BULLETIN OF THE L.N. GUMILYOV EURASIAN NATIONAL UNIVERSITY. Mathematics. Computer science. Mechanics series

Full C(N)D–research of the problem of recovery functions from the generalized Sobolev class

Adilzhan Utesov — Sat, 30 Dec 2023 00:00:00 +0000

In this paper a complete C(N)D-research of the problem of recovery functions from the generalized Sobolev class $W^{\omega_{r}}_{2}$ is carried out in the case, where numerical information of volume $N$ about the function $f$ being restored is removed from linear functionals. Namely, firstly, the exact order of error of restoring functions from classes $W^{\omega_{r}}_{2};$ is established in the metric $L^{q},2\leq q\leq \infty;$ secondly, a specific computing unit is proposed, that implements the exact order and its limiting error $\bar{\varepsilon}_{N}$ is found, that preserves the exact order and not improved in order; thirdly, it is proved that any computing unit constructed by the Fourier coefficients of the function being restored does not have a limiting error, better (in order) than $\bar{\varepsilon}_{N}.$

Self-Supervised Training for the Kazakh Speech Recognition System

Zhanibek Kozhirbayev — Sat, 30 Dec 2023 00:00:00 +0000

In recent times, advancements in neural models trained using extensive multilingual textual and spoken data have displayed promising potential for enhancing the situation of languages that lack resources. This study is centered on conducting experiments utilizing cutting-edge speech recognition models, specifically Wav2Vec2.0 and Wav2Vec2-XLSR, applied to the Kazakh language. The primary aim of this research is to assess the efficacy of these models in transcribing spoken Kazakh content. Additionally, the investigation seeks to explore the feasibility of leveraging data from other languages for initial training, and to assess whether refining the model with target language data can enhance its performance. As such, this study offers valuable insights into the viability of employing pre-trained multilingual models in the context of underresourced languages. The fine-tuned wav2vec2.0-XLSR model achieved exceptional results, boasting a character error rate (CER) of 1.9 and a word error rate (WER) of 8.9 when evaluated against the test set of the kazcorpus dataset. The outcomes of this analysis hold potential to advance the creation of robust and efficient Automatic Speech Recognition (ASR) systems tailored for the Kazakh language. These developments stand to benefit a range of applications, including speech-to-text translation, voice-activated assistants, and speech-driven communication tools.

Mathematical aspects of occupational risk and its classification considering statistical indicators

Sh. Abikenova, R. Marcelloni, Sh. Aitimova — Sat, 30 Dec 2023 00:00:00 +0000

Occupational risk emerges as a comparatively novel concept within the regulation of social and labor relations in the Republic of Kazakhstan, serving as an indicator of the risk for loss of work capacity or the death of an employee during the execution of their work-related (official) duties due to an occupational accident. This article elucidates the statistical aspects of оccupational risk content, affirming the mathematical ubiquity and the validity of the scientific framework employed. The methodological foundation for assessing оccupational risk incorporates universally accepted statistical methods for comparison, grouping, and systematisation through the juxtaposition of credible statistical data. The purpose of the study is to employ one of the mathematical approaches to categorise types of economic activity into classes of occupational risk within the mandatory insurance system against accidents for workers fulfilling their occupational duties. The application of machine learning theory with big data across 132 types of economic activities (two-digit, including some five-digit codes) facilitated the execution of a classification procedure, resulting in the segmentation into 22 classes of occupational risk. It revealed the necessity for introducing a five-digit classification and further detailing the types economic activities as the class of оccupational risk increases, such as including the five-digit codes 07101 "Underground mining of iron ores" and 43991 "Mine construction" into class 22 (the highest), delineating from the two-digit codes of the mining and construction industry accordingly. The scientific results were obtained within the framework of program-targeted funding by the Ministry of Labor and Social Protection of the Population of the Republic of Kazakhstan (scientific and technical program No. BR11965728 "Economic problems of safe work and institutional transformations of the insurance mechanism in the Republic of Kazakhstan").