How Machine Learning Shapes the Future of Agriculture

2019-10-17

In 1962, Arthur Samuel shocked the world. He created a computer that could challenge the reigning checkers champion, Robert Nealy. The machine emerged victorious, but the significance of this victory goes beyond the game itself. Samuel did not program all the potential moves on the checkerboard into the computer; instead, he programmed it to respond based on past game instructions. Today, the principles of "machine learning" that Samuel applied in his early experiments are used in almost every field, including one seemingly unrelated to artificial intelligence.



Using Data to Grow More Efficiently


Botanists have been searching for methods to help farmers grow crops more efficiently while using fewer natural resources. However, to ensure that plants retain beneficial traits, researchers must identify the correct gene sequences. The mystery of which sequence is the correct one has persisted. Machine learning algorithms can develop probability models based on decades of raw data, such as the performance of crops in various climates or how they inherit specific traits. These models can predict which genes are most likely to express beneficial traits, greatly narrowing the search among millions of combinations using digital tools and advanced software.



Testing Potential Scenarios on Farms


With machine learning, plant breeding becomes more accurate, efficient, and capable of evaluating a wider range of variables. Scientists are using computer simulations for early testing to assess how different varieties perform under various climates, soil types, weather patterns, and other factors. This digital testing allows botanists to bypass cumbersome physical experiments and more accurately predict crop growth.


Nalini Polavarapu, Head of Data Science Strategy at Bayer, said, "By using machine learning, we can save an entire year in the testing phase. For farmers, machine learning will help them make 40 critical decisions, including planting, irrigation, disease prevention, pest control, and weed removal."



Finding Solutions with Pattern Recognition


Early and accurate identification is crucial when tracking any plant disease. Traditional methods of identifying plant diseases rely on human recognition, which is inefficient and prone to errors. For a trained computer, diagnosing plant diseases is essentially pattern recognition. After curating hundreds of thousands of photos of diseased plants, machine learning algorithms can assess disease types, severity, and many other factors.


Machine learning in agriculture enables more accurate disease diagnosis, resource conservation, and prevention of false data. Farmers can upload images of diseased plants to the Climate FieldView™ platform, where machine learning can identify potential issues on the farm and propose management plans.



The Future of Machine Learning


Today, both small startups and large organizations are using machine learning to shape the future of agriculture. By combining machine learning with human creativity, a $5,000 supercomputer theoretically has the potential to make significant breakthroughs in plant breeding.


Similar to software, the improvements in machine learning seem limitless. Agricultural researchers are testing their theories on a larger scale and helping make more accurate predictions about crops. Through machine learning, we have the possibility to create more methods to sustain our growing world, adapt to climate change, and protect our water, land, and other resources.

RECENT AI NEWS

RECENT AI TOOLS