Wyniki wyszukiwania - BazTech

Ograniczanie wyników

Znaleziono wyników: 3

Liczba wyników na stronie

Wyniki wyszukiwania

Sortuj według:

Ogranicz wyniki do:

Canine age classification using Deep Learning as a step towards preventive medicine in animals

Mazurek Szymon, Wielgosz Maciej, Caputa Jakub, Frączek Rafał, Karwatowski Michał, Grzeszczyk Jakub, Łukasik Daria, Śmiech Anna, Russek Paweł, Jamro Ernest, Dąbrowska-Boruch Agnieszka, Pietroń Marcin, Sebastian Koryciak, Wiatr Kazimierz

Annals of Computer Science and Information Systems

2022

Vol. 30

169--172

The main goal of this work was to implement a reliable machine learning algorithm that can classify the age of a dog, given only a photograph of its face. The problem, which seems simple for humans, presents itself as very difficult for the machine learning algorithms due to differences of facial features among the dogs population. As convolutional neural networks (CNNs) performed poorly in this problem, authors took other approach of creating novel architecture consisting of combination of CNN and vision transformer (ViT) and examining the age of the dogs separately for every breed. Authors were able to achieve better results than those in initial works covering the problem.

Compressing sentiment analysis CNN models for efficient hardware processing

Wróbel Krzysztof, Karwatowski Michał, Wielgosz Maciej, Pietroń Marcin, Wiatr Kazimierz

Computer Science

2020

T. 21 (1)

25--41

Convolutional neural networks (CNNs) were created for image classification tasks. Shortly after their creation, they were applied to other domains, including natural language processing (NLP). Nowadays, solutions based on artificial intelligence appear on mobile devices and embedded systems, which places constraints on memory and power consumption, among others. Due to CNN memory and computing requirements, it is necessary to compress them in order to be mapped to the hardware. This paper presents the results of the compression of efficient CNNs for sentiment analysis. The main steps involve pruning and quantization. The process of mapping the compressed network to an FPGA and the results of this implementation are described. The conducted simulations showed that the 5-bit width is enough to ensure no drop in accuracy when compared to the floating-point version of the network. Additionally, the memory footprint was significantly reduced (between 85 and 93% as compared to the original model).

Using neural networks with data quantization for time series analysis in LHC superconducting magnets

Wielgosz Maciej, Skoczeń Andrzej

International Journal of Applied Mathematics and Computer Science

2019

Vol. 29, no. 3

503--515

The aim of this paper is to present a model based on the recurrent neural network (RNN) architecture, the long short-term memory (LSTM) in particular, for modeling the work parameters of Large Hadron Collider (LHC) superconducting magnets. High-resolution data available in the post mortem database were used to train a set of models and compare their performance for various hyper-parameters such as input data quantization and the number of cells. A novel approach to signal level quantization allowed reducing the size of the model, simplifying the tuning of the magnet monitoring system and making the process scalable. The paper shows that an RNN such as the LSTM or a gated recurrent unit (GRU) can be used for modeling high-resolution signals with the accuracy of over 0.95 and a small number of parameters, ranging from 800 to 1200. This makes the solution suitable for hardware implementation, which is essential in the case of monitoring the performance critical and high-speed signal of LHC superconducting magnets.