Cristian Cruz-Carrasco

Plant diseases cause economic losses and health risks, such as aflatoxins linked to liver cancer. These toxins, produced by fungi like Aspergillus flavus in figs, are often detected late through invasive methods or visual inspection. Since Spain, particularly Extremadura, is a key fig producer, alternative detection methods are essential to preventing aflatoxins in the food chain. The aim of this research is the early detection of Aspergillus flavus fungus using non-invasive techniques with hyperspectral imaging and applying artificial intelligence techniques, in particular deep learning. The images were taken after inoculation of the microtoxin using 3 different concentrations, related to three different classes and healthy figs (healthy controls). The analysis of the hyperspectral images was performed at the pixel level. Firstly, a fully connected neural network was used to analyze the spectral signature associated with each pixel; secondly, the wavelet transform was applied to each spectral signature. The resulting images were fed to a convolutional neural network. The hyperparameters of the proposed models were adjusted based on the parameter tuning process that was performed. The results are promising, with 83% accuracy, 82.75% recall, and 83.25% F1-measure for the fully connected neural network. The high F1-measure demonstrates that the model’s performance is good. The model has a low incidence of false positives for samples that contain aflatoxin, while a higher number of false positives appears in healthy controls. Due to the presence of false negatives, this class also has a high recall. The convolutional neural network results, accuracy, recall, and F1 are 77.25%, indicating moderate model performance. Only class 3, with higher aflatoxin concentration, achieves high precision and low false positive incidence. Healthy controls exhibit a high presence of false negatives. In conclusion, we demonstrate the effectiveness of pixel-level analysis in identifying the presence of the fungus and the viability of the non-invasive techniques applied in improving food safety. Although further research is needed, in this study, the fully connected neural network model shows good performance with lower energy consumption.

Plant diseases not only cause economic losses, but they also pose a health risk. In this regard, aflatoxins are linked to the development of some type of liver cancer, especially the hepatocellular carcinoma. Aflatoxins are usually produced by the fungi Aspergillus flavus and Parsiticus in figs, and their identification is usually carried out using invasive and complex methods or by visual inspection at an advanced stage. Given that Spain is one of the main producers of figs and Extremadura is the leading region in the country, it is necessary to explore alternative ways of detecting the presence of aflatoxins in figs and preventing them from entering the food chain. The aim of this research is to address the early detection of Aspergillus flavus fungus using non-invasive techniques with the aid of hyperspectral imaging and the application of artificial intelligence techniques, in particular deep learning. The images were taken after inoculation of the microtoxin using 3 different concentrations, related to three different classes and healthy figs (healthy controls). The analysis of the hyperspectral images was performed at the pixel level. Firstly, a full connected neural network was used to analyse the spectral signature associated with each pixel; second, the wavelet transform was applied to each spectral signature. The resulting images are fed to a convolutional neural network. The hyperparameters of the proposed models were adjusted based on the parameter tuning process also performed. The results are promising, with 83% accuracy, 82.75% recall and 83.25% F1-measure for the single neural network, and 77.25% accuracy, recall and F1 for the convolution neural network.

With the advent of next-generation networks, it is crucial to persist in the research and development of key enabling technologies such as software-defined networking (SDN). This involves assessing prospective network deployments, mechanisms, or ideas; an undertaking performed by both network operators and academia to assess the advantages and limitations of the developed proposals related to programmable networks. In this context, simulators are envisioned as essential tools for replicating experiments, offering the required realism, adaptability, and scalability within a controlled environment. However, current solutions have limitations related to the SDN capabilities and indicators that allow for optimizing network performance, which is crucial for Beyond 5G (B5G) and 6G. To overcome this challenge, we propose SDNSimPy, a Python-based simulation framework built on a discrete event simulator. The proposed simulator features a modular architecture with various functional abstractions related to programmable networks, which have been partitioned into distinct modules to streamline its development and facilitate future extensions. Moreover, SDNSimPy has undergone a verification phase to check its implementation. Results obtained from the simulator reveal a significant distinction in the operation modes (proactive and reactive) with respect to end-to-end delay. This parameter is crucial in Beyond 5G (B5G) services and can impact the quality of service (QoS) of network communications.