Informacja

Drogi użytkowniku, aplikacja do prawidłowego działania wymaga obsługi JavaScript. Proszę włącz obsługę JavaScript w Twojej przeglądarce.

Wyszukujesz frazę "deep neural networks" wg kryterium: Temat


Tytuł:
Analiza treningu i własności głębokich sieci neuronowych typu GAN (Generative Adversarial Networks)
Analysis of training and properties of deep neural networks of the GAN type (Generative Adversarial Networks)
Autorzy:
Kadłubowski, Paweł
Opis:
The diploma thesis focused on the analysis of training and properties of deep neural networks ofthe GAN type. The impact of various parameters on the quality of images generated by the DCGANnetwork was investigated. To accomplish this, the CelebA dataset with varying amounts of data,different architectures, and optimizers was utilized. The experiments involved the followingoptimizers: Adam, NAdam, RAdam, AdaMax, AdaGrad, as well as two distinct network architecturesdiffering in the depth of the feature map. During the experiment, a CelebA dataset was prepared, consisting of 202554 color images ofcelebrity faces with a resolution of 178x218 pixels. Three subsets were created from this dataset,containing 13453, 45073 and 202554 images, respectively. The resolution of these images wasreduced to 64x64 pixels for further analysis. For three subsets of different sizes, a neural network training process was conducted, leading to thegeneration of artificially generated images. Upon completion of the network training process, a comparative analysis of the generated imageswas conducted, and graphs depicting the relationship between loss and the number of trainingiterations were generated. Next, a comparison of the quality of generated images was conducted using two different types ofnetworks: WGAN and DCGAN. The CelebA dataset, containing 200000 celebrity face images, wasdivided into three subsets, each containing 10000, 45000, and 200000 color images, all with aresolution of 64x64 pixels. For each of the three subsets of data, a training process was conducted for both WGAN andDCGAN networks to generate images. Finally, a comparison of the quality of the generated imageswas performed. The next experiment aimed to investigate how the length of the input noise vector affects thequality and diversity of generated images using the MNIST dataset, which contains 60000 handwritten digits. The experiment was conducted for various values of the noise vector length (nz), generating atotal of 10,000 images for each of these values. Based on these images, a tensor representing theaverage image was calculated, and the variance was computed. Subsequently, a chart illustrating therelationship between variance and the length of the noise vector was created.
Praca dyplomowa dotyczyła analizy treningu i własności głębokich sieci neuronowych typu GAN.Badano wpływ różnych parametrów na jakość obrazów generowanych przez sieć DCGAN.W tym celu wykorzystano zbiór CelebA o zróżnicowanej liczbie danych, różnych architekturachi optymalizatorach. Eksperymenty obejmowały optymalizatory: Adama, NAdama, RAdama, AdaMax,AdaGrad oraz dwie różne architektury, które różniły się głębokością mapy cech. W trakcie doświadczenia przygotowano zbiór CelebA, składający się z 202554 kolorowychobrazów twarzy celebrytów o rozdzielczości 178x218 pikseli. Na podstawie tego zbioru stworzonotrzy podzbiory zawierające odpowiednio: 13453, 45073 i 202554 obrazów. Rozdzielczość zdjęć wtych podzbiorach została zmniejszona do 64x64 pikseli w celu dalszej analizy.Dla trzech podzbiorów o różnych rozmiarach, przeprowadzono proces uczenia sieci neuronowej,który doprowadził do uzyskania sztucznie wygenerowanych obrazków. Po zakończonym procesie uczenia sieci przeprowadzono analizę porównawczą generowanychobrazów oraz wygenerowano wykresy, przedstawiające zależność straty (Loss) od liczby iteracjitreningowych.Następnie przeprowadzono porównanie jakości generowanych obrazów przy użyciu dwóchróżnych typów sieci: WGAN oraz DCGAN. Zbiór CelebA zawierający 200000 zdjęć twarzycelebrytów, został podzielony na trzy podzbiory zawierające odpowiednio: 10000, 45000 i 200000kolorowych obrazków, wszystkie o rozdzielczości 64x64 piksele.Dla każdego z trzech podzbiorów danych przeprowadzono proces uczenia zarówno dla sieciWGAN, jak i DCGAN, w celu generacji obrazów. Na zakończenie przeprowadzono porównaniejakości generowanych obrazków.Kolejnym eksperymentem było zbadanie, jak długość wektora szumu wejściowego wpływa najakość i różnorodność generowanych obrazów przy użyciu zbioru danych MNIST, który zawiera60000 ręcznie napisanych cyfr.Doświadczenie przeprowadzono dla różnych wartości długości wektora szumu (nz), generującłącznie 10000 obrazków dla każdej z tych wartości. Na podstawie tych obrazów obliczono tensorreprezentujący średni obrazek oraz obliczono wariancję. Następnie stworzono wykres ilustrującyzależność między wariancją, a długością wektora szumu.
