Search
Search results for: PHYSICS-GUIDED NEURAL NETWORKS
-
Physics-guided neural networks (PGNNs) to solve differential equations for spatial analysis
PublicationNumerous examples of physically unjustified neural networks, despite satisfactory performance, generate contradictions with logic and lead to many inaccuracies in the final applications. One of the methods to justify the typical black-box model already at the training stage and lead to many inaccuracies in the final applications. One of the methods to justify the typical black-box model already at the training stage involves extending...
-
Global Surrogate Modeling by Neural Network-Based Model Uncertainty
PublicationThis work proposes a novel adaptive global surrogate modeling algorithm which uses two neural networks, one for prediction and the other for the model uncertainty. Specifically, the algorithm proceeds in cycles and adaptively enhances the neural network-based surrogate model by selecting the next sampling points guided by an auxiliary neural network approximation of the spatial error. The proposed algorithm is tested numerically...
-
Iterative Global Sensitivity Analysis Algorithm with Neural Network Surrogate Modeling
PublicationGlobal sensitivity analysis (GSA) is a method to quantify the effect of the input parameters on outputs of physics-based systems. Performing GSA can be challenging due to the combined effect of the high computational cost of each individual physics-based model, a large number of input parameters, and the need to perform repetitive model evaluations. To reduce this cost, neural networks (NNs) are used to replace the expensive physics-based...
-
Neural Oscillation During Mental Imagery in Sport: An Olympic Sailor Case Study
PublicationThe purpose of the current study was to examine the cortical correlates of imagery depending on instructional modality (guided vs. self-produced) using various sports-related scripts. According to the expert-performance approach, we took an idiosyncratic perspective analyzing the mental imagery of an experienced two-time Olympic athlete to verify whether different instructional modalities of imagery (i.e., guided vs. self-produced)...
-
Deep neural networks for data analysis 24/25
e-Learning CoursesThis course covers introduction to supervised machine learning, construction of basic artificial deep neural networks (DNNs) and basic training algorithms, as well as the overview of popular DNNs architectures (convolutional networks, recurrent networks, transformers). The course introduces students to popular regularization techniques for deep models. Besides theory, large part of the course is the project in which students apply...
-
Olgun Aydin dr
PeopleOlgun Aydin finished his PhD by publishing a thesis about Deep Neural Networks. He works as a Principal Machine Learning Engineer in Nike, and works as Assistant Professor in Gdansk University of Technology in Poland. Dr. Aydin is part of editorial board of "Journal of Artificial Intelligence and Data Science" Dr. Aydin served as Vice-Chairman of Why R? Foundation and is member of Polish Artificial Intelligence Society. Olgun is...
-
Neural networks and deep learning
PublicationIn this chapter we will provide the general and fundamental background related to Neural Networks and Deep Learning techniques. Specifically, we divide the fundamentals of deep learning in three parts, the first one introduces Deep Feed Forward Networks and the main training algorithms in the context of optimization. The second part covers Convolutional Neural Networks (CNN) and discusses their main advantages and shortcomings...
-
NEURAL NETWORKS
Journals -
Deep neural networks for data analysis
e-Learning CoursesThe aim of the course is to familiarize students with the methods of deep learning for advanced data analysis. Typical areas of application of these types of methods include: image classification, speech recognition and natural language understanding. Celem przedmiotu jest zapoznanie studentów z metodami głębokiego uczenia maszynowego na potrzeby zaawansowanej analizy danych. Do typowych obszarów zastosowań tego typu metod należą:...
-
A survey of neural networks usage for intrusion detection systems
PublicationIn recent years, advancements in the field of the artificial intelligence (AI) gained a huge momentum due to the worldwide appliance of this technology by the industry. One of the crucial areas of AI are neural networks (NN), which enable commer‐ cial utilization of functionalities previously not accessible by usage of computers. Intrusion detection system (IDS) presents one of the domains in which neural networks are widely tested...
-
Marine and Cosmic Inspirations for AI Algorithms
PublicationArtificial Intelligence (AI) is a scientific area that currently sees an enormous growth. Various new algorithms and methods are developed and many of them meets practical, successful applications. Authors of new algorithms draw different inspirations. Probably the most common one is the nature. For example, Artificial Neural Networks were inspired by the structure of human brain and nervous system while the classic Genetic Algorithm...
-
Performance Analysis of Convolutional Neural Networks on Embedded Systems
PublicationMachine learning is no longer confined to cloud and high-end server systems and has been successfully deployed on devices that are part of Internet of Things. This paper presents the analysis of performance of convolutional neural networks deployed on an ARM microcontroller. Inference time is measured for different core frequencies, with and without DSP instructions and disabled access to cache. Networks use both real-valued and...
