6:45 – 3:00 ASRC IRAD
- A Physicist Who Models ISIS and the Alt-Right
- Working on adding the labels to LabeledTensor. Making an inner class that handles all the unique pieces of the dimension. It will also handle the checks, etc. Mostly done, but I need to make sure it handles axis of different size.
- Well drat: “Our operations teams are currently working to mitigate a Distributed Denial of Service (DDoS) attack on our networks. ” I can’t get to SVN. Yay! Got it!
- It just struck me that if I make this work with MapReduce, I can have Geomesa in something like 500 lines of code. I think I’ll call it GeoPlain
- Got different length axis working
- Important thing to think about with respect to maps: I find word clouds strange b/c they arrange summaries of text data in a way we would never handle summaries of numeric data.
- Data-driven Advice for Applying Machine Learning to Bioinformatics Problems
- 10:00 – 12:00 Song Chen’s Fraud Detection in Healthcare dissertation defense.
- Fiedler Vector
- What about spikes, such as a particularly bad flu or natural disaster? Is there a sense of normal at a particular time?
- bias in synthetic data?
- Community detection
- There are some good public datasets for testing healthcare. Heritage Foundation Prize dataset is used most in this work
- cosine similarity by procedure codes
- Is community pruning a generally effective optimization technique?
- What if community membership changes over time? Fraudulent doctors who move to a different state, for example?
- Parallel Heuristics for Scalable Community Detection