Informatica

In this paper, we focus on the problem of whether efficient Lithuanian large language models (LLMs) can be achieved from Llama2 LLMs, which lack Lithuanian-specific components. Although the Llama2 architecture was previously successfully utilised to derive various regional LLMs, we propose and describe the first open Llama2 LLMs for the Lithuanian language (7 and 13 billion parameter versions), an accompanying question/answer (Q/A) dataset, and translations of popular language understanding benchmarks (Arc, Belebele, Hellaswag, MMLU, TruthfulQA, and Winogrande), which contribute to the standardisation of Lithuanian LLM evaluation. We empirically evaluate the proposed models by investigating their perplexity and performance in the translated language understanding benchmarks. The perplexity experiments show that it decreases consistently during pretraining, reflecting enhanced next-token prediction capabilities. Benchmarking the proposed LLMs against language understanding tasks reveals that high-quality pretraining datasets may be essential to achieve models that perform efficiently on these benchmarks. Comparison of the proposed LLMs with the latest open multilingual LLM shows that our model with 13 billion parameters is ranked 4th of 8 models in tasks such as Arc, Hellaswag, and Winogrande, but is generally outperformed in other tasks. These benchmarks allow us to hypothesise that from recent LLMs more efficient Lithuanian language models can be derived in the future. The complete realisations of the LLMs and other contributed components are available in the accompanying open repository https://huggingface.co/neurotechnology.

For more than two decades, Lithuanian speech recognition has been researched solely in Lithuania due to the need for deep knowledge of Lithuanian. AI advancements now allow high-quality speech-to-text systems to be built without native knowledge, given sufficient annotated data is available. This study evaluated as many as 18 Lithuanian speech transcribers using a small piece of recording; 7 best ones were selected and evaluated using extensive data. The top system achieved a WER of 5.1% for Lithuanian words, with three others showing 8.7–9.2%. For other word-size tokens, such as numbers, speech disfluencies, abbreviations, foreign words, a classification adapted to the Lithuanian language was proposed. Different processing strategies for tokens of these classes were examined and it was assessed which transcribers tend to follow which strategies.

Electricity demand estimation is vital for the optimal design and operation of microgrids, especially in isolated, unelectrified, or partially electrified areas where demand patterns evolve with electricity adoption. This study proposes a causal model for electricity demand estimation that explicitly considers the electrification process along with key factors such as hour, month, weekday/weekend distinction, temperature, and humidity, effectively capturing both temporal and environmental demand patterns.

Seismic hazard analysis plays a vital role in evaluating the potential earthquake risk in a given region. Northeast India is one of the most seismically active zones due to its tectonic positioning at the collision boundary of the Indian and Eurasian plates. This study aims to implement a comprehensive Seismic Hazard Assessment (SHA) framework using Fuzzy Multi-Criteria Decision Making (MCDM) techniques to improve the accuracy and reliability of Peak Ground Acceleration (PGA) estimates in Northeast India. The methodology integrates Trapezoidal Fuzzy Full Consistency Method (TrF-FUCOM) and Neutrosophic-TOPSIS under Single Valued Neutrosophic Set (SVNS) environment (Neutrosophic-TOPSIS), effectively addressing the limitations of traditional seismic hazard assessment methods, particularly in selecting and weighting Ground Motion Prediction Equations (GMPEs). An extensive earthquake catalogue covering the period from 1762 to 2024 has been analysed, and after declustering, fault zones have been delineated based on earthquake density along active faults. The analysis provides a detailed spatial distribution of Peak Ground Acceleration (PGA) across the region, with the highest PGA value reaching 1.43g using the Deterministic Seismic Hazard Assessment (DSHA) method. The findings of this study offer crucial insights for disaster preparedness, urban planning, and the design of earthquake-resistant infrastructure, helping to mitigate seismic risks and enhance the resilience of communities in Northeast India.

This paper introduces a novel approach that bridges the floating-point (FP) format, widely utilized in diverse fields for data representation, with the μ-law companding quantizer, proposing a method for designing and linearizing the μ-law companding quantizer to yield a piecewise uniform quantizer tailored to the FP format. A key outcome of the paper is a closed-form approximate expression for closely and efficiently evaluating the FP format’s performance for data with the Laplacian distribution. This expression offers generality across various bit rates and data variances, markedly reducing the computational complexity of FP performance evaluation compared to prior methods reliant on summation of a large number of terms. By facilitating the evaluation of FP format performance, this research substantially aids in the selection of the optimal bit rates, crucial for digital representation quality, dynamic range, computational overhead, and energy efficiency. The numerical calculations spanning a wide range of data variances provided for some commonly used FP versions with an 8-bit exponent have demonstrated that the proposed closed-form expression closely approximates FP format performance.

Sustainable practices are essential for long-term societal development, minimizing environmental impacts while promoting the efficient use of resources. Multi-criteria decision-making (MCDM) approaches can play a vital role in assessing and prioritizing sustainability solutions by considering diverse economic, social, and environmental factors. This study proposes a multi-criteria group decision-making approach based on the Objective Pairwise Adjusted Ratio Analysis (OPARA) method in a fuzzy environment and presents its application for the assessment of sustainable agriculture solutions. In the proposed approach, the evaluation criteria weights are determined by combining subjective weights from experts and objective weights obtained from the MEREC (Method Based on the Removal Effects of Criteria) method. The Relative Preference Relation (RPR) approach is employed for ranking fuzzy numbers and final evaluation. Sensitivity analysis and comparison with other methods are conducted to assess the robustness and validity of the proposed approach. The results demonstrate the effectiveness of the proposed approach in evaluating solutions. Based on the final evaluation from the case study, the most important criteria are “Availability and quality of water”, “Focus on immediate economic returns”, and “Financial incentives and access to credit”, while the most suitable solutions for advancing sustainable agriculture are “Financial and credit support”, “Education and enhancement of farmers’ knowledge”, and “Enhancement of research and development”.