-
Explainable AI for Inspecting Adversarial Attacks on Deep Neural Networks
PublicationDeep Neural Networks (DNN) are state of the art algorithms for image classification. Although significant achievements and perspectives, deep neural networks and accompanying learning algorithms have some important challenges to tackle. However, it appears that it is relatively easy to attack and fool with well-designed input samples called adversarial examples. Adversarial perturba-tions are unnoticeable for humans. Such attacks...
-
Deep neural networks for data analysis 27/28
e-Learning Courses -
Deep neural networks for data analysis 25/26
e-Learning Courses -
Deep neural networks for data analysis 26/27
e-Learning Courses -
Exploring Neural Networks for Musical Instrument Identification in Polyphonic Audio
PublicationThe purpose of this paper is to introduce neural network-based methods that surpass state-of-the-art (SOTA) models, either by training faster or having simpler architecture, while maintaining comparable effectiveness in musical instrument identification in polyphonic music. Several approaches are presented, including two authors’ proposals, i.e., spiking neural networks (SNN) and a modular deep learning model named FMCNN (Fully...
-
Adjusted SpikeProp algorithm for recurrent spiking neural networks with LIF neurons
PublicationA problem related to the development of a supervised learning method for recurrent spiking neural networks is addressed in the paper. The widely used Leaky-Integrate-and-Fire model has been adopted as a spike neuron model. The proposed method is based on a known SpikeProp algorithm. In detail, the developed method enables gradient descent learning of recurrent or multi-layer feedforward spiking neural networks. The research included...
-
Deep neural networks approach to skin lesions classification — A comparative analysis
PublicationThe paper presents the results of research on the use of Deep Neural Networks (DNN) for automatic classification of the skin lesions. The authors have focused on the most effective kind of DNNs for image processing, namely Convolutional Neural Networks (CNN). In particular, three kinds of CNN were analyzed: VGG19, Residual Networks (ResNet) and the hybrid of VGG19 CNN with the Support Vector Machine (SVM). The research was carried...
-
Selected Technical Issues of Deep Neural Networks for Image Classification Purposes
PublicationIn recent years, deep learning and especially Deep Neural Networks (DNN) have obtained amazing performance on a variety of problems, in particular in classification or pattern recognition. Among many kinds of DNNs, the Convolutional Neural Networks (CNN) are most commonly used. However, due to their complexity, there are many problems related but not limited to optimizing network parameters, avoiding overfitting and ensuring good...
-
Robustness in Compressed Neural Networks for Object Detection
PublicationModel compression techniques allow to significantly reduce the computational cost associated with data processing by deep neural networks with only a minor decrease in average accuracy. Simultaneously, reducing the model size may have a large effect on noisy cases or objects belonging to less frequent classes. It is a crucial problem from the perspective of the models' safety, especially for object detection in the autonomous driving...
-
Clothes Detection and Classification Using Convolutional Neural Networks
PublicationIn this paper we describe development of a computer vision system for accurate detection and classification of clothes for e-commerce images. We present a set of experiments on well established architectures of convolutional neural networks, including Residual networks, SqueezeNet and Single Shot MultiBox Detector (SSD). The clothes detection network was trained and tested on DeepFashion dataset, which contains box annotations...
-
Outlier detection method by using deep neural networks
PublicationDetecting outliers in the data set is quite important for building effective predictive models. Consistent prediction can not be made through models created with data sets containing outliers, or robust models can not be created. In such cases, it may be possible to exclude observations that are determined to be outlier from the data set, or to assign less weight to these points of observation than to other points of observation....
-
A MODEL FOR FORECASTING PM10 LEVELS WITH THE USE OF ARTIFICIAL NEURAL NETWORKS
PublicationThis work presents a method of forecasting the level of PM10 with the use of artificial neural networks. Current level of particulate matter and meteorological data was taken into account in the construction of the model (checked the correlation of each variable and the future level of PM10), and unidirectional networks were used to implement it due to their ease of learning. Then, the configuration of the network (built on the...
-
Use of Neural Networks in Diagnostics of Rolling-Element Bearing of the Induction Motor
PublicationBearing defect is statistically the most frequent cause of an induction motor fault. The research described in the paper utilized the phenomenon of the current change in the induction motor with bearing defect. Methods based on the analysis of the supplying current are particularly useful when it is impossible to install diagnostic devices directly on the motor. The presented method of rolling-element bearing diagnostics used indirect...
-
A Novel Iterative Decoding for Iterated Codes Using Classical and Convolutional Neural Networks
PublicationForward error correction is crucial for communication, enabling error rate or required SNR reduction. Longer codes improve correction ratio. Iterated codes offer a solution for constructing long codeswith a simple coder and decoder. However, a basic iterative code decoder cannot fully exploit the code’s potential, as some error patterns within its correction capacity remain uncorrected.We propose two neural network-assisted decoders:...
