Machine learning has enabled computer systems to detect skin cancer, sort cucumbers, and even detect escalators in need of repair. In this Cloud AI Adventures episode, Yufeng Guo walks us through a basic example of machine learning and how to use it to get answers from data. The process includes gathering data, preparing it for use, choosing a model, training it, evaluating it, tuning hyperparameters, and finally, using the model to predict or infer something useful. By the end, we learn that the power of machine learning lies in its ability to differentiate between objects using models rather than relying on human judgment and manual rules.
1. Machine learning has granted computer systems new abilities.
2. Machine learning involves training a model to accurately answer questions based on collected data.
3. The training data for a model that determines whether a drink is beer or wine includes the color and alcohol content of each drink.
4. In the training phase, the model's values are adjusted based on the accuracy of its predictions.
5. The evaluation phase tests the model's accuracy on data that it has not seen before.
6. Hyperparameters, which affect the model's performance, can be adjusted during the training phase.
7. The final step in machine learning is prediction or inference.
8. The TensorFlow Playground is a browser-based machine learning sandbox for experimenting with different parameters.
9. Future episodes will delve deeper into the nuances of machine learning.