"Automated Phenotyping and Genotype-Phenotype Association for Accelerating Plant Breeding via..."

"Automated Phenotyping and Genotype-Phenotype Association for Accelerating Plant Breeding via Machine Learning: A Report from the TERRA Project" -- Naoki Abe, Research Staff Member and Senior Manager, IBM The TERRA (Transportation Energy Resources from Renewable Agriculture) project, a large scale project funded by Department of Energy's ARPA-E, involving a dozen of centers and spanning over 3 and a half years, is coming to its completion. Several members of our machine learning team from IBM Research participated in this project, in collaboration with Purdue University (and University of Queensland), developing machine learning approaches to help with the overall project goal - to develop an automated high-throughput system for determining how variations in the sorghum genome impact field performance, agricultural productivity and energy potential as biofuel. The two main technical challenges we faced in achieving this goal are (1) automating phenotyping using sensor data from ground-based mobile and airborne platforms, and (2) integrated analytics on high-dimensional genomic data and multi-modal field image/sensor data. In this talk, I will describe the progress we made in both of these areas by developing and applying advanced machine learning approaches, focusing on areas contributed primarily by IBM Research. The talk reflects work done by Dr. Peder Olsen, Dr.Aurelie Lozano, Dr. Karthikeyan Ramamurthy (IBM Research), Min-Hwan Oh (Columbia U.), Prof. Mitch Tuinstra (Purdue U.), Prof. Addie Thompson (Michigan S. U.), Dr. Javier Ribera (Purdue U.) among many others.