-
Neural Networks, Support Vector Machine and Genetic Algorithms for Autonomous Underwater Robot Support
PublicationIn this paper, artificial neural networks, a classification technique called support vector machine and meta-heuristics genetic algorithm have been considered for development in autonomous underwater robots. Artificial neural networks have been used for seabed modelling as well as support vector machine has been applied for the obstacles classification to avoid some collision problems. Moreover, genetic algorithm has been applied...
-
An Automated Method for Biometric Handwritten Signature Authentication Employing Neural Networks
PublicationHandwriting biometrics applications in e-Security and e-Health are addressed in the course of the conducted research. An automated graphomotor analysis method for the dynamic electronic representation of the handwritten signature authentication was researched. The developed algorithms are based on dynamic analysis of electronically handwritten signatures employing neural networks. The signatures were acquired with the use of the...
-
A Selection of Starting Points for Iterative Position Estimation Algorithms Using Feedforward Neural Networks
PublicationThis article proposes the use of a feedforward neural network (FNN) to select the starting point for the first iteration in well-known iterative location estimation algorithms, with the research objective of finding the minimum size of a neural network that allows iterative position estimation algorithms to converge in an example positioning network. The selected algorithms for iterative position estimation, the structure of the...
-
Supply current signal and artificial neural networks in the induction motor bearings diagnostics
PublicationThis paper contains research results of the diagnostics of induction motor bearings based on measurement of the supply current with usage of artificial neural networks. Bearing failure amount is greater than 40% of all engine failures, which makes their damage-free operation crucial. Tests were performed on motors with intentionally made bearings defects. Chapter 2 introduces the concept of artificial neural networks. It presents...
-
A novel approach exploiting properties of convolutional neural networks for vessel movement anomaly detection and classification
PublicationThe article concerns the automation of vessel movement anomaly detection for maritime and coastal traffic safety services. Deep Learning techniques, specifically Convolutional Neural Networks (CNNs), were used to solve this problem. Three variants of the datasets, containing samples of vessel traffic routes in relation to the prohibited area in the form of a grayscale image, were generated. 1458 convolutional neural networks with...
-
Dynamically positioned ship steering making use of backstepping method and artificial neural networks
PublicationThe article discusses the issue of designing a dynamic ship positioning system making use of the adaptive vectorial backstepping method and RBF type arti cial neural networks. In the article, the backstepping controller is used to determine control laws and neural network weight adaptation laws. e arti cial neural network is applied at each time instant to approximate nonlinear functions containing parametric uncertainties....
-
Hybrid of Neural Networks and Hidden Markov Models as a modern approach to speech recognition systems
PublicationThe aim of this paper is to present a hybrid algorithm that combines the advantages ofartificial neural networks and hidden Markov models in speech recognition for control purpos-es. The scope of the paper includes review of currently used solutions, description and analysis of implementation of selected artificial neural network (NN) structures and hidden Markov mod-els (HMM). The main part of the paper consists of a description...
-
INFLUENCE OF DATA NORMALIZATION ON THE EFFECTIVENESS OF NEURAL NETWORKS APPLIED TO CLASSIFICATION OF PAVEMENT CONDITIONS – CASE STUDY
PublicationIn recent years automatic classification employing machine learning seems to be in high demand for tele-informatic-based solutions. An example of such solutions are intelligent transportation systems (ITS), in which various factors are taken into account. The subject of the study presented is the impact of data pre-processing and normalization on the accuracy and training effectiveness of artificial neural networks in the case...
-
Predicting the impact of traffic–induced vibrations on buildings using artificial neural networks
PublicationTraffic–induced vibrations may constitute a considerable load to a building, cause cracking of plaster, cracks in load–bearing elements or even a global structural collapse of the whole structure [1-4]. Vibrations measurements of real structures are costly and laborious, not justified in all cases. The aim of the paper is to create an original algorithm, to predict the negative dynamic impact on the examined residential building...
-
USING ARTIFICIAL NEURAL NETWORKS FOR PREDICTING SHIP FUEL CONSUMPTION
PublicationIn marine vessel operations, fuel costs are major operating costs which affect the overall profitability of the maritime transport industry. The effective enhancement of using ship fuel will increase ship operation efficiency. Since ship fuel consumption depends on different factors, such as weather, cruising condition, cargo load, and engine condition, it is difficult to assess the fuel consumption pattern for various types...
-
The application of neural networks in forecasting the influence of traffic-induced vibrations on residential buildings
PublicationTraffic-induced vibrations may cause the cracking of plaster, damage to structural elements and, in extreme cases, may even lead to the structural collapse of residential buildings. The aim of this article is to analyse the effectiveness of a method of forecasting the impact of vibrations on residential buildings using the concept of artificial intelligence. The article presents several alternative forecasting systems for which...
