Asia Pacific Academy of Science Pte. Ltd.
Not a member yet
3612 research outputs found
Sort by
Fine-tuned large language models into traditional back-end web architectures
Full text linkIntegrating Large Language Models (LLMs) into traditional back-end systems can significantly reduce development overhead and enhance flexibility. This paper presents a novel approach using a fine-tuned LLama3 model as a modular back-end component capable of processing JSON-formatted inputs and outputs. We developed a specialized dataset through advanced prompt engineering with the Phi-3.5 model and fine-tuned LLama3 using Quantized Low-Rank Adaptation (QLoRA) on a single NVIDIA T4 GPU. An API layer was designed to facilitate seamless communication between clients and the LLM, effectively replacing conventional application logic. Our fine-tuned model achieved an average accuracy of 76.5¥% and a response time of 3.56 s across 100 test cases, demonstrating its effectiveness in handling back-end tasks. This work underscores the potential of LLMs to transform AI-driven back-end architectures, offering scalable and efficient solutions for modern web services
Co-dynamics of measles and hand-foot-mouth disease
Full text linkThis study develops two compartmental models to analyze the co-dynamics between measles and hand, foot, and mouth disease (HFMD): a four-compartment model and a seven-compartment HFMD-Measles co-infection model. For the four-compartment model, we systematically analyzed the co-dynamics of measles and hand, foot, and mouth disease (HFMD), and employed the next-generation matrix method to calculate the basic reproduction number of measles and that of HFMD. Through the analytical study of these two types of basic reproduction numbers, we rigorously determined the existence of the disease equilibrium points, with their quantitative relationship were clearly illustrated through graphical representations. The global asymptotic stability of these equilibria is established by applying LaSalle invariance principle, with stability regions of the four equilibrium points precisely defined. The analysis reveals that within the stability region of the disease-free equilibrium, both diseases will eventually die out, preventing any outbreaks. In the stability region corresponding to the measles equilibrium, HFMD is eliminated while measles remains endemic. Conversely, in the stability region of the HFMD-only equilibrium, measles dies out whereas HFMD persists. Finally, within the stability region of the coexistence equilibrium, both diseases persist and become endemic. Numerical simulations further validate the consistency and reliability of these theoretical results. For the seven-compartment infectious disease model, we calculated the basic reproduction number and verified the threshold theorem. We derived the conditions for both local and global asymptotic stability of the disease-free equilibrium. In particular, the disease-free equilibrium is locally stable when the basic reproduction number is less than one, and we also provided conditions for its global stability. Model validation is performed by fitting empirical data from China on HFMD and measles cases
Technology integration in supply chain management: A systematic literature review on driving marketing success and efficiency
Full text linkThis study examines how supply chain management (SCM) uses information technology (IT) and how that integration affects marketing effectiveness. IT plays a critical role in improving supply chain management (SCM) procedures as organisations aim for operational excellence in a world going digital. The current body of knowledge regarding the connection between IT integration in supply chain management (SCM) and marketing outcomes is compiled in this study through the use of a systematic literature review (SLR). Important conclusions show that using IT solutions, including cloud computing and advanced analytics, greatly enhances the communication, accessibility, and reactivity of data in marketing initiatives. In order to maximise the allocation of marketing resources and drive client interaction, the conversation highlights the revolutionary potential of IT. The study ends with proposals for additional research to examine the changing landscape of IT in SCM as well as tips for practitioners
The challenges for superconducting wires and tapes today and tomorrow
