4 ms·

I want to know what Daphne Koller and her machine learning (probabilistic modeling?) team are doing. I want to know this badly.

• sjg007 9y ago
  In jest, you should just build a graphical model of likely states and then feed in some evidence to get a conclusion. Couple ideas: mapping regulatory networks behind some phenotype or biological pathways. Building sensor networks based on some biological inputs. Encoding biological data into some kind of probabilistic database. Simulations. Applied neural networks of some kind. In anti-aging research you would breed worms, mice, and fruit flies and look for anti-aging biomarkers in some set of conditions thought to promote life extension. You would also try to map known pathways between organisms to each other. You would sequence a ton of RNA... Also you would sequence genomes and RNA of people who are old and have lived a long time etc... As a control sequence everyone else. Feed all of that into a NN and see what pops out.