Találati lista | Tudóstér

001-es BibID:	BIBFORM125237
Első szerző:	Godó Zoltán (biológus, informatikus)
Cím:	Neural Network Visualization for The Study of Parallel Algorithm Building Skills / Zoltán Attila Godó
Dátum:	2024
Megjegyzések:	In our previous studies, we have shown that algorithm building skills provide the most important foundation for later programming ability. To measure this, we developed a one-dimensional light programming system on a web interface that allows for age-independent evaluation. However, the system only tested single-threaded algorithm creation. In modern programming paradigms, multithreaded program units are coded. This requires much more complex thinking. The underlying parallel algorithm building skill can only be tested in 4D visualization. The visualizer we developed for our multiprocessor neural network, which maps each processor to a node, is precisely suited to visualize a spatial, timevarying light program. The movement of the spatial bodies, meeting and stopping each other's cascadelike starts, emulates the running of parallel programs. The subject is able to simultaneously move the spatial bodies and their interdependent behaviour. The subject tries to replicate as closely as possible the light program of the test sample. The test program measures the correctness of the solution to the task, the mathematical distance to the correct solution, the number of incorrect clicks and several time parameters such as the time between clicks. A total of 36 parameters are recorded during a single measurement, from which further calculated data series can be generated. The evaluation of the test results was compared with measures of programming ability, such as students' grades in programming or the results of solving specific programming problems. The evaluations show a strong correlation between test scores and programming ability. The visualization of the neural network is thus suitable for a valid measure of parallel algorithm building skills. The system can help in the early identification of programming talent and in advising on a career in programming. Which may even help to alleviate the shortage of programmers in the computer science profession.
ISBN:	978-625-98911-4-9
Tárgyszavak:	Természettudományok Matematika- és számítástudományok előadáskivonat könyvrészlet light programming algorithm building skill neural network visualization
Megjelenés:	ICONST EST 2024 International Conferences on Science and Technology Engineering Science and Technology: Engineering Science and Technology Abstracts and Proceedings Book / Dr. Mustafa Karaboyac; Dr. Abdullah Beram; Dr. Hamza Kandemir; Dr. Serkan Özdemir. - p. 21-21. -
Internet cím:	Szerző által megadott URL Intézményi repozitóriumban (DEA) tárolt változat
Borító:
Saját polcon: