Accepted Papers

ABSTRACT

The paper investigates how the principles of the task approach and the concept of semantic programming can be applied to address various unresolved issues related to trusted artificial intelligence. First of all, this concerns such problems as: the AI centralization problem, the AI Black Box problem, and the AI audit problem, which leads to the problem of trust. This paper presents the Delta framework, which allows the implementation of AI algorithms as smart contracts in a decentralized environment based on multi-blockchain structures. This allows us to achieve transparency, reliability, and decentralization of AI systems, which brings us one step closer to trusted artificial intelligence.

KEYWORDS

Artificial intelligence, trusted AI, strong AI, explainable AI, task approach, semantic programming, multi-blockchains, smart contracts, Bitcoin, ChatGPT, Telegram, cryptocurrency, Delta framework.

Bayesian Intelligent And Soft Measurement

Svetlana Prokopchina1 and Veronika Zaslavskaia2, 1Financial University under the Government of the Russian Federation, Moscow, Russia, 2Zello Russia, "Artificial Intelligence" Committee of "RUSSOFT" Association, St. Petersburg, Russia

ABSTRACT

Modern measurement tasks are solved under conditions of uncertainty Significant information uncertainty is caused by the lack of complete and accurate knowledge about the models of measurement objects, influencing factors, measurement conditions, and the diversity of experimental data. In the article briefly discusses the history of the development of methods for the intellectualization of measurement processes, which are oriented to the situation with uncertainty also and the classification of measurements and measurement systems. The basic requirements for intelligent measurement systems and technologies are formulated. The article considers the conceptual aspects of intelligent measurements as measurements based on the integration of metrologically certified data and knowledge and defines intelligent measurements. The properties of intelligent measurements are determined. The article considers the main properties of soft measurements and their differences from deterministic classical measurements of physical quantities. Cognitive, systemic, and global measurement are marked as new types of measurement. In this paper, the methodology and technologies of Bayesian intelligent measurements based on the regularizing Bayesian approach are considered in detail. In this type of measurement, a new concept of measurement is implemented, in which the measurement problem is posed as the inverse problem of pattern recognition in accordance with the postulates of the Bayesian approach. Within the framework of this concept, new types of models and measurement scales are proposed in the form of models and coupled scales with dynamic constraints that provide for the creation of developing measurement technologies in order to implement the processes of cognition and interpretation of measurement results by means of measurement systems. The new type of scale allows the integration of numerical (for numerical data) and linguistic (for information in the form of knowledge) information in order to improve the quality of measurement solutions. A new set of metrological characteristics of intelligent measurements is proposed, including indicators of accuracy, reliability (error levels of the 1st and 2nd kind), reliability, risk, and entropy characteristics. The paper presents formulas for the implementation of the measurement process with a complete metrological justification of the solutions. An example of solving an applied problem by means of an intelligent measuring complex for monitoring the state of water supply networks based on the methodology and technologies of Bayesian intelligent measurements is considered. In conclusion, the advantages and prospects of using intelligent measurements are formulated, both for solving applied problems, and for the development and integration of artificial intelligence and measurement theory technologies.

KEYWORDS

Measurement Theory, Bayesian approach, Uncertainty

Teaching Reading Skills More Effectively

Julia Koifman, Beit Ekstein Rupin high school, Emek Hefer, Israel

ABSTRACT

Reading is one of the most crucial skills in learning. Children learn to read very early, and before they start school, they are supposed to be able to read. Nevertheless, some of them struggle. For instance, some of them confuse letters or may have difficulty reading comprehension, while others may have difficulty remembering, which might be the consequence of learning difficulties (LD), for instance, dyslexia, one of the most common cognitive disorders. It often affects reading and language skills. Researchers estimate that more than 40 million people in the USA have dyslexia, but only about 2 million of them have been diagnosed with dyslexia. At the same time, about 30% of people diagnosed with dyslexia also suffer from autism spectrum disorders (ASD) and attention deficit hyperactivity disorder (ADHD) to one degree or another.

KEYWORDS

Dyslexia, dysgraphia, dyspraxia, ADHD, ASD, learning difficulties, neurodiversity.

