KPI Science News

RATIONAL SEARCH ALGORITHM FOR SEARCHING FOR A LEGAL POSITION IN A SET OF JUDICIAL DECISIONS OF THE JUDICIARY OF UKRAINE USING LARGE LANGUAGE MODELS

Pavlo Maslianko — 2025-07-21

Background. The need to develop a reasoning search algorithm that is capable of automating the search and generation of context-based legal positions and an algorithm for generating legal positions of judicial decisions from existing judicial decisions in the judicial system of Ukraine, as a preparatory stage for a reasoning search algorithm.

Objective. Development of the concept of a reasoning search algorithm that is capable of automating the search and generation of context-based legal positions. Development and implementation of an algorithm for generating legal positions of judicial decisions available in the judicial system of Ukraine.

Methods. Development of the concept of a model of a reasoning search algorithm with three main components: search, formation of alternatives and generation of an answer. Development and implementation of an algorithm for generating legal positions of court decisions that searches for used legal norms, decomposes court decisions into a sequence of actions and generates legal positions.

Results. Mathematical model of the algorithm for generating legal positions of court decisions, which ensures automatic formation of a database of structured legal positions. Validation of the algorithm for generating legal positions of court decisions by comparing the results of the algorithm analysis and existing legal positions in the Database of Legal Positions of the Supreme Court.

Conclusions. The need for automating the search and generation of context-based legal positions is substantiated. The concept of a reasoning search algorithm is proposed, which is capable of automating the search and generation of context-based legal positions. An algorithm for generating legal positions using large language models and Data Science methods has been implemented, the validation of which has shown that the results correspond to the existing positions of the Supreme Court

TALL ARRAY METHOD EFFICIENCY IN DATASET DIMENSIONALITY REDUCTION BY PRINCIPAL COMPONENT ANALYSIS

Vadim Romanuke — 2025-07-21

Background. Exploratory data analysis has been extensively growing since early 2000s. As of 2025, most real-practice datasets are classified as Big Data. The Big Data analytics workflow includes the data preprocessing step, which is the starting point of Big Data computational handling. At this step, the data are tried to get simplified as much as possible. The main paradigm is dimensionality reduction allowing to simplify and visualize high-dimensional datasets. Principal component analysis (PCA) is a linear dimensionality reduction technique. The PCA can be sped up by applying Tall Arrays, if the data are stored on disk. The Tall Array PCA (TAPCA) computes principal components incrementally using a divide-and-conquer strategy.

Objective. The objective is to determine when the TAPCA is factually efficient for dimensionality reduction. The two numeric types to be studied are double and single precision.

Methods. To achieve the said objective, random large datasets are generated as matrices of a specified numeric type. Then computational time of the ordinary MATLAB PCA applied to generated matrices is measured. Next, computational time of converting in-memory arrays (generated matrices) into tall arrays is measured. Computational time of the TAPCA applied to those generated matrices, to which the PCA is applied before, is measured as well.

Results. A comparative analysis of the averaged computational times reveals that computational time complexity of both the PCA and TAPCA is rather polynomial than strictly quadratic or cubic. There is a nearly-hyperbolic margin, which alternatively could be called the TAPCA efficiency threshold, in a plane of the number of dataset observations and the number of dataset features, by which the TAPCA and the ordinary PCA take approximately the same time to compute principal components.

Conclusions. In computing principal components for dimensionality reduction of large datasets stored on disk, the Tall Array method becomes efficient by two parallel processor workers if a dataset has at least 5 to 6 million entries. The Tall Array method is more efficient on datasets with double precision whose efficiency threshold is nearly 6 million entries, whereas the efficiency threshold for datasets with single precision is between 5 to 15.2 million entries

A SKIN CANCER CLASSIFICATION MODEL INCORPORATING A DISCRIMINATOR INTO THE LADDER NEURAL NETWORK ARCHITECTURE

Oleksii Zarytskkyi — 2025-07-21

Background. Semi-supervised learning (SSL) is one of the most promising areas of deep learning, especially for tasks where label acquisition is a complex and expensive process, such as skin cancer classification. Existing approaches, such as Adversarial Autoencoders (AAE) and Ladder Networks (LN), effectively use unlabeled data, but have limitations in reconstruction and regularization accuracy.

Objective. Development and study of a model for classifying skin cancers based on the integration of a discriminator into the architecture of ladder neural networks.

Methods. The developed model combines the regularizing properties of ladder neural networks with the use of a discriminator-based loss function that evaluates the quality of image reconstruction based on their structure, shape, and key visual features. The experiments were conducted on the HAM10000 dataset with different ratios of labeled and unlabeled data (30%, 10%, 5%).

Results. The experiments showed that the proposed model improved the -score for the malignancy class by 4% compared to the baseline ladder network with 5% labeled data. With 30% labeled samples, the score reached 74.8%, which is only 1% less than the fully supervised model. The relative indicator at 5% labeled data was 0.939, exceeding the similar coefficient for STFL (0.877), which confirms the effectiveness of using unlabeled data in the proposed model.

