This next article is pretty technical in nature but I found it interesting in how networks or groupings of objects are discovered. Historically, there have been two primary methods of mapping these networks: statistical inference (guessing the next connection statistically) and/or spectral analysis (grouping nodes focusing on the flow of information - finding hubs or hot spots). This works well in most cases of denser populations but for sparse networks, it doesn't present a reliable map. The method described in this articlemakes a small modification that makes for more accurate mapping. This has implications in a number of areas such as finding links between colleagues at work as well as mapping the food cycle and the spread of diseases in communities.
http://www.santafe.edu/news/item/pnas-moore-spectral-redemption/
No comments:
Post a Comment