1764. Ляхов П.А.: Параллельная математика высокопроизводительных вычислений и нейронные сети

Issue 1764. June 24, 2026. "Parallel Mathematics of High-Performance Computing and Neural Networks." About the use of an ancient Chinese remainder theorem for high-performance parallel computing, the scientific work of Professor Nikolai Ivanovich Chervyakov, how to teach a computer to understand data sets with sufficient accuracy, how artificial neurons work, weighting coefficients and training neural networks, the theory of explainable AI, the research of Stavropol mathematicians in the field of machine learning, what is "catastrophic forgetting," how high-performance computing helps recognize cancer cells and aids in precision agriculture, whether it is possible to deploy large language models without giant computers, the dragonfly brain and edge computing, computational quantization and distillation, model compression and training a neural network with an already trained neural network, and the training of young mathematicians at the North Caucasus Federal University—watch and listen to the first conversation in the "Young Science of the Caucasus" series. Pavel Alekseevich Lyakhov, PhD in Physics and Mathematics and Head of the Department of Mathematical Modeling in the Faculty of Mathematics and Computer Science at the North Caucasus Federal University named after N.I. Chervyakov (Stavropol), is a guest at the Russian Radio University. The episode premiered on June 24, 2026.