June 6, 2005

And, now some modicum of relief...

It looks like Kiera and the baby will be okