For this meeting, Christian McDaniel and Andrew Durden from Shannon Quinn's group at UGA made the hike over to the ATL to talk with us about the projects they presented at SciPy 2018.

"Developing an LSTM Pipeline for Accelerometer Data." Christian McDaniel

Christian gave us the low-down on the pipeline he developed to benchmark recurrent neural networks that recognize different types of human activity from accelerometer data recorded from cell phones. Check out the SciPy proceedings paper and code here:
Also check out the presentation he shared with us for lots of helpful pointers to info, tutorials, etc. about machine learning, in particular about LSTMs and other types of recurrent neural networks.

"Dynamic Social Network Modeling of Diffuse Subcellular Morphologies." Andrew Durden

Andrew talked with us about how he's building tools to track changes in sub-cellular structures that have until now been hard to analyze because of their diffuse structure, such as mitochondria and actin. Their pipeline OrNET addresses this gap by leveraging the power of social network analysis in combination with other tools from image analysis and computer vision. You can read the Proceedings paper here:
And for some updates on how the work has progressed, see Andrew's presentation which he kindly shared.