We discovered that simply creating deeper neural networks will not always result in a significant increase in accuracy. Here I show that the right activation function can help boost performance, especially for deep networks.
I stumbled upon Martin Gorner's introduction to TensorFlow on Youtube and I loved it so much that it inspired me to create this notebook for everyone who wants to learn about neural networks and how to write it into code.
The Google Cloud Platform is a hot mess right now with the more recent offerings getting better documentation. For example, it was really confusing to get the App Engine standard SDK working on my Mac because first of all, there are two App Engines, and the way the SDK installs is really weird. Solving this issue is a real pain especially if you don't know where to look.
Bioinformatics and computational biology have been gaining popularity even with my benchtop biologist colleagues. To help them out, I recently gave a short seminar on how programming in Python can help solve common bioinformatics problems like calculating a sequence's GC content. Check out my lecture I summarized as an IPython notebook if you too want to learn more about programming.
One thing I love about PyCharm is that it truly is an integrated developement environment in that you can do most of your work from creating virtual environments and installing Python modules to version control without ever leaving the IDE. However, when a crucial part of your stack updates its API or drastically changes how it works, then that tight integration becomes the worst pain in the butt.
When I decided to make this blog, I looked into setting up a virtual private server or a cloud service to host it. However, these services seemed too complicated for what I wanted to do and didn’t seem to be worth the time and money.