-
International Journal of Neural Networks
Journals -
IEEE TRANSACTIONS ON NEURAL NETWORKS
Journals -
An Intelligent Approach to Short-Term Wind Power Prediction Using Deep Neural Networks
PublicationIn this paper, an intelligent approach to the Short-Term Wind Power Prediction (STWPP) problem is considered, with the use of various types of Deep Neural Networks (DNNs). The impact of the prediction time horizon length on accuracy, and the influence of temperature on prediction effectiveness have been analyzed. Three types of DNNs have been implemented and tested, including: CNN (Convolutional Neural Networks), GRU (Gated Recurrent...
-
Application 2D Descriptors and Artificial Neural Networks for Beta-Glucosidase Inhibitors Screening
PublicationBeta-glucosidase inhibitors play important medical and biological roles. In this study, simple two-variable artificial neural network (ANN) classification models were developed for beta-glucosidase inhibitors screening. All bioassay data were obtained from the ChEMBL database. The classifiers were generated using 2D molecular descriptors and the data miner tool available in the STATISTICA package (STATISTICA Automated Neural...
-
System for monitoring road slippery based on CCTV cameras and convolutional neural networks
PublicationThe slipperiness of the surface is essential for road safety. The growing number of CCTV cameras opens the possibility of using them to automatically detect the slippery surface and inform road users about it. This paper presents a system of developed intelligent road signs, including a detector based on convolutional neural networks (CNNs) and the transferlearning method employed to the processing of images acquired with video...
-
Multi-Camera Vehicle Tracking Using Local Image Features and Neural Networks
PublicationA method for tracking moving objects crossing fields of view of multiple cameras is presented. The algorithm utilizes Artificial Neural Networks (ANNs). Each ANN is trained to recognize images of one moving object acquired by a single camera. Local image features calculated in the vicinity of automatically detected interest points are used as object image parameters. Next, ANNs are employed to identify the same objects captured...
-
Graph Neural Networks and Structural Information on Ionic Liquids: A Cheminformatics Study on Molecular Physicochemical Property Prediction
PublicationIonic liquids (ILs) provide a promising solution in many industrial applications, such as solvents, absorbents, electrolytes, catalysts, lubricants, and many others. However, due to the enormous variety of their structures, uncovering or designing those with optimal attributes requires expensive and exhaustive simulations and experiments. For these reasons, searching for an efficient theoretical tool for finding the relationship...
-
Application of Artificial Neural Networks to Predict Insulation Properties of Lightweight Concrete
PublicationPredicting the properties of concrete before its design and application process allows for refining and optimizing its composition. However, the properties of lightweight concrete are much harder to predict than those of normal weight concrete, especially if the forecast concerns the insulating properties of concrete with artificial lightweight aggregate (LWA). It is possible to use porous aggregates and precisely modify the composition...
-
Recognition of Emotions in Speech Using Convolutional Neural Networks on Different Datasets
PublicationArtificial Neural Network (ANN) models, specifically Convolutional Neural Networks (CNN), were applied to extract emotions based on spectrograms and mel-spectrograms. This study uses spectrograms and mel-spectrograms to investigate which feature extraction method better represents emotions and how big the differences in efficiency are in this context. The conducted studies demonstrated that mel-spectrograms are a better-suited...
-
Modeling of Surface Roughness in Honing Processes by UsingFuzzy Artificial Neural Networks
PublicationHoning processes are abrasive machining processes which are commonly employed to improve the surface of manufactured parts such as hydraulic or combustion engine cylinders. These processes can be employed to obtain a cross-hatched pattern on the internal surfaces of cylinders. In this present study, fuzzy artificial neural networks are employed for modeling surface roughness parameters obtained in finishing honing operations. As...
-
Approximation of Fractional Order Dynamic Systems Using Elman, GRU and LSTM Neural Networks
PublicationIn the paper, authors explore the possibility of using the recurrent neural networks (RNN) - Elman, GRU and LSTM - for an approximation of the solution of the fractional-orders differential equations. The RNN network parameters are estimated via optimisation with the second order L-BFGS algorithm. It is done based on data from four systems: simple first and second fractional order LTI systems, a system of fractional-order point...
-
Artificial Neural Networks for Comparative Navigation
Publication -
Neural Networks and the Evolution of Environmental Change
PublicationZmiany środowiskowe na Ziemii są odwieczne i liczą około 4 miliardy lat. Homo sapiens wpłynął na każdy aspekt środowiska ziemskiego w wyniku rozwoju ludzkości na przestrzeni ostatnich milionów lat. Ale nic tak nie wpłynęło na wzrost i szybkość zmian na Ziemi jak ludzka aktywność w ciągu ostatnich dwóch stuleci. Po raz pierwszy zmiany ekosystemów były tak intensywne i zachodziły na tka wielką skalę i z taką szybkością jak nigdy...