https://repositorio.ifpe.edu.br/xmlui/handle/123456789/492 Fri, 17 Apr 2026 07:36:42 GMT 2026-04-17T07:36:42Z https://repositorio.ifpe.edu.br/xmlui/handle/123456789/2017 Avaliação do uso de grandes modelos de linguagem para detecção de alucinações. Recently, generative artificial intelligence is quickly advancing. However, despite great investment in improvements, the output generated still obtains an elevated level of hallucinations, compromising the reliability of the content. This work aims to mitigate this problem by offering a comparison between models Llama and GPT when used to detect hallucinations. After checking accuracy values, both LLMs performed similarly, with the two models displaying 84% and 68% for input and context hallucinations. For factual hallucinations, there was a 16% difference in results. Finally, results suggest that using GenAI generated content without further human analysis is not recommended for complex activities. Thu, 18 Dec 2025 00:00:00 GMT https://repositorio.ifpe.edu.br/xmlui/handle/123456789/2017 2025-12-18T00:00:00Z https://repositorio.ifpe.edu.br/xmlui/handle/123456789/1987 Desenvolvimento de Aplicativo para Análise de Padrões de Vibração de Compressores This work presents an application for analyzing vibration patterns in compressors used in air conditioning systems. The proposal provides an affordable solution by utilizing low-cost sensors (on the order of BRL 30.00) to assist in fault detection and predictive maintenance of these devices, promoting greater operational efficiency and reduced maintenance costs. The application is capable of capturing, processing, and analyzing vibration signals generated by compressors, delivering accurate and timely diagnostics regarding the health status of the equipment. The vibration signals are displayed through an interface that enables intuitive data visualization and interpretation, ensuring that professionals in the field can make quick and informed decisions. Thu, 13 Nov 2025 00:00:00 GMT https://repositorio.ifpe.edu.br/xmlui/handle/123456789/1987 2025-11-13T00:00:00Z https://repositorio.ifpe.edu.br/xmlui/handle/123456789/1931 Desenvolvimento de um Bot Inteligente para Atendimento ao Cliente com Base em Perguntas Frequentes Usando o Modelo GPT com Fine-Tuning. The growing presence of technological advancements across various fields of study, production, and service delivery in society is remarkable, and among these, the customer service sector stands out. Given the need to optimize the customer journey and enhance customer loyalty and satisfaction while also keeping up with its competition, more and more companies are investing in technologies aimed at improving customer service. Among these technologies, chatbots are particularly noteworthy, as they allow the automation of certain stages of the customer journey while striving to maintain the desired quality and satisfaction. This work addresses the development of an intelligent customer service bot, based on fine-tuning techniques applied on the GPT model. The training dataset consists of frequently asked questions (FAQs), extracted from real customer interactions within a Customer Relationship Management (CRM) system. The proposal aims to meet the system's demand to deliver a satisfactory and efficient service experience, thereby aligning with the necessary aspects to remain competitive in the customer service market. Fri, 31 Oct 2025 00:00:00 GMT https://repositorio.ifpe.edu.br/xmlui/handle/123456789/1931 2025-10-31T00:00:00Z https://repositorio.ifpe.edu.br/xmlui/handle/123456789/1928 Estudo do efeito de variáveis de estado e funções de recompensa no desempenho de algoritmos de enxames combinados com aprendizagem por reforço. This work investigates the integration between Reinforcement Learning and Swarm Intelligence applied to optimization problems, focusing on the analysis of the impact of state variables and reward functions on the performance of agent combinations. Swarm Intelligence, inspired by the collective behavior of animals, seeks solutions through the decentralized cooperation of agents, while Reinforcement Learning teaches an agent to make decisions by trial and error, optimizing rewards accumulated in the interaction with the environment. The study adopts an approach in which a Proximal Policy Optimization agent is responsible for dynamically selecting between three swarm metaheuristics: Global Particle Swarm Optimization, Local Particle Swarm Optimization, and Grey Wolf Optimizer. The experimental environment was developed by incorporating variables associated with swarm behavior and two reward functions: Reward 1, which already exists and is based on incremental fitness improvement, and Reward 2, proposed in this study to penalize stagnation. The methodology involved applying the ablation technique, allowing the evaluation of the relevance of groups of state variables in learning. The experiments were conducted on benchmark functions, named F1 and F2, under different dimensionalities (10, 30, and 50), in order to identify how the configurations of observables and rewards influence the adaptation and convergence of the agent in optimization scenarios. The results showed that Reward 1 stood out for its stability and consistent performance, while removing fitness variables reduced the computational cost without compromising convergence. Tue, 18 Nov 2025 00:00:00 GMT https://repositorio.ifpe.edu.br/xmlui/handle/123456789/1928 2025-11-18T00:00:00Z