Dostawca treści:
Repozytorium Uniwersytetu Jagiellońskiego
Inne
Tytuł:
Automatic Fault Classification for Journal Bearings Using ANN and DNN
Autorzy:
Narendiranath Babu, T.
Aravind, A.
Rakesh, A.
Jahzan, M.
Rama Prabha, D.
Ramalinga Viswanathan, M.
Tematy:
journal bearing
fault classification
artificial neural networks
deep neural networks
Pokaż więcej
Wydawca:
Polska Akademia Nauk. Czytelnia Czasopism PAN
Powiązania:
https://bibliotekanauki.pl/articles/177579.pdf  Link otwiera się w nowym oknie
Opis:
Journal bearings are the most common type of bearings in which a shaft freely rotates in a metallic sleeve. They find a lot of applications in industry, especially where extremely high loads are involved. Proper analysis of the various bearing faults and predicting the modes of failure beforehand are Essentials to increase the working life of the bearing. In the current study, the vibration data of a journal Bering in the healthy condition and in five different fault conditions are collected. A feature extraction metod is employed to classify the different fault conditions. Automatic fault classification is performed using artificial neural networks (ANN). As the probability of a correct prediction goes down for a higher number of faults in ANN, the method is made more robust by incorporating deep neural networks (DNN) with the help of autoencoders. Training was done using the scaled conjugate gradient algorithm and the performance was calculated by the cross entropy method. Due to the increased number of hidden layers in DNN, it is possible to achieve a high efficiency of 100% with the feature extraction method.
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Track finding with Deep Neural Networks
Autorzy:
Kucharczyk, Marcin
Wolter, Marcin
Tematy:
deep neural networks
machine learning
tracking
HEP
Pokaż więcej
Wydawca:
Akademia Górniczo-Hutnicza im. Stanisława Staszica w Krakowie. Wydawnictwo AGH
Powiązania:
https://bibliotekanauki.pl/articles/305791.pdf  Link otwiera się w nowym oknie
Opis:
High energy physics experiments require fast and efficient methods for reconstructing the tracks of charged particles. The commonly used algorithms are sequential and the required CPU power increases rapidly with the number of tracks. Neural networks can speed up the process due to their capability of modeling complex non-linear data dependencies and finding all tracks in parallel. In this paper, we describe the application of the deep neural network for reconstructing straight tracks in a toy two-dimensional model. It is planned to apply this method to the experimental data obtained by the MUonE experiment at CERN.
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Low-cost, low-resolution IR system with super-resolution interpolation of thermal images for industrial applications
Autorzy:
Więcek, P.
Sankowski, D.
Tematy:
Super-resolution
residual deep neural networks
image interpolation
Pokaż więcej
Wydawca:
Stowarzyszenie Inżynierów i Techników Mechaników Polskich
Powiązania:
https://bibliotekanauki.pl/articles/114076.pdf  Link otwiera się w nowym oknie
Opis:
In this paper authors present application of deep neural networks for super-resolution interpolation of infrared images. A residual neural network with reduced number of layers was used. The transfer learning using RGB visual images was applied in this research. The validation of the network was performed for 32×24 and 160×120 pixels infrared images, with the up-sampling scale factors 2, 3, 4, 5 and 6. Monitoring of high temperature industrial processes like inductive heating and thermal hardening is the main application of proposed methods.
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Applying a neural network ensemble to intrusion detection
Autorzy:
Ludwig, Simone A.
