Skip to main content
ARS Home » Plains Area » Clay Center, Nebraska » U.S. Meat Animal Research Center » Livestock Bio-Systems » Research » Publications at this Location » Publication #347181
Title: Incorporating deep learning into the analysis of diverse livestock data

Author
item Keel, Brittney

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 10/23/2017
Publication Date: 11/13/2017
Citation: Keel, B.N. 2017. Incorporating deep learning into the analysis of diverse livestock data. [Abstract]. In Proceedings: Livestock High-Throughput Phenotyping and Big Data Analytics (Livestock HTP and Big Data), Nov. 13-14, 2017, Beltsville, MD. www.animalgenome.org/share/meetings/LivestockHTP.

Interpretive Summary:

Technical Abstract: Technological advances in high-throughput phenotyping and multiple omics fields have led to an explosion in the volume of data across the whole spectrum of biology, allowing researchers to integrate data of different types to inform hypotheses and expand the scope of their research questions. However, the rapid increase in biological data dimension and the complexity of the relationships between various data sources pose challenges to conventional analysis strategies. Machine learning methods are broadly applicable approaches to learn functional relationships from data without the need to define them a priori. In the past, the application of machine learning to high dimensional data sets has been limited because deriving the most informative features of these data sets can be highly labor intensive due to the vast number of input combinations. Deep learning techniques, such as artificial neural networks, have recently emerged as a solution to this bottleneck. Deep learning is one of the most active fields in machine learning, and it has been shown to improve performance in image and speech recognition. Deep learning could greatly enhance the field of livestock science by allowing researchers to better exploit the availability of increasingly large and high dimensional data sets (e.g. high-throughput phenotypes, genomics, proteomics, transcriptomics, and metabolomics). Here, we present a broad overview of deep learning and discuss potential applications to livestock data. In particular, we discuss the application of deep artificial neural networks to the prediction of swine feeding behavior based on thermal conditions.