Full text linkThe main advantages of superconducting wires and tapes is lossless flow of transport current, lossless storage of energy, and generation of high magnetic fields. These advantages can help solve the most critical problems that world faces. Currently, humanity's primary concern is energy security. Humanity for stable development needs cheap, stable, and non-polluting energy sources. A fusion reactor can meet such expectations because it will produce a large amount of stable energy without pollution of the natural environment. This reactor needs cheap superconducting coils and superconducting wires with high critical parameters. Various superconducting wires and tapes e.g. NbTi, Nb3Sn, MgB2, HTS, and iron-based superconductors are being considered and tested. It seems that in the future, MgB2 wires with isotopic boron (11B) may be necessary for fusion reactors because irradiation does not degrade the transport critical current density and has a short decay time of about 1 year [1]. Another crucial application of superconductivity could be the combination of renewable energy sources with superconducting magnetic energy storage. This opens up prospects for cheap and clean energy on a small scale (small family businesses)
Quality and safety indicators of multigrain bread with freshly prepared rye malt using phytoantioxydants
Full text linkThe quality and safety indicators of a new type of multigrain bread using freshly obtained rye malt were studied. Hydrogen peroxide as a safe prooxidation activator and aqueous phytoextracts as natural antioxidants were used for the bioactivation of grain. The choice of phytoextract was made by comparing the antioxidant properties of thyme and oregano obtained with warm (60 ℃) or hot (80 ℃) extraction. The rate of accumulation of organic acids and reducing substances was maximum during malting under the influence of thyme extract (80 ℃). Oregano extracts were used to make rye-textured flour. The paper describes a new method for producing emulsified malt, known as “rye cream”. Unlike other methods, it is proposed to use not dried malt, but freshly obtained rye malt extract. This made it possible to form the adaptogenic properties of the food system, which was evident from the results of the antioxidant activity assessment. This assessment was carried out at different stages of the bread production process. It was shown that at the stage of rye cream production, the oxidation-reduction balance of the food system was shifted towards prooxidants, so the use of phytoextracts with antioxidant properties was necessary. At the dough production stage, the balance changed in the opposite direction, and antioxidants began to dominate the food system as functional nutritional components. Multigrain bread had antioxidant activity twice as high as the standard (Borodinsky bread) and had better characteristics compared to two analogues. Under the influence of multigrain bread, the production of free radicals was inhibited more gently than under the influence of rye bread “BIO”. Another analogue, Fitness bread, had the properties of a strong prooxidant, not an antioxidant. According to the results of an objective assessment in independent laboratories, multigrain bread with rye cream met regulatory requirements for quality and safety, which was confirmed by the results of physicochemical and microbiological analysis
Virtual design and manufacturing technology used for rapid parametric design of mechanical model
No full textIn the traditional design methods, there are challenges in quickly adjusting a three-dimensional model according to the variations in mechanical parts. In view of the above problems, a rapid design method for gear reducer based on parametric design is proposed. Instead of directly constructing geometric models based on three-dimensional software, this paper comprehensively considers the parameters and constraints needed for three-dimensional models. By analyzing the parameters and constraints of parts, the design interface of mechanical parts is constructed. Then, the design interface is connected with the parameters and constraints of three-dimensional model through the programming language. Finally, a parametric design system for mechanical parts is developed for rapid design and modification of three-dimensional model. Experimental results demonstrate that the rapid parametric design reduces the modeling time by 28% (for single-parameter parts) and 57% (for three-parameter parts) compared to traditional methods, validating the efficiency of the proposed methodology
Research on deep learning analysis and optimization of humanoid robot based on Yushu Technology
No full textHumanoid robots, as core carriers of embodied intelligence, rely on their deep learning and behavior prediction capabilities to break through the bottleneck in general-task execution. Taking Unitree as a case study, this research conducts an in-depth analysis of the current technical status, challenges, and optimization paths of humanoid robots in this field. A dynamic environment perception-decision-execution closed-loop system is constructed, encompassing a multimodal perception layer, a hybrid decision-making layer, and a realtime execution layer. It is proposed that hardware iteration must be deeply coordinated with AI algorithms. In terms of model optimization, a multi-task lightweight model architecture is established, which innovatively combines dynamic environment adaptation algorithms with transfer learning mechanisms. Meanwhile, efforts are being made to develop a native multimodal industry-specific large-scale model for robots, exploring the engineering
