researching made in poultry abattoir and data achieved this way. Results presented in this article applies to hourly prediction for energy consumption achieved with Data Mining techniques with utilization of input variables subjected to various transformations..

11

A predictive model of daily seismic activity induced by mining, developed with data mining methods

Jakubowski J.

Geoinformatica Polonica

|

2014

|

T. 13

8--19

PL

W artykule przedstawiono budowę i ocenę predykcyjnego modelu klasyfikacyjnego dobowej energii sejsmicznej indukowanej eksploatacją ścianową węgla w partii XVI kopalni Piast. Model wykorzystuje dane o energii wstrząsów oraz podstawowe dane charakteryzujące front eksploatacji i wydobycie w tej partii w okresie od lipca 1987 do marca 2011. Dwustanową zmienną prognozowaną jest wystąpienie dobowej sumy energii wstrząsów w ścianie większej lub równej wartości progowej 105 J. Zastosowano trzy metody analityczne data mining: regresję logistyczną, sieci neuronowe i drzewa wzmacniane. Jako najlepszy dla celów prognozy wybrano model drzew wzmacnianych. Wyniki na próbie walidacyjnej pokazują jego wysoką zdolność predykcyjną, biorąc pod uwagę złożoność zjawiska. Wskazywać to może na przydatność zastosowanego modelu do konstrukcji bieżącej, krótkookresowej prognozy zagrożenia sejsmicznego.

12

Predictive regression models of monthly seismic energy emissions induced by longwall mining

Jakubowski J., Tajduś A.

Archives of Mining Sciences

|

2014

|

Vol. 59, no. 3

705--720

EN

This article presents the development and validation of predictive regression models of longwall mining-induced seismicity, based on observations in 63 longwalls, in 12 seams, in the Bielszowice colliery in the Upper Silesian Coal Basin, which took place between 1992 and 2012. A predicted variable is the logarithm of the monthly sum of seismic energy induced in a longwall area. The set of predictors include seven quantitative and qualitative variables describing some mining and geological conditions and earlier seismicity in longwalls. Two machine learning methods have been used to develop the models: boosted regression trees and neural networks. Two types of model validation have been applied: on a random validation sample and on a time-based validation sample. The set of a few selected variables enabled nonlinear regression models to be built which gave relatively small prediction errors, taking the complex and strongly stochastic nature of the phenomenon into account. The article presents both the models of periodic forecasting for the following month as well as long-term forecasting.

PL

W artykule przedstawiono budowę i walidację predykcyjnych modeli regresyjnych sejsmiczności indukowanej eksploatacją w ścianie, opartych na obserwacjach w 63 ścianach kopalni Bielszowice prowadzonych w 12 pokładach w latach 1992-2012. Zmienna prognozowaną jest logarytm miesięcznej sumy energii sejsmicznej wstrząsów w ścianie. Zestaw predyktorów składa się z siedmiu zmiennych ilościowych i jakościowych opisujących wybrane czynniki górnicze i geologiczne w ścianach. Do budowy modeli zastosowano dwie metody uczenia się maszyn: drzewa wzmacniane oraz sieci neuronowe. Zastosowano dwa rodzaje walidacji modeli: na losowej próbie walidacyjnej oraz na czasowej próbie walidacyjnej. Zestaw kilku wybranych zmiennych pozwolił na zbudowanie nieliniowych modeli regresyjnych, które, biorąc pod uwagę złożoną i silnie stochastyczną naturę zjawiska, dają względnie małe błędy pro gnozy. W artykule przedstawiono zarówno modele do prognozy okresowej na kolejny miesiąc jak i do prognozy długoterminowej.