Mimo Mobile-to-mobile 5g Communication Systems Along Elliptical Geometrical Channel Modeling for Analysis of Channel Parameters

Samra Urooj Khan1, Sundas Naqeeb Khan2, Zoya Khan3, 1Department of Electrical Engineering Technology, Punjab University of Technology, Rasool, Mandi Bahauddin, Punjab, Pakistan, 2Department of graphics, Computer vision and digital systems, Silesian University of Technology, Gliwice, Poland

ABSTRACT

The condition of the propagation environment is essential in the design and execution of any transmission medium. As a result, mathematical modeling of transport routes has been a focus of study for centuries. Geometrical channel modeling, as demonstrated by researchers and theorists, is best suited for mobile-to-mobile (M2M) communication settings. Several hollow cylindrical geometrical systematic collections have been thoroughly studied in this research. According to the literature, an elliptical modeling technique could more centralize national the transmission channel. Furthermore, the influence of different channel coefficients across multiple-in-multiple-out (MIMO) resonators has been illustrated utilizing geometrical models. Moreover, the velocity of a mobile station (MS) inside the M2M presenter has still not been assessed among the MIMO resonators. For 5G communications networks, a study of several MS variables would be given.

KEYWORDS

M2M, MIMO, MS, Geometrical modeling, Transmission, 5G, Communication.

Batch-stochastic Sub-gradient Method for Solving Non-smooth Convex Loss Function Problems

KasimuJuma Ahmed, Mathematics Unit, Department of General Studies, Federal Polytechnic Bali, Taraba State, Nigeria

ABSTRACT

Mean Absolute Error (MAE) and Mean Square Error (MSE) are the two loss functions that can fit well into machine learning techniques to make accurate prediction on a continuous data. MAE has non-differentiable property but penalizes outliers unlike MSE which has differentiable property but does not penalize outliers. Batch sub-gradient method is expensive but stable because iteration is over the entire dataset while stochastic sub-gradient method is less expensive but not stable because epoch is over a single data point. A batch-stochastic sub-gradient method is developed for its computational efficiency than batch and stability than stochastic because epoch is defined over collection of data. We tested the computational efficiency of the method using Structured Query Language (SQL). The new method shows greater stability, efficiency, accuracy and convergence than any other existing method.

KEYWORDS

Machine learning, Loss function, sub-gradientMean Absolute Error (MAE) and Prediction.

Microbe2Pixel: Taxonomy informed deep-learning models and explanations

B. Voermans1, 2, M.C. de Goffau2, 3, M. Nieuwdorp1, 4 , and E. Levin2, 1Department of Experimental Vascular Medicine, Amsterdam University Medical Centers, University of Amsterdam , Meibergdreef 9, Amsterdam, the Netherlands, 2HORAIZON Technology BV, Marshallaan 2, Delft, the Netherlands, 3Tytgat Institute for Liver and Intestinal Research, Amsterdam University Medical Centers, Meibergdreef 69-71, Amsterdam,the Netherlands, 4Department of Vascular Medicine, Amsterdam University Medical Centers, University of Amsterdam, Meibergdreef 9, Amsterdam, the Netherlands

ABSTRACT

In recent years, machine learning, especially deep learning, has garnered substantial attention in the biomedical field. For instance, deep learning has become a preferred method for medical image analysis tasks. However, in other areas like fecal metagenomics analysis, the application of deep learning remains underdeveloped. This can be attributed to the tabular nature of metagenomics data, feature sparsity, and the complexity of deep learning techniques, which often lead to perceived inexplicability. In this paper, we introduce Microbe2Pixel, an innovative technique that applies deep neural networks to fecal metagenomics data by transforming tabular data into images. This transformation is achieved by inferring location from the taxonomic information inherently present in the data. A significant advantage of our method is the use of transfer learning, which reduces the number of samples required for training compared to traditional deep learning. Our method aims to develop a local model-agnostic feature importance algorithm that provides interpretable explanations. We evaluate these explanations against other local image explainer methods using quantitative (statistical performance) and qualitative (biological relevance) assessments. Microbe2Pixel outperforms all other tested methods from both perspectives. The feature importance values align better with current microbiology knowledge and are more robust concerning the number of samples used to train the model. This is particularly significant for the application of deep learning in smaller interventional clinical trials (e.g., fecal microbial transplant studies), where large sample sizes are unattainable and model interpretability is crucial.

KEYWORDS

metagenomics, interpretable deep learning, local explanations.