Office Hours

A place for Brandon to chat with students and for students to chat with each other

Testing your code brings peace of mind

Testing a machine learning framework is challenging. When an ML algorithm runs, it operates on data to produce a result. It usually incorporates an element of randomness. Data is often unknown unfamiliar. It’s hard to automatically determine whether everything is working as it...

READ MORE

Cottonwood now lets you build complex models

The autoencoder project we built with Cottonwood in the 310 course sequence was able to get by using a sequential graph. For the most part, each layer connected to one layer ahead and one behind. Progression through the layers started at the beginning and sequentially proceeded through the...

READ MORE

Group enrollment discounts

A couple of companies sent me really thoughtful notes saying that their teams were getting a lot out of e2eml courses. That really made my day. Just a reminder: If you decide to provide e2eML courses as a WFH professional development resource for your teams, don't pass up the group...

READ MORE

Breaking Ground for Course 321: Convolutional Neural Networks in One Dimension

Hi everyone, I'm excited to announce that I've stared work on Course 321. It will be the first of several courses covering convolutional neural networks (CNNs) in end-to-end applications. CNNs have some unfamiliar concepts. In order to make them as clear as we possibly can, we're going to start...

READ MORE

Course Certificates

Q. Are there completion certificates at the end of each course? A. Yes! All the 200- and 300-series courses have completion certificates. I encourage you to list them on your resume. If any potential employers contact me to verify I'll be happy to confirm the material that you covered in the...

READ MORE

Mars demo user interface

In our autoencoder compression case study, I’m particularly proud of the fact that we took the time to think through how a new user might want to use the code in practice. We built a small API around the method, making it so that very little code is required do use it. Here is the final...

READ MORE

Cottonwood's first steps

With the completion of the 310-series of e2eML courses (312, 313, 314), Cottonwood has taken its first steps. If you'd like to skip to the final product, the autoencoder-based compression tool that the courses construct has been released as a case study. Cottonwood is a deep learning framework...

READ MORE

Course 314 is complete

This is a milestone for our curriculum. Where the 200-series courses traversed ML case studies end-to-end in a single course, neural networks were too big a topic. Instead, our autoencoder case study has been broken up into four separate courses, 311, 312, 313, and 314. Now the arc is complete,...

READ MORE

Research story with a twist

I looked for a neural network regularization method that limited the number of non-zero node activities (L0 group sparsity?) I wanted to incorporate it into our course 314 compression work. I couldn’t find one. So I made a layer that preserved the k node activities with the highest...

READ MORE

End-to-End Machine Learning online address is now officially e2eML.school

Hi all, We have a new online address! Update your bookmarks to e2eml.school The old brohrer.github.io content isn't going anywhere, but it's not going to be updated either. I've included some handy shortcuts too e2eml.school/courses (course listing) e2eml.school/library (tutorial...

READ MORE