arXiv subjects

Alexander Rogachev

Publications and source records attributed to Alexander Rogachev.

4 recordsLinked to original sources

arXiv2024

Attacks on multimodal models

Today, models capable of working with various modalities simultaneously in a chat format are gaining increasing popularity. Despite this, there is an issue of potential attacks on these models, especially considering that many of them include open-source components. It is important to study whether the vulnerabilities of these components are inherited and how dangerous this can be when using such models in the industry. This work is dedicated to researching various types of attacks on such models and evaluating their generalization capabilities. Modern VLM models (LLaVA, BLIP, etc.) often use pre-trained parts from other models, so the main part of this research focuses on them, specifically on the CLIP architecture and its image encoder (CLIP-ViT) and various patch attack variations for it.

cs.CV↗

arXiv2022

GAN with an Auxiliary Regressor for the Fast Simulation of the Electromagnetic Calorimeter Response

High energy physics experiments essentially rely on simulated data for physics analyses. However, running detailed simulation models requires a tremendous amount of computation resources. New approaches to speed up detector simulation are therefore needed. The generation of calorimeter responses is often the most expensive component of the simulation chain for HEP experiments. It was shown that deep learning techniques, especially Generative Adversarial Networks, may be used to reproduce the calorimeter response. However, those applications are challenging, as the generated responses need evaluation not only in terms of image consistency: different physics-based quality metrics should be also taken into consideration. In our work, we develop a multitask GAN-based framework with the goal to speed up the response generation of the Electromagnetic Calorimeter (ECAL) of the LHCb detector at LHC. We introduce the Auxiliary Regressor as a second task to evaluate a proxy metric of the given input that is used by the Discriminator of the GAN. We show that this approach improves the stability of GAN and the model produces samples with better physics distributions.

physics.data-an↗

arXiv2006

The new cosmological model founded on the Scale Covariant Theory of Gravitation and on the Dirac's Large Number Hypothesis. (Part 1)

The new scale-covariant formulation of the Dirac's Large Number Hypothesis (LNH) is proposed. The basic equations of LNH are formulated in the scale-covariant and "G-invariant" (invariant on the transformation law for G) form. On the basis of the Scale Covariant Theory of Gravitation and Dirac's LNH the cosmological model is constructed that gives as result the closed static Einstein's Universe in the gravitational system of units (g.s.u.) and the closed expanding Universe in the atomic system of units (a.s.u.). The simple dynamical model of atomic clock is proposed that leads to the same connection between the a.s.u. and the g.s.u. as the LNH and allows to specify the parameters of the model. The question about the equivalence of inertial and gravitational masses in the arbitrary system of units is ivestigated. It is deduced that the equivalence of inertial and gravitational masses in the arbitrary system of units takes place only for the unique transformation law for G even if the equivalence of two kinds of mass in the g.s.u. is assumed. It is shown that the model predicts the conservation of the black-body spectrum of the microvave background radiation independently of choice of the transformation law for G. It is argued that the some paradoxes of the standard cosmology can be overcomed in the given model without the supposition about the existence in the past of a phase of inflationary expansion.

gr-qc↗

arXiv2002

Quantum Motion as the Motion in the Integrable Weyl Space

A new version of hidden variables theory founded on the generalisation of world's geometry is proposed. The quantum-mechanical motion as the motion in some "inner space", which has a structure of the integrable Weyl space is examined. Equations of motion for a quantum particle as well as equations for the quantum-mechanical field, "guiding" the particle, are deduced. The wave equation for quantum ensembles, the Born's interpretation of wave function and the "guiding equation" as consequences of proposed model are obtained.

quant-ph↗