• I'm watching TensorFlow tutorial on FreeCodeCamp and it's 6 hours long
• I realized that there are three types of learning: supervised learning, unsupervised learning, reinforcement learning.
• Supervised is like a classifier. Given a machine knows that a good portion of children survived the titanic. What are the chances a random will survive the titanic?
• Unsupervised learning is like having data points and trying to find the "center of mass" at each of these points. I honestly don't know too much about this one besides the fact that the autoencoder uses this.
• Reinforcement learning is like what we think of when a machine plays tetris, snake, or chess.
• I currently know of two algorithms for supervised learning: Linear Regression and DNN Classifier
• Linear Regression just tried to make a best fit line given a bunch of points. You now have a line predicting the relationship between 2 (or more) variables based on a bunch of existing points
• DNN (Deep Neural Network) classifier is what I think of when I think of Neural Network. It takes a neural network with weights and biases. It changes based on some linear algebra and calculus stuff when there is a difference between its current guess and the training data
• I am learning one unsupervised learning algorithm called K-near classifier which takes the "center of mass" of some points