Tematy:
ensemble learning
Deep Neural Networks
NSL-KDD data set
Pokaż więcej
Wydawca:
Społeczna Akademia Nauk w Łodzi. Polskie Towarzystwo Sieci Neuronowych
Powiązania:
https://bibliotekanauki.pl/articles/91620.pdf  Link otwiera się w nowym oknie
Opis:
An intrusion detection system (IDS) is an important feature to employ in order to protect a system against network attacks. An IDS monitors the activity within a network of connected computers as to analyze the activity of intrusive patterns. In the event of an ‘attack’, the system has to respond appropriately. Different machine learning techniques have been applied in the past. These techniques fall either into the clustering or the classification category. In this paper, the classification method is used whereby a neural network ensemble method is employed to classify the different types of attacks. The neural network ensemble method consists of an autoencoder, a deep belief neural network, a deep neural network, and an extreme learning machine. The data used for the investigation is the NSL-KDD data set. In particular, the detection rate and false alarm rate among other measures (confusion matrix, classification accuracy, and AUC) of the implemented neural network ensemble are evaluated.
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Application of EMD ANN and DNN for Self-Aligning Bearing Fault Diagnosis
Autorzy:
Narendiranath, B. T.
Aravind, A.
Rakesh, A.
Jahzan, M.
Rama, P. D.
Tematy:
self-aligning bearing
fault classification
artificial neural networks
deep neural networks
Pokaż więcej
Wydawca:
Polska Akademia Nauk. Czytelnia Czasopism PAN
Powiązania:
https://bibliotekanauki.pl/articles/176889.pdf  Link otwiera się w nowym oknie
Opis:
Self-aligning roller bearings are an integral part of the industrial machinery. The proper analysis and prediction of the various faults that may happen to the bearing beforehand contributes to an increase in the working life of the bearing. This study aims at developing a novel method for the analysis of the various faults in self-aligning bearings as well as the automatic classification of faults using artificial neural network (ANN) and deep neural network (DNN). The vibration data is collected for six different faults as well as for the healthy bearing. Empirical mode decomposition (EMD) followed by Hilbert Huang transform is used to extract instantaneous frequency peaks which are used for fault analysis. Time domain and time-frequency domain features are then extracted which are used to implement the neural networks through the pattern recognition tool in MATLAB. A comparative study of the outputs from the two neural networks is also performed. From the confusion matrix, the efficiency of the ANN has been found to be 95.7% and using DNN has been found to be 100%.
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Petrov-Galerkin formulation equivallent to the residual minimization method for finding an optimal test function
Autorzy:
Paszyński, Maciej R.
Służalec, Tomasz
Opis:
Numerical solutions of Partial Differential Equations with Finite Element Method have multiple applications in science and engineering. Several challenging problems require special stabilization methods to deliver accurate results of the numerical simulations. The advection-dominated diffusion problem is an example of such problems. They are employed to model pollution propagation in the atmosphere. Unstable numerical methods generate unphysical oscillations, and they make no physical sense. Obtaining accurate and stable numerical simulations is difficult, and the method of stabilization depends on the parameters of the partial differential equations. They require a deep knowledge of an expert in the field of numerical analysis. We propose a method to construct and train an artificial expert in stabilizing numerical simulations based on partial differential equations. We create a neural network-driven artificial intelligence that makes decisions about the method of stabilizing computer simulations. It will automatically stabilize difficult numerical simulations in a linear computational cost by generating the optimal test functions. These test functions can be utilized for building an unconditionally stable system of linear equations. The optimal test functions proposed by artificial intelligence will not depend on the right-hand side, and thus they may be utilized in a large class of PDE-based simulations with different forcing and boundary conditions. We test our method on the model one-dimensional advection-dominated diffusion problem.
Dostawca treści:
Repozytorium Uniwersytetu Jagiellońskiego
Artykuł
Tytuł:
TSProto : fusing deep feature extraction with interpretable glass-box surrogate model for explainable time-series classification
Autorzy:
Bobek, Szymon
Nalepa, Grzegorz
Opis:
Deep neural networks (DNNs) are highly effective at extracting features from complex data types, such as images and text, but often function as black-box models, making interpretation difficult. We propose TSProto – a model-agnostic approach that goes beyond standard XAI methods focused on feature importance, clustering important segments into conceptual prototypes—high-level, human-interpretable units. This approach not only enhances transparency but also avoids issues seen with surrogate models, such as the Rashomon effect, enabling more direct insights into DNN behavior. Our method involves two phases: (1) using feature attribution tools (e.g., SHAP, LIME) to highlight regions of model importance, and (2) fusion of these regions into prototypes with contextual information to form meaningful concepts. These concepts then integrate into an interpretable decision tree, making DNNs more accessible for expert analysis. We benchmark our solution on 61 publicly available datasets, where it outperforms other state-of-the-art prototype-based methods and glassbox models by an average of 10% in the F1 metric. Additionally, we demonstrate its practical applicability in a real-life anomaly detection case. The results from the user evaluation, conducted with 17 experts recruited from leading European research teams and industrial partners, also indicate a positive reception among experts in XAI and the industry. Our implementation is available as an open-source Python package on GitHub and PyPi.
