Machine Learning:

How A Game of Checkers is Shaping Agriculture

In 1962 Arthur Samuel shocked the world. He built a computer that could challenge then-reigning checkers champion, Robert Nealy. The machine won, but it wasn't the triumph alone that grabbed headlines. It was the software behind the victory that would change the world. 


Rather than programming the 500 quintillion2 potential scenarios from a checkerboard into his computer, he instructed the device to react based on games it had played in the past. After playing game after game — weighing dozens of factors, calculating risk, and planning the next, most efficient moves — the computer “learned” to master the board.  


Today, the same principles of “machine learning” that Samuel applied in his early experiments are being used in almost every discipline — including one field that is not often associated with the rise of artificial intelligence. Agriculture.


Using data to grow more efficiently

Plant breeders are constantly searching for specific traits. They look for qualities that help farmers grow crops more efficiently while using fewer natural resources. For a plant to inherit a beneficial trait, however, researchers must find the right sequence of genes. But exactly which sequence is the right one, is a mystery in the beginning.  


Deep learning algorithms can take a decade of raw field data — like insights about how crops have performed in various climates, or how they have inherited certain characteristics — and use this data to develop a probability model. With all of this information, machine learning can predict which genes will most likely contribute beneficial traits. Of the millions of combinations, digital tools and advanced software greatly narrow the search. 


Testing what’s possible on the farm

With the aid of machine learning, plant breeding is becoming more accurate, efficient, and capable of evaluating a wider set of variables. Scientists are using computer simulations to conduct early tests to evaluate how a variety may perform when faced with different sub climates, soil types, weather patterns, and other factors. This digital testing does not replace physical field trials but allows plant breeders to more accurately predict the performance of crops. By the time a new variety reaches the soil, machine learning has helped breeders create a more thoroughly vetted product than ever before. 


"We were able to save an entire year of testing in our pipeline by using machine learning,” says Nalini Polavarapu, Enterprise Data Science Strategy Lead at Bayer, "For a farmer, machine learning will help create personalized answers to 40 key decisions they make in a growing season — from planting to in-season management of irrigation, diseases, pests, and weeds to harvesting."


Using patterns to find solutions

When tracking any disease, early and accurate identification is essential. The traditional method of identifying plant disease is done by visual examination. However, this process can be plagued with inefficiencies and prone to human error. For a trained computer, diagnosing plant disease is essentially pattern recognition. After sorting through hundreds of thousands of photos of diseased plants, a machine learning algorithm can assess disease type, severity, and a number of other issues.


Machine learning in agriculture allows for a more accurate disease diagnosis while preserving energy and preventing false data. Farmers can upload field images taken by satellites, UAVs, land based rovers, smartphones, and tools like the Climate FieldView™ platform, which can identify potential issues on the farm and recommend a management plan.


Feeding our growing population 

Crop disease is a major cause of famine and food insecurity around the world.3 Modern agriculture seeks to create seeds and crop protection products that can provide relief to these global challenges. 


One of the many benefits of machine learning is how this technology can make more accurate and precise improvements to a process. In plant breeding, machine learning is helping create more efficient seeds. Such advancements offer the potential to create even more adaptable, productive crops that better feed the planet, all while preserving our precious natural resources. 


The future of machine learning

What used to be reserved for major institutions is now within reach for all. Small startups and large organizations alike are using machine learning to shape the future of agriculture. When paired with human ingenuity, a $5,000 supercomputer could theoretically create a huge breakthrough in plant breeding. Just a decade ago, this notion was a fantasy. 


Much like software, improvements in machine learning have seemingly endless possibilities. Researchers in agriculture are testing their theories on a greater scale and helping make more accurate, real-time predictions about crops. What we can imagine through machine learning has the potential to uncover even more ways to feed our growing world, adapt to climate change, and conserve our water, land, and energy.

4 min read