Stanislav Protasov

Innopolis University · Machine learning and knowledge representation laboratory

Artificial neural networks are currently applied in a wide variety of fields, and they are near to achieving performance similar to humans in many tasks. Nevertheless, they are vulnerable to adversarial attacks in the form of a small intentionally designed perturbation, which could lead to misclassifications, making these models unusable, especiall...

Medical imaging is a useful tool that simplifies the diagnosing and treatment. Having high-quality images doctor can make a statement about disease and correct the further recovery steps. The image quality depends on such factors as equipment condition, scanning correctness and patient’s body structure. The solid inclusions inside the human body ca...

Computational material science aims to simulate substances to understand their physical properties. Bioelectronics is an interdisciplinary field that studies biological material from the conductivity point of view. In case of proteins, the folding is an important feature that directly influences physical and chemical properties. The folding modelli...

Quantum programs allow to process multiple bits of information at the same time, which is useful in multidimensional data handling. Images are an example of such multidimensional data. Our work reviews 14 quantum image encoding works and compares implementations of 8 of them by 3 metrics: number of utilized qubits, quantum circuit depth, and quantu...

m.lisnichenko@innopolis.university, s.protasov@innopolis.ru Абстракт Вычислительное материаловедение направлено на моделирование веществ для изучения их физических свойств. Биоэлектроника-это междисциплинарная область, изучающая биологические материалы с точки зрения проводимости. Пространственная структура (или свертка) белка напрямую влияет на ег...

Researchers have put a lot of effort into reducing the gap between current quantum processing units (QPU) capabilities and their potential supremacy.One approach is to keep supplementary computations in the CPU, and use QPU only for the core of the problem. In this work, we address the complexity of quantum algorithm of arbitrary quantum state init...

Poster for the Quantum Techniques in Machine Learning (QTML) conference under research topic "Case study on quantum convolutional neural network scalability"

Computational material science aims to simulate substances to understand their physical properties. Bioelectronics is an interdisciplinary field that studies biological material from the conductivity point of view. In case of proteins, the folding is an important feature that directly influences physical and chemical properties. The folding modelli...

Bioelectronics is a perspective future of electronics. The modelling of the protein is an important part that allows to search the appropriate folding structure with applicable conductivity properties. The classical computers struggle from modelling large structures because of number degrees of freedom. The mathematical modelling inside the quantum...

Quantum processing units (QPU) in theory propose a computational supremacy in a significant number of tasks. Quantum programs are well suited for vector and matrix data processing. The greatest concern is whether physical implementations will step over the noise and decoherence limitations: today in the noisy intermediate-scale quantum (NISQ) era t...

Medical imaging is a useful tool that simplifies the diagnosing and treatment. Having high-quality images doctor can make a statement about disease and correct the further recovery steps. The image quality depends on such factors as equipment condition, scanning correctness and patient's body structure. The solid inclusions inside the human body ca...

Contemporary quantum algorithms, being efficient theoretically, fail to run on real QPUs. One reason for these failures is the algorithm's probabilistic nature, which is amplified by imperfections in hardware implementation. Grover search is such a probabilistic method, which enables other methods in quantum optimization and machine learning. In th...

Large multidimensional data sets are hard to visualize. Most existing methods dedicate visual space to multiple items or multiple features. In this work, we explore dimensionality reduction methods to capture both properties. We show that self-organizing maps (SOM) are the good choice for screen and paper visualization. We involve colors to make mu...

Researchers have put a lot of effort into reducing the gap between current QPU capabilities and potential supremacy. One practical approach is to unload, at least partially, computations from quantum computers to the CPU. Arbitrary quantum state initialization is an example of such computations. The known algorithms are exponentially complex if cou...

One of the crucial tasks in computer science is the processing time reduction of various data types, i.e., images, which is important for different fields-from medicine and logistics to virtual shopping. Compared to classical computers, quantum computers are capable of parallel data processing, which reduces the data processing time. This quality o...

In this paper we reproduce theoretical Buffon needle experiment. We run 7 separate tests with different initial parameters. We numerically assess solution accuracy and method convergence and provide suggestions for future experiments in this area. The obtained approximation is π = 3.07505 ± 1.31692 and it does not contradict original hypothesis.

Quantum processing units (QPU) in theory propose a computational supremacy in a significant number of tasks. Quantum programs are well suited for vector and matrix data processing. The greatest concern is whether physical implementations will step over the noise and decoherence limitations: today in the NISQ (noisy intermediate-scale quantum) era t...

Quantum programs allow to process multiple bits of information simultaneously, which is useful in multidimensional data handling. Images are an example of such multidimensional data. Our work reviews 15 quantum image encoding works and compares 8 of them by 3 metrics: number of utilized qubits, quantum circuit depth, and quantum volume. Our work in...