Dostawca treści:
Repozytorium Uniwersytetu Jagiellońskiego
Artykuł
Tytuł:
Cooling fan controlled by embedded vision system
Autorzy:
Kula, Sebastian
Tematy:
computer vision
deep neural networks
electromechanical systems
human computer interaction
Pokaż więcej
Wydawca:
Politechnika Poznańska. Wydawnictwo Politechniki Poznańskiej
Powiązania:
https://bibliotekanauki.pl/articles/376210.pdf  Link otwiera się w nowym oknie
Opis:
The HMI (human machine interaction) systems are widely used to control machines and variety of devices. Currently the HMI solutions, based on touch screens are almost commonly used in many domains, however the number of devices, which interaction with the user is based on speech recognition or user gesture recognition increases systematically. The paper focuses on the electromechanical system, which applies gestures and handwritten digits to control the speed of the DC cooling fan. The system crucial elements are the AVR microcontroller and the developer board, equipped with the embedded supercomputer NVIDIA Jetson TX1. To create the software part of the system artificial intelligence algorithms and deep neural networks were applied. The paper describes the complete routine of data preprocessing, deep neural network training and testing with the use of the GPU Tesla K20 and with the use of the DIGITS (Deep Learning GPU Training System), deployment of the trained model on Jetson TX1 board and the system execution. The system enables to control the fan through the two gestures (“stone”, ”paper”) or through four handwritten digits.
Dostawca treści:
Biblioteka Nauki
Artykuł
Tytuł:
Power-Ground Plane Impedance Modeling Using Deep Neural Networks and an Adaptive Sampling Process
Autorzy:
Goay, Chan Hong
Cheong, Zheng Quan
Low, Chen En
Ahmad, Nur Syazreen
Goh, Patrick
Tematy:
adaptive sampling
deep neural networks
deep learning
power-ground plane
Z-parameters
Pokaż więcej
Wydawca:
Polska Akademia Nauk. Czytelnia Czasopism PAN
Powiązania:
https://bibliotekanauki.pl/articles/2200709.pdf  Link otwiera się w nowym oknie
Opis:
This paper proposes a deep neural network (DNN) based method for the purpose of power-ground plane impedance modeling. A composite DNN model, which is a combination of two DNNs is used to predict the Z-parameters of power ground planes from their design parameters. The first DNN predicts the normalized Z-parameters whereas the second DNN predicts the original maximum and minimum values of the nonnormalized Z-parameters. This allows the method to retain a high accuracy when predicting responses that have large variations across designs, as is the case with the Z-parameters of the power-ground planes. We use the adaptive sampling algorithm to generate the training and validation samples for the DNNs. The adaptive sampling algorithm starts with only a few samples, then slowly generates more samples in the non-linear regions within the design parameters space. The level of non-linearity of the regions is determined by a surrogate model which is also trained using the generated samples as well. If the surrogate model has poor prediction accuracy in a region, then the adaptive sampling algorithm will generate more samples in that region. A shallow neural network is used as the surrogate model for non-linearity determination of the regions since it is faster to train and update. Once all the samples have been generated, they will be used to train and validate the composite DNN models. Finally, we present two examples, a square-shaped power ground plane and a squareshaped power ground plane with a hollow square at the center to demonstrate the robustness of the DNN composite models.
Dostawca treści:
Biblioteka Nauki
Artykuł

Ta witryna wykorzystuje pliki cookies do przechowywania informacji na Twoim komputerze. Pliki cookies stosujemy w celu świadczenia usług na najwyższym poziomie, w tym w sposób dostosowany do indywidualnych potrzeb. Korzystanie z witryny bez zmiany ustawień dotyczących cookies oznacza, że będą one zamieszczane w Twoim komputerze. W każdym momencie możesz dokonać zmiany ustawień dotyczących cookies