Produção Científica e Didática
Produção científica recente
Framework to Detect Adulterated Honey Using Hyperspectral Imaging and Machine Learning
Honey is a natural product renowned for its nutritional and medicinal properties. However, the increasing market demand in Pakistan has led to frequent instances of adulterated and fraudulent honey. Conventional detection methods are widely used but are often ineffective in accurately identifying adulteration. This study proposes a novel approach to detecting adulteration in honey, specifically the mixing of sugar, using machine learning (ML) techniques in conjunction with hyperspectral imaging (HSI). By leveraging HSI, we capture distinct spectral features of both pure and adulterated honey. These spectral features are processed and analyzed using ML algorithms trained on a dataset comprising pure and mixed honey samples. Experiments involve binary, as well as, multiclass (5 classes) honey samples with different levels of adulteration. In addition, data balancing is also carried out using synthetic minority oversampling technique (SMOTE). The results prove the proposed approach to be efficient in detecting adulteration across various honey varieties, with linear regression achieved 99.9% while support vector machine and multilayer perceptron obtaining a 98.7% accuracy for binary class. For the multiple classes of honey, the multilayer perceptron shows a 96.67% accuracy outperforming the remaining ML and deep learning models. Among the deep learning models, the convolutional neural network (CNN) achieved the best performance with an accuracy of 94.67% after applying the SMOTE × 3 (three times of over samples of original samples of each class) data balancing strategy, whereas recurrent neural network (RNN), long short-term memory (LSTM), and gated recurrent unit (GRU) models showed comparatively lower performance.
Cryptocurrency Market Trends: A Machine Learning-Driven Time Series Forecasting with Twitter Sentiment Integration
Accurate forecasting of cryptocurrency prices remains an open challenge because classical statistical models cannot capture the non-linear, sentiment-driven dynamics of these markets. This study compares three hybrid deep learning architectures—VAR-LSTM, XGBoost-LSTM, and CNN-LSTM—to determine which best forecasts Bitcoin (BTC), Ethereum (ETH), and Dogecoin (DOGE) closing prices, and to quantify the marginal predictive value of Twitter sentiment integration. Six years of hourly OHLCV data (2017–2023) are augmented with VADER-scored Twitter sentiment polarity. Each model is formulated mathematically, implemented with documented hyperparameters (epochs, dropout, units;), and trained for one-step-ahead next-hour price prediction. Performance is measured by RMSE, MAE, MAPE, R 2 , and Directional Accuracy (DA) across five random seeds, with paired Wilcoxon significance tests. XGBoost-LSTM achieves the best performance (RMSE = 81.547, R 2 = 0.9254, DA = 80.0%), outperforming all nine literature baselines. Removing Twitter sentiment degrades DA by 14.3 percentage points ( p < 0.01 ), confirming that social media signals carry independent predictive information. Hybrid architectures consistently outperform single-model baselines; XGBoost-LSTM offers the best accuracy-to-compute ratio. VADER-enriched Twitter sentiment is a significant predictor beyond price history. Limitations include reliance on a single sentiment platform and a training window that predates several structural market events
Parental Decision-Related Factors Are Associated with Discretionary Ultra-Processed Food Consumption Among Children and Adolescents Living in the Mediterranean Area
Background/Objectives: Nutrition during childhood and adolescence is a key determinant of long-term health, influencing metabolic homeostasis, neurocognitive development, and immune system maturation. Globalization and technological advances have reshaped food production and consumption, increasing the availability of ultra-processed foods (UPF) of low nutritional quality. This study aimed to investigate the relationship between parental factors, namely food literacy, perceived barriers and enablers, dietary attitudes, and healthy eating behaviors, and the consumption of discretionary UPF among children and adolescents living in 5 Mediterranean countries. Methods: This cross-sectional study was based on a survey completed by 2011 parents of children and adolescents aged 6–17 years from 5 Mediterranean countries, who reported on their children’s dietary and lifestyle habits. Adherence to the Mediterranean diet was assessed through the KIDMED index. Parental food literacy was measured using the Short Food Literacy Questionnaire (SFLQ). Perceived barriers and enablers were assessed based on the Theory of Planned Behavior, and parents’ attitudes toward their child’s diet were evaluated using the Healthy-Eating Attitudes Questionnaire (HEAQ). Finally, the Theory of Internet Use Related to Health (TIUH) questionnaire was used to assess parents’ tendencies related to health information use online. Results: Higher perceived barriers and enablers were significantly associated with lower discretionary UPF consumption across all models. Parental food literacy (SFLQ) showed a positive association with discretionary UPF consumption, remaining significant in the fully adjusted model, although with reduced magnitude. Healthy-eating attitudes (HEAQ) were initially positively associated with discretionary UPF intake but lost statistical significance after full adjustment. Regarding health-related internet use (TIUH), the Health Information dimension showed a strong positive association with discretionary UPF consumption, while other dimensions (Consciousness and Beliefs) showed inconsistent and non-significant associations in the fully adjusted model. Conclusions: Children’s consumption of discretionary UPF is shaped by several interrelated factors, such as family environment, eating patterns, and parents’ perceptions, rather than solely by knowledge or attitudes.
The legal governance of ai upskilling and reskilling: towards a new employer duty
This article examines whether the rapid adoption of artificial intelligence (AI) justifies a broader interpretation of employer responsibilities regarding continuous upskilling and reskilling. An interdisciplinary qualitative methodology was employed, combining doctrinal-comparative legal analysis with a systematic literature review conducted according to the PRISMA 2020 guidelines. Scientific evidence was collected from Scopus, Web of Science, ScienceDirect and SpringerLink, complemented by reports from the International Labour Organization (ILO), the Organisation for Economic Co-operation and Development (OECD) and the World Economic Forum (WEF). The findings indicate that AI-driven workforce transformation extends beyond technological innovation and significantly affects employability, requiring organisations to anticipate skills disruption through strategic workforce development. The study identifies a conceptual gap between labour law, strategic human resource management and corporate governance concerning employer responsibility for workforce capability development. To address this gap, it proposes the AI Employability Governance Framework (AEGF), an original model integrating strategic workforce foresight, continuous upskilling, reskilling pathways, shared employer responsibility, labour rights protection and sustainable employability. The framework contributes to the emerging debate on responsible AI governance by providing a conceptual foundation for more inclusive, resilient and legally informed models of workforce transformation.
Advanced Wafer Hotspot Detection through Image Segmentation and Stacked Model
The wafer map is a data visualization of a thin semiconductor fabric made of crystalline silicon, such as defects or test results. The wafer map is a base for creating electronic coordinate circuits and photovoltaic cells. During the wafer map production, any fault results in a product failure. The wafer map faults are undetectable to the naked eye, which is a big challenge. Hotspot detection in wafer maps is significantly important to evaluate the manufacturing process and. improve product yield. The hotspot detection in the wafer maps is the primary aim of this research. A novel wafer map hotspot detector (WHD) is proposed based on three stack fully connected conventional neural network layers and a dense layer. Data augmentation uses the segmented images of the wafers to build the proposed model. The proposed model is evaluated through several evalua-tion parameters and state-of-the-art studies comparative analysis. The proposed model achieved a 94% training and 90% testing performance accuracy for hotspot detection and shows better results than existing approaches. This study helps semiconductor engineers improve wafer manufacturing designs and efficiency in the semiconductor industry.