Most existing traffic trajectory recommendation methods don’t consider the driver-car-road preferences, resulting in the poor ability to meet the driver-car-road requirements. To address this issue, we propose a traffic Trajectory Recommendation scheme based on Edge-cloud computing Driver-car-road Preferences (named as TREDP). TREDP reduces the computational, storage, and energy burden on the edge through edge-cloud collaborative computing. TREDP enhances the recommended accuracy by considering driver-car-road requirements and the relationship among driver-car-road in different traffic trajectories. Meanwhile, TREDP increases the computational efficiency through edge-cloud computing. Thus, it improves the driver experience of intelligent traffic trajectory recommendation systems.

One of the most recent innovations in the field of fuzzy sets has been continuous intuitionistic fuzzy sets (CINFUSs), where membership and non-membership degrees are defined by nonlinear functions, as a direct extension of intuitionistic fuzzy sets (IFSs). The membership and non-membership degrees of CINFUSs can account for uncertainty at every point since they are represented by continuous structures that change based on how the decision-maker responds to uncertainty. On the other hand, Pythagorean fuzzy sets (PFSs) allow for a more accurate representation of the data and a better way to handle uncertainty in decision issues by reflecting the hesitations of decision-makers over a larger range. Due to these superior advantages of CINFUSs and the fact that PFSs are more comprehensive than IFSs, in this study, continuous Pythagorean fuzzy sets (CPFUSs) have been aimed at introducing to define uncertainty more broadly and accurately by representing PFSs with a continuous structure as in IFSs. In this study, firstly, the basic principles and mathematical operators of CPFUSs have been developed and presented. Then, multi-attribute decision-making (MADM) models have been developed by considering different aggregation operators to indicate the feasibility and effectiveness of the continuous Pythagorean fuzzy (CPFU) extension. The developed CPFU-MADM models have been implemented to the solution of two different decision problems: green supplier selection and waste disposal site selection problems. In addition, sensitivity analyses have been conducted on criterion weights, expert weights and parameters in order to demonstrate the reliability and stability of the developed models. Furthermore, the validity and superiority of the developed models have been indicated by the comparative analysis conducted with IFSs and PFSs-based MADM models in the literature. MADM applications have shown that continuous Pythagorean fuzzy sets can successfully represent the expert decisions with different attitudes within the same model. It has been observed that the rankings of alternatives according to aggregation operators do not change even when there are differences in the score values of the alternatives.

Quantum computing has come to stay in our lives. Companies are investing billions of dollars in it because of the potential benefits that it can achieve, providing promising applications in almost every business sector. Although quantum computing is evolving at an exponential rate, the development of tools, techniques, or frameworks for the evolution of current information systems towards quantum software systems is still proving to be a challenge. This research contributes to the evolution of current information systems towards hybrid information systems (combining the classical and quantum computing paradigm). We propose a software modernization process, by following model-driven engineering principles, adapted to the quantum paradigm, based on modified versions of standards for reverse engineering of classical, quantum software assets, and for the design of the target system. In particular, this paper focuses on the restructuring transformation from KDM to UML models, where KDM models have been generated from Q# code. This proposal has been validated through a case study involving 17 programmes. The results obtained show optimistic values regarding the complexity of the UML models generated, their expressiveness and scalability. The main implication of this research is that UML models can indeed help the software evolution of/toward hybrid information systems.

The accelerated progress of aquaponics offers a promising remedy for food production in arid regions, where success heavily hinges on sustaining optimal water quality parameters of aquaponic system. However, managing water parameters in large-scale aquaponic farms, given their complex and interconnected nature, poses significant challenges. Various control approaches have been introduced over the years, but selecting the most suitable one is vital for ensuring stability, efficiency, and high productivity. In this study, a novel fuzzy-based Multiple Criteria Decision Making (MCDM) methodology is proposed, which combines the Intuitionistic Fuzzy Ordinary Priority Approach (OPA-IF) with the Neutrosophic-TOPSIS strategy. This methodology aims to identify the most appropriate control strategy for large-scale aquaponic systems. The OPA-IF analysis reveals that the ‘Capability to Handle MIMO Systems’ is the most critical criterion, leading to the conclusion, through the Neutrosophic-TOPSIS approach, that ‘Model Predictive Control (MPC)’ is the optimal choice for managing large-scale aquaponic systems. Additionally, a comparative analysis using the BWM-Neutrosophic-TOPSIS strategy further supports the findings of the proposed method. The results are further validated through statistical analysis and sensitivity testing, ensuring their robustness and reliability. Overall, this study not only contributes to the scientific understanding of control strategies in aquaponics but also offers practical insights for farmers and aquaponic practitioners. The ultimate goal is to enhance the sustainability and efficiency of aquaponic systems, promoting their adoption and long-term success in sustainable food production.