Conclusions. The proposed model improves on existing semi-supervised learning methods (Ladder Networks, Adversarial Autoencoders) by providing high efficiency of using unlabeled data to regularize the encoder latent space in the task of skin cancer classification. Prospects for further research include using a discriminator to compare latent spaces and improving the Reconstruction Cost function to expand its application to other medical image analysis tasks

METHOD OF INCREASING THE WEAR RESISTANCE OF A NICKEL-GRAPHITE-BASED COMPOSITE PLASMA COATING BY ADDING COPPER ALLOYS

Oleksii Kushchev — 2025-07-21

Background. The article considers the possibility of expanding the areas of application of plasma coatings based on the nickel-graphite system by adding lead-free copper alloys to the composition, which have an affinity for the nickel shell of the nickel-graphite powder and, in addition to preserving the tribotechnical properties of nickel-graphite, can provide the necessary level of cohesion of the coating.

Objective. The aim of this work is to study the effect of the content of the copper alloy CuAl10Ni5Fe4 when added to nickel-graphite (NPG-75) on the tribotechnical characteristics of the composite plasma coating.

Methods. As part of the work to determine the influence of the copper alloy content in nickel-graphite on the tribotechnical properties of the plasma composite coating, friction and wear tests were conducted on samples with 35, 50, 75 and 90 wt.% copper alloy CuAl10Ni5Fe4 in nickel-clad graphite NPG-75 (75 wt.% nickel). The double copper-zinc alloy CuZn10 was used as a standard.

Results. It has been experimentally established that plasma coatings based on NPG-75 with the addition of 50 wt.% CuAl10Ni5Fe4 have wear resistance increased by 10 times compared to pure NPG-75 with a friction coefficient significantly lower (0.36) than the CuZn10 standard (0.52). The coating on the counterbody of DIN C45 (hardness 45-48 HRC) is the smallest among the friction pair that were investigated in this work. Reducing the copper alloy content to 35 wt.% is impractical, since this results in active wear of the coating material and its spreading onto the counter body. Further increase in the content of CuAl10Ni5Fe4 powder in the composition leads to the occurrence of setting in the contact zone, as evidenced by a sharp (10-fold) increase in the wear of the coating samples, an increase in the temperature in the contact zone, and a ~ 2-fold increase in the friction coefficient.

Conclusions. The optimal content of CuAl10Ni5Fe4 in a plasma coating based on NPG-75 is about 50 wt.%. This material allows you to reduce the friction coefficient from 0.52 (CuZn10) to 0.36 when paired with steel DIN C45. The mass wear of the steel counterbody is also significantly reduced, probably due to a change in the wear mechanism from adhesive (seizing) to oxidative wear. This allows us to significantly expand the scope of application of the Ni-C + CuAl10Ni5Fe4 material in aviation and space technology

ANALYSIS OF CIRCUIT DESIGN METHODS FOR LINEARIZING THE TEMPERATURE CHARACTERISTICS OF NTC THERMISTORS AND THE MEASUREMENT CHANNEL CHARACTERISTICS

Sergii Matvienko — 2025-07-21

Background. Temperature measurement devices play a crucial role in the automation of various systems and in monitoring diverse physico-chemical parameters. thermoresistors, thermistors, or thermocouples are commonly used as temperature sensors. However, the nonlinear temperature-resistance characteristic of thermistors presents challenges for accurate temperature-to-electric signal conversion. Therefore, the issue of linearizing the output characteristic of the temperature measurement channel remains highly relevant. The task of linearizing the temperature dependence of the measurement channel can be addressed using hardware (circuit-based), software, or combined hardware-software approaches.

Objective. The objective of this research is to analyze the effectiveness of circuit-based (schematic) methods for linearizing the temperature characteristics of NTC-type thermistors and to evaluate their impact on the overall performance of the temperature measurement channel.

Methods. A structural model of a temperature measurement device utilizing an NTC thermistor as a sensor was developed in the MATLAB Simulink environment. Five different circuit configurations for connecting the thermistor to the measurement channel were analyzed to assess the effectiveness of linearization within the temperature ranges of ±15 °C and ±25 °C. Based on the proposed connection schemes, simulation of the temperature characteristics of NTC thermistors was carried out.

Results. The results of the study for five circuit-based methods of linearizing the temperature characteristics of NTC thermistors provide an opportunity to evaluate the effectiveness of applying different circuit-based methods for linearizing the temperature characteristics of NTC thermistors and the measurement channel characteristics in two temperature ranges: ±15°C and ±25°C. Based on the presented simulations, it can be stated that by applying these circuit-based methods for linearizing the temperature characteristics of NTC thermistors and the measurement channel characteristics, the measurement error can be reduced by 40% in a limited temperature range compared to using a balanced measurement bridge, and corrections for compensating the self-heating effect of the thermistor can be determined.

Conclusions. By modeling the circuit-based methods for linearizing the temperature characteristics of NTC thermistors, it becomes possible to evaluate the effectiveness of linearizing the temperature dependence of the measurement channel while investigating various options for connecting the thermistor to the measurement channel. Additionally, the parameters of the measurement channel elements of the temperature measuring device can be determined