In this study, the performance of a set of machine learning algorithms was evaluated to classify the freshness levels of frozen-thawed shrimps during storage in refrigerator conditions. The algorithms used include CNN, DenseNet121, InceptionV3, ResNet50, CNN + LSTM, and YOLO_v8. Shrimps were categorised into three classes according to their freshness: fresh, medium, and stale. According to the results, the CNN + LSTM model was the most successful algorithm with a 99.07% accuracy rate. The YOLO_v8 algorithm stood out with 99.23% accuracy in the validation set. However, it was observed that the classification performances of models such as DenseNet121 and ResNet50 were relatively low. In addition, to increase the explainability of the decision processes, the Grad-CAM algorithm was used to visualise the areas that the CNN + LSTM model considers when determining the freshness states. Grad-CAM effectively emphasised visual cues focusing on the surface texture and signs of deterioration of the frozen-thawed shrimps. The results of the study showed that the proposed algorithms provide a non-destructive and accurate method for shrimp freshness classification. This approach offers a potential solution to improve food safety in the seafood industry.

The primary aim of this study is to predict the Marshall stability and flow values of hot-mix asphalt samples prepared according to the Marshall design method using regression-based machine learning algorithms. To overcome the limited number of experimental observations, synthetic data generation was applied using the Conditional Tabular Generative Adversarial Network (CTGAN), while the structural consistency of the generated data was validated through Principal Component Analysis (PCA). Two datasets containing 17 physical and mechanical input variables were analyzed, and multiple regression models were compared, including Extra Trees, Random Forest, Gradient Boosting, AdaBoost, and K-Nearest Neighbors. Among these, the Extra Trees Regressor consistently achieved the best results with near-perfect accuracy in flow predictions (MAE ≈ 4.06 × 10⁻¹⁵, RMSE ≈ 4.97 × 10⁻¹⁵, Accuracy ≈ 99.99%) and high performance in stability predictions (MAE = 109.52, RMSE = 150.67, accuracy = 90.45%). Furthermore, model interpretability was ensured by applying SHapley Additive Explanations (SHAP), which revealed that parameters such as softening point, VMA, penetration, and void ratios were the most influential features. These findings demonstrate that regression-based ensemble models, combined with synthetic data augmentation and explainable AI methods, can serve as reliable and interpretable tools in asphalt mixture design.