Опубликована: Март 25, 2024
Здобувач факультету природничої
Academic Journal of Science and Technology, Год журнала: 2024, Номер 10(1), С. 50 - 55
Опубликована: Март 26, 2024
Human behavior recognition refers to the classification task of identifying specific actions human characters based on characteristics body and completed through a algorithm. It has wide range applications in intelligent surveillance, video retrieval so on. The main challenge this direction is accurately extract semantic information each describe its dynamic changes space time. Therefore, article introduces latest research progress field recognition. Through deep learning techniques, particularly convolutional neural networks recurrent networks, movements data can be effectively identified. However, models lack interpretability, which practical applications. researchers also introduce application traditional methods learning-based recognition, explore advantages processing multi-time scale introducing attention mechanisms. Finally, paper summarizes potential technology combined with multimodal behavioral analysis, provides prospects for smart fitness, health care other fields.
Язык: Английский
Процитировано
10
International Journal of Computer Science and Information Technology, Год журнала: 2024, Номер 2(1), С. 36 - 44
Опубликована: Март 6, 2024
With the development of medical technology, imaging technology plays an increasingly important role in diagnosing and monitoring diseases. At same time, machine learning has shown strong capabilities data processing analysis, providing new possibilities for detection auxiliary diagnosis images. This paper will deeply discuss application images, especially deep how to achieve effective image through steps such as preprocessing, feature extraction, model training optimization.
Язык: Английский
Процитировано
9
International Journal of Computer Science and Information Technology, Год журнала: 2024, Номер 2(1), С. 348 - 358
Опубликована: Март 24, 2024
A data warehouse is a subject-oriented, integrated, relatively stable collection of that reflects historical changes and used to support management decisions. Common tools for building are IBM Cognos SAP BO. However, both the above use centralized single-node mode build warehouses. This type has poor scalability, due rapid increase in scale Internet, traditional warehouses can no longer meet actual needs use. paper mainly introduces integration cloud machine learning as well importance application parallel methods. First, describes how combination warehousing promote business innovation output. It then discusses challenges managing models production environments, role addressing these challenges. Subsequently, computing Snowflake, implementation steps processes approach also introduced detail. Finally, results method analyzed, it considered good prospect development potential warehouse.
Язык: Английский
Процитировано
8
Journal of Theory and Practice of Engineering Science, Год журнала: 2024, Номер 4(03), С. 176 - 182
Опубликована: Март 25, 2024
As large language models gain traction in the financial sector, they are revolutionizing workflows of professionals. From data analysis and market forecasting to risk assessment customer management, these demonstrate significant potential value. By automating processing tasks, enhance productivity empower professionals derive deeper insights make more precise decisions. This article explores application conversational intelligent reporting systems, leveraging artificial intelligence models, within industry. It examines how systems streamline processes, facilitate efficient communication, contribute informed decision-making, ultimately reshaping landscape operations.
Язык: Английский
Процитировано
8
Journal of Theory and Practice of Engineering Science, Год журнала: 2024, Номер 4(03), С. 183 - 190
Опубликована: Март 25, 2024
The integration of artificial intelligence (AI) with gene editing technologies like CRISPR-Cas9 holds immense promise for advancing biomedical research and personalized medicine. This article highlights the crucial role AI in predicting minimizing off-target effects, thereby enhancing precision efficiency editing. Researchers have developed algorithms BE-DICT to accurately predict base outcomes, showcasing potential AI-driven strategies optimizing processes. By combining bioengineering, this interdisciplinary approach aims automate refine DNA modifications, paving way innovative applications therapy biofabrication. Ultimately, endeavors revolutionize life sciences field, leading significant breakthroughs healthcare biotechnology.
Язык: Английский
Процитировано
5
Опубликована: Июнь 8, 2024
In addressing the computational and memory demands of fine-tuning Large Language Models (LLMs), we propose LoRASP (Streamlined Partial Parameter Adaptation), a novel approach utilizing randomized half-selective parameter freezing within Low-Rank Adaptation (LoRA) framework. This method efficiently balances pre-trained knowledge retention adaptability for task-specific optimizations. Through mechanism, LoRA-SP determines which parameters to update or freeze, significantly reducing requirements without compromising model performance. We evaluated across several benchmark NLP tasks, demonstrating its ability achieve competitive performance with substantially lower resource consumption compared traditional full-parameter other parameter-efficient techniques. LoRA-SP's innovative not only facilitates deployment advanced models in resource-limited settings but also opens new research avenues into effective efficient adaptation strategies
Язык: Английский
Процитировано
5
International Journal of Computer Science and Information Technology, Год журнала: 2024, Номер 2(1), С. 359 - 367
Опубликована: Март 24, 2024
With the development of network technology, APT(advanced persistentthreat) attacks are increasing, and research on security enterprise assets requires effective detection digital in space management through screening combing, which is key to real-time monitoring safe operation system. However, accompanying malware also poses a threat user's property privacy, so an method detecting Android necessary. In this direction, although feature processing capability traditional machine learning has been improved, there problems that extraction relies expert experience accuracy low. Therefore, paper combines deep reinforcement technology with operating system detect defend advance, as achieve security.
Язык: Английский
Процитировано
4
Опубликована: Март 4, 2024
Introductions.Packed scrubbers and electric filters are widely used in various industries for controlling air pollution removing particulate matter harmful gases from industrial emissions.Understanding their efficiency is crucial ensuring compliance with environmental regulations minimizing the impact of pollutants on public health environment.In this study, we conduct a comprehensive analysis packed to evaluate performance pollutant removal [1].Aim.The aim study analyze emissions.By assessing under different operating conditions loadings, provide insights into optimizing design operation improved quality control [2].Materials methods.The was conducted through combination experimental measurements computational simulations.Experimental involved collecting data inlet outlet concentrations, gas flow rates, pressure drops across settings.Computational simulations were performed using advanced modeling techniques simulate fluid dynamics mechanisms within [3][4].Results discussion.Our revealed significant emissions.We observed variations conditions, such as temperature.Furthermore, identified factors influencing ARCHITECTURE, CONSTRUCTION THEORETICAL AND PRACTICAL ASPECTS OF THE DEVELOPMENT SCIENCE EDUCATION 22 filters, packing material characteristics, field strength, residence time [5].Conclusions.In conclusion, our provides valuable control.By operational parameters features, selection, configuration, distribution, these devices can be enhanced.Future research may focus developing materials technologies improve reduce energy consumption.Overall, contributes advancement supports efforts mitigate emissions human environment.
Язык: Английский
Процитировано
1
Опубликована: Фев. 19, 2024
ПОНЯТТЯ ФІНАНСОВОЇ ДІЯЛЬНОСТІ В
Процитировано
0
Опубликована: Фев. 26, 2024
ВПЛИВ ОБРОБІТКУ ҐРУНТУ
Процитировано
0