Contemporary quantum algorithms, being efficient theoretically, fail to execute on real QPUs. One reason for these failures is the algorithm's probabilistic nature, which is amplified by hardware implementation imperfections. Grover search is such a probabilistic method, which enables other methods in quantum optimization and machine learning. In t...

K-nearest neighbours (kNN) is a very popular instance-based classifier due to its simplicity and good empirical performance. However, large-scale datasets are a big problem for building fast and compact neighbourhood-based classifiers. This work presents the design and implementation of a classification algorithm with index data structures, which w...

We present an approach to fully automatic metal artifact detection in the computed tomography (CT) studies. Artifacts are parts of the image that corrupt information. Our technique localizes artifacts as the first step of the artifact removal algorithm. Artifacts appear in CT images as 3D connected components. We segment the artifact's area as a co...

K-nearest neighbours (kNN) is a very popular instance-based classifier due to its simplicity and excellent empirical performance. However, large-scale datasets are a big problem for building fast and compact neighbourhood-based classifiers. This work targets the design and implementation a classification algorithm to be used with efficient data str...

With exponential data growth search engines require more memory for storage and time for search. The data is indexed to increase search speed, which requires additional memory. In this study we develop a fully functional search engine for Wikipedia articles and compare different indexing techniques. Using vector quantization for compression we fit...

Big problem visualization at a single plot can be considered as a handy tool for making managerial decisions. Thesaurus can be used to represent important properties of a large text corpus. In this work I propose fully automatic method of visualizing text dataset thesaurus on a plain graph which can be used for domain exploration. Resulting visuali...

Метод k ближайших соседей (kNN) обязывает хранить всю обучающую выборку, что ограничивает его применимость. В работе рассматривается подход, позволяющий уменьшить размер хранимых данных, а также увеличить устойчивость решения на границе классов за счёт процедуры голосования, учитывающей пересечение границы. Метод превосходит оригинальный алгоритм п...

В работе представлен однопроходный детектор связных объектов, использующий модифицированую структуру union-find. Предложенные модификации позволяют за константное время осуществлять слияние компонент и таким образом достигать линейного времени работы алгоритма в зависимости от площади изображения.

The semantic video indexing problem is still underexplored. Solutions to the problem will significantly enrich the experience of video search, monitoring, and surveillance. This paper concerns scene detection and annotation, and specifically, the task of video structure mining for video indexing using deep features. The paper proposes and implement...

Hyperspectral image classification with a limited number of training samples without loss of accuracy is desirable, as collecting such data is often expensive and time-consuming. However, classifiers trained with limited samples usually end up with a large generalization error. To overcome the said problem, we propose a fuzziness-based active learn...

The Indian Pines (corrected) dataset, consisting of 145*145 samples and 220 spectral bands with a spatial resolution of 20 m and a spectral range of 0.4–2.5 μm. Twenty noisy bands were removed prior to the analysis, whereas the remaining 200 bands were used in our experimental setup. The removed bands are 104–108, 150–163, and 220. The original In...

This file contains MATLAB code for reshaping and rewriting of the original Indian Pines dataset and ground truths into text files as per journal requirements. (M)

The original Indian Pines ground truths consist of 16 classes. The ground truth classes and the number of samples per class (class name-number of samples) are as follows: ““Alfalfa-46”, “Corn Notill-1428”, “Corn-Mintel-830”, “Corn-237”, “Grass Pasture-483”, “Grass Trees-730”, “Grass Pasture Mowed-28”, “Hay Windrowed-478”, “Oats-20”, “Soybean Notill...

Automatic analysis of the video is one of most complex problems in the fields of computer vision and machine learning. A significant part of this research deals with (human) activity recognition (HAR) since humans, and the activities that they perform, generate most of the video semantics. Video-based HAR has applications in various domains, but on...

Hyperspectral image analysis often requires selecting the most informative bands instead of processing the whole data without losing the key information. Existing band reduction (BR) methods have the capability to reveal the nonlinear properties exhibited in the data but at the expense of loosing its original representation. To cope with the said i...

Автоматизация контроля успеваемости студентов является важной задачей для повышения производительности преподавателей и управляемости учебного процесса. При этом необходимо всесторонне анализировать результаты автоматизации, выявлять как позитивные, так и негативные итоги. В нашей работе мы рассматриваем результаты внедрения системы управления обуч...

Smartphones have ubiquitously integrated into our home and work environments. It is now a common practice for people to store their sensitive and confidential information on their phones. This has made it extremely important to authenticate legitimate users of a phone and block imposters. In this paper, we demonstrate that the motion dynamics of sm...

1]. Большинство существующих алгоритмов стерео-сопоставления может быть разбито на две категории решений: локальную и глобальную. Вообще говоря, локальные алгоритмы являются с точки зрения вычислений менее затратными, а глобальные алгоритмы формируют более точные ДД. Задачу построения сеточной модели по множеству точек в трёхмерном пространстве мож...