With the completion of the Advanced Neural Networks Methods course, the Cottonwood machine learning framework is becoming a useful playground for experimentation. We haven’t yet added convolutional layers or recurrent neural networks (that will happen in the upcoming series of courses) but we are putting in a firm foundation of the basics including fully connected networks, and a healthy variety of optimizers and regularizers.

My hope for Cottonwood is that it will become a tool for exploring neural network concepts and machine learning experimentation. As you work your way through the existing neural networks courses on the fundamentals ( Course 312 and 313) and the upcoming courses on hyperparameter tuning, convolutional neural networks, and recurrent networks (Courses 314, 315, and 316) that you will try out variations and ideas of your own. When you build a thing, feel free to share them with me! If you are comfortable with it, I will be pleased to share it around my network. But even if you prefer to keep it quiet, I would really love to see the things you’ve built.

Good luck on your learning path.

Happy building!