MLIT computing power, as well as machine learning methods in high energy physics and applied tasks, presented at XXVI Khariton Talks On 14 April, representatives of the Meshcheryakov Laboratory of Information Technologies spoke at the XXVI Khariton Scientific Talks “Artificial Intelligence and Big Data in Technical, Industrial, Natural and Social Systems”. The conference is held on 14–18 April 2025 at the Russian Federal Nuclear Center (RFNC-VNIITF) in Sarov. MLIT Scientific Leader Vladimir Korenkov delivered a talk at the plenary section of the XXVI Khariton Talks “Artificial Intelligence and Big Data in Scientific Research”, which was chaired by Scientific Director of the National Center for Physics and Mathematics, RAS Academician Alexander Sergeev. His plenary report was devoted to methods and technologies for intelligent data processing and analysis in megascience projects. Vladimir Korenkov spoke about the structure and capabilities of the JINR Multifunctional Information and Computing Complex based on MLIT, paying special attention to the organization of distributed computing, data processing and storage for the experiments at the Large Hadron Collider (LHC) at CERN and at the NICA accelerator complex at JINR. Presenting the computing concept for NICA, Vladimir Korenkov dwelled upon the distributed heterogeneous computing environment created on top of the DIRAC Interware platform. “This computing infrastructure is developing and expanding, and in the future it can be used not only for NICA, but also for all megascience projects implemented in Russia,” MLIT Scientific Leader highlighted. He reported that in 2023, during the 8th physics run of the ВМ@N experiment, a complete reconstruction of 400 terabytes of raw experimental data was performed using this computing infrastructure over five days. The talk also discussed the establishment of a Consortium for IT support of Russian megascience projects based on the RDIG-M infrastructure. It was noted that at present, JINR, based on MLIT resources, provided about 52% of the international scientific traffic in Russia. Two sessional reports were delivered at the XXVI Khariton Talks. MLIT Deputy Director Nikolay Voytishin gave a talk on the application of machine learning methods (MLMs) to process data from high-energy physics experiments. Event reconstruction using MLMs and neural networks on the basis of measurement data in the track detectors of the BM@N, MPD and SPD experiments at NICA and BES-III in China, as well as particle identification in MPD, was considered. Nikolay Voytishin presented MLIT’s experience in applying neural network approaches when working with large data volumes using the example of the CBM experiment and high-luminosity LHC experiments, which can be applied in SPD in the future. MLIT Senior Researcher Alexander Uzhinskiy spoke about the employment of machine learning and artificial intelligence in scientific and applied tasks solved at MLIT. In addition to the application of MLMs and computer vision in JINR tasks in the field of radiation biology, Alexander Uzhinskiy presented MLIT’s experience in solving environmental monitoring tasks using neural networks based on data collected with the help of moss biomonitors and satellite images. Approaches to training neural networks, the software development process and the status of a project on plant disease detection and prediction, as well as the experience of creating digital twins of greenhouse complexes, were covered. The conference is devoted to a wide range of issues related to developments in the field of artificial intelligence and its application for the predictive modeling of the life cycle of complex technical systems and production processes, the formation of theoretical foundations of artificial intelligence, the creation of neurohybrid computing systems, cybersecurity, the development of methods for multi-agent decision-making and management, as well as the creation of promising intelligent systems and methods employed in the social sphere.
