- May 1, 2019
When working on classification problems, it is very common to use the softmax function to transform the output’s logits into a probability for each class. It is also important that the probability given to the predicted class reflects how certain we can be of that result. That is not the case when a model is uncalibrated.
- Apr 23, 2019
During college I had the chance of creating a group about machine learning with a couple of friends. Together, we were responsible for preparing workshops, organizing events with invited lecturers, etc. One of the cool things that also came out of that project was a list of resources.