Machine Learning In a Nutshell
What is Machine Learning? Machine learning is a branch of Artificial Intelligence (AI) and computer science which focuses on the use of data and algorithms to imitate the way humans learn, gradually improving its accuracy. To borrow a description from TensorFlow’s “Intro to Machine Learning” video series, traditional programming involves programming complex rules into a computer program that are used to analyze input data and output an answer. If the input data is an image of a flower, and if the programming/rules can recognize the flower, then it outputs the answer “flower.” Having a traditional program recognize the differences between multiple different kinds of flowers would require significantly more complex programming, especially if the images are allowed to be at various angles and orientations, and not directly centered in the image. Instead, machine learning focuses on providing examples to a machine learning algorithm or “model” – providing data and answers – and allowing the model to build its own rules to determine the relationships between the examples provided to it. Just like a human, during each “step” of the training process the model makes a refined guess about the relationships between the known examples and then tests those guesses against examples not yet seen. By training a model over successive steps, the model attempts to improve its accuracy in correctly identifying previously unseen variations of the data. In this way, training a model to correctly recognize multiple types of data requires no more source code than recognizing a single type, it only requires creating more examples for the model to learn.
In FIRST Tech Challenge, the machine learning platform used is TensorFlow. TensorFlow is an open source platform for machine learning with a comprehensive, flexible ecosystem of tools, libraries, and community resources to enable developers to create tools such as the FIRST Tech Challenge Machine Learning tool. TensorFlow has been utilized in FIRST Tech Challenge for a number of years, allowing teams to recognize individual game pieces and clusters of game pieces via pre-built models developed by FIRST Tech Challenge engineers. Now FIRST Tech Challenge is empowering teams to build their own custom models!