Prediction of outcomes is critical in both swine breeding and management. This necessitates the development of predictive models that address challenges in swine farming. For predictive modeling, there have been significant advances in deep learning. Nevertheless, there are needs to adapt deep learning-based models for specific swine farming problems including genomic prediction and behavior analysis. Furthermore, there is not yet a clear guideline on how to validate a model in this field. The overarching goal of this dissertation was to validate a collection of predictive models for improved swine farming with applications to precision management, phenotyping, and breeding. The first study addressed the pig genomic prediction problem. Differential evolution was utilized to optimize deep learning (DL) hyperparameters that affected the predictive performance of DL models. Performance of optimized DL was compared with "best practice" DL architectures selected from literature and baseline DL models with randomly specified hyperparameters. Optimized models showed clear improvement. Further, differential evolution saved considerable time compared to traditional optimization approaches e.g., grid search. Despite the success of genomic prediction, phenotyping has become a bottleneck in breeding programs as it is still time-consuming and labor-intensive. Computer vision (CV) can be used to automate the phenotyping process. Nonetheless, there are limited amount of public data for CV development in livestock farming. Most published CV applications to livestock farming were developed using rather small datasets, and their broader validity remained unknown. Therefore, the second study aimed at reviewing publicly available image datasets that were used for CV algorithms in livestock farming and the validation methods in the related work. Through the review, we could not find public datasets that addressed pigs' agonistic behaviors (negative social behaviors), which is an important topic in swine farming. Given this, the third study aimed at collecting a video dataset to study pig's agonistic behavior and adapting a state-of-the-art DL pipeline to classify pigs' agonistic behaviors through video analysis. The pipeline was validated through various training-validation data partitions, where the training data were used for model development and the validation data were used for model evaluation. Results showed that splitting the training and validation sets at random led to over-optimistic estimates of model performance. The last study focused on developing and validating a statistical model for the analysis of pigs' social interactions. Generalized linear mixed models were fitted, and a Bayesian framework was used for parameter estimation and posterior predictive model checking. The predictive performance of the models varied depending on the validation strategy, where three strategies were defined: random cross-validation, block-by-social-group cross-validation, and block-by-focal-animals validation. In conclusion, this dissertation provides information about how state-of-the-art models can be adapted for and validated in swine farming applications. Future directions of this research could aim at creating reference imagery datasets in swine farming that provides a platform for CV applications and developing integrated computer vision systems, which eventually assists in prediction tasks for improved pig management and breeding.
Read