implementation plan for humanoid robot behavior prediction. Experimental verification not only tests the performance of Unitree’s humanoid robots but also identifies technical bottlenecks such as insufficient chip computing power, lack of industry-specific large-scale models, and dependence on remote control, along with targeted optimization suggestions. Finally, this study looks ahead to the development trends of humanoid robot technology, including breakthroughs in general AI models, the implementation of neuromorphic computing, and aspects of social impact and ethical reconstruction, aiming to promote the development of the humanoid robot industry and expand its applications in diverse scenarios such as industry and households
Thermodynamic nexus between UNCBD and UNFCCC: Bounded openness over natural information and the Yasuní-ITT initiative
No full textConsilience of biodiversity and climate change is achievable through reduction to non-equilibrium thermodynamics. Subsequent construction supports a policy to decrease carbon emissions and species loss in capital-poor countries that are mega-diverse and carbon-rich. Incentives for conservation underpinned The Yasuní-ITT Initiative, whereby Ecuador sought payment not to drill for oil. The same logic holds for not opening highways in the Amazon. Consilience marries Living within Limits by Garrett Hardin with Half-Earth by E.O. Wilson. Artificial Intelligence facilitates persuasion of the Conference of the Parties to the United Nations Framework Convention on Climate Change and the United Nations Convention on Biological Diversity
Fostering sustainable behavior through green leadership: The mediating role of environmental consciousness and moderating effect of goal clarity
No full textThis study investigates the mechanisms through which green transformational leadership (GTL) and green authentic leadership (GAL) influence employees’ green behavior for sustainable development (EGB). Drawing from social cognitive and goal-setting theories, we examined the mediating role of environmental consciousness (EC) and the moderating role of goal clarity (GC) using survey data collected from 532 employees across diverse organizational contexts in China. Partial least squares structural equation modeling (PLS-SEM) via SmartPLS was employed to analyze the data. Results indicate that GTL and GAL positively impact EC and EGB directly. EC significantly mediates the relationship between both leadership styles and EGB, highlighting its critical psychological function in translating leadership practices into concrete environmental actions. Additionally, GC significantly strengthens the positive relationship between EC and EGB, demonstrating the importance of clearly articulated sustainability objectives in fostering sustainable workplace behaviors. These findings provide essential theoretical insights and practical implications for enhancing sustainability performance through targeted leadership practices, environmental awareness initiatives, and effective sustainability goal-setting strategies
Application of ZnO nanoparticles modified with bacterial proteins for the removal of trypan blue effluent
Full text linkIn this study, a bacterial protein-coated nanoparticle system is modified as a new biosorbent. Escherichia coli (E. coli), and Staphylococcus aureus (S. aureus) bacterial proteins are collected, and successfully coated onto zinc oxide nanoparticles (ZnONPs). The new biosorbents are combined between the attractive surface properties of the nanoparticles and the adsorbed protein corona. ZnONPs, ZnONPs/E. coli, and ZnONPs/S. aureus were characterized using X-ray diffraction (XRD), Brunauer-Emmett-Teller (BET), Scanning Electron Microscopy (SEM), and Infrared spectroscopy (FT-IR) studies analysis. The Bradford method was used to ensure the presence of protein corona on the modified surface. The addition of bacterial proteins to the surface of the ZnONPs alters its activity through modifications of its size, shape, surface charge, and other characteristics. The improvement of the functional groups and surface charge of the modified biosorbents makes it more efficient for the removal of dyes. ZnONPs, ZnONPs/E. coli, and ZnONPs/S. aureus were used for the removal of trypan blue (TB) dye from contaminated wastewater. The TB dye was completely removed (98%–100%) using ZnONPs/E. coli, and ZnONPs/S. aureus within 25–30 min, whether in the dark or light conditions, over a wide pH range (5–9). The negative values of ∆G showed the spontaneous nature of the removal process. The ∆H values confirmed an endothermic removal in the dark and an exothermic removal in the light. ZnONPs/E. coli, and ZnONPs/S. aureus were applied for the removal of TB dye from real wastewater samples, and their efficiencies were proven. The average removal rate of TB dye using ZnONPs E. coli, and ZnONPs/S. aureus was 92 % which is more efficient than that of ZnONPs (87 %), and the average value of RSD% was 1.7 % (n = 5)