Pattern classification of ingestive behavior in grazing animals has extreme importance in studies related to animal nutrition, growth and health. In this paper, a system to classify chewing patterns of ruminants in in vivo experiments is developed. The proposal is based on data collected by optical fiber Bragg grating sensors (FBG) that are processed by machine learning techniques. The FBG sensors measure the biomechanical strain during jaw movements, and a decision tree is responsible for the classification of the associated chewing pattern. In this study, patterns associated with food intake of dietary supplement, hay and ryegrass were considered. Additionally, two other important events for ingestive behavior were monitored: rumination and idleness. Experimental results show that the proposed approach for pattern classification is capable of differentiating the five patterns involved in the chewing process with an overall accuracy of 94%.

Thousand-seed weight (TSW) is used to calculate the sowing density, number of seeds present in a sample, and predict seed size. Our aim was to establish an alternative methodology for thousand-seed weight determination in forest species using image analysis techniques. The traditional methodology was compared with the alternative one, testing them on seeds from 16 forest species. The traditional method was performed by manually counting eight repetitions of 100 seeds, weighing of samples, and computation of thousand-seed weight. The alternative methodology was performed by counting seed samples captured by images and processing in ImageJ® software, followed by weighing a single sample, and computation. All steps were timed, and each methodology was repeated ten times per species. The TSW obtained by the alternative methodology was similar to that obtained by the traditional one, and the average execution time of the activities was reduced by 62%. The proposed thousand-seed weight determination by image analysis is an efficient and optimized alternative to the traditional method. Thousand-seed weight determination from image analysis reduces execution time for the seed analyst and forest seedling producer.

A major concern in multi-agent coordination is how to select algorithms that can lead agents to learn together to achieve certain goals. Much of the research on multi-agent learning relates to reinforcement learning (RL) techniques. One element of RL is the interaction model, which describes how agents should interact with each other and with the environment. Discrete, continuous and objective-oriented interaction models can improve convergence among agents. This paper proposes an approach based on the integration of multi-agent coordination models designed for reward-sharing policies. By taking the best features from each model, better agent coordination is achieved. Our experimental results show that this approach improves convergence among agents even in large state-spaces and yields better results than classical RL approaches.

The identification of seeds from native species is a complex assessment due to the high Brazilian biodiversity and varied characteristics between species. The objective was to apply different machine learning classifiers associated with image analysis to identify seeds of forest species. In total, 155 native species belonging to 42 botanical families were analyzed. In addition, to determine the appropriate machine learning classifier, five supervised learning classification techniques were implemented: decision trees (DT), artificial neural networks (ANN), k-nearest neighbors (k-NN), Naive-Bayes classifier (NBC) and support vector machine (SVM), which had their performance evaluated. For modeling, 66% of the seeds’ morphobiometric data were used to train the classifiers, while 34% were reserved for validation. The classifiers are promising tools for identifying species from seed images. The decision tree (DT) classifier showed greater accuracy for correct species identification (82.8%), followed by ANN (81.7%), k-NN (81.7%), NBC (81.1%) and SVM (78.7%). Therefore, it is possible to identify seeds of native species from images and machine learning with a satisfactory accuracy rate. Finally, the decision tree classifier is recommended.

Biometric seed analysis can be used to characterize and differentiate forest species. However, forest species are generally studied using manual methods such as measurements with a digital caliper, which provides a limited amount of information on plant morphological characteristics, whereas agronomic species are analyzed using expensive and often inaccessible equipment. Thus, the objective of the present study was to demonstrate that seed image analysis and processing tools can help characterize and differentiate Brazilian forest species. For this purpose, the seeds of 155 forest species belonging to 42 families were photographed and analyzed to extract data on their morphometric descriptors using a new methodological approach. A total of 18 characteristics were assessed, namely eight dimensions, four shape characteristics, and six color characteristics. A set of approximately 1.827 million data was extracted from 101,521 seed images. Digital image processing efficiently characterized the studied seeds and the obtained characteristics allowed us to differentiate between species, including those belonging to the same botanical family and genus. Therefore, seed image analysis using the proposed methodology can be used to characterize, differentiate, and automatedly identify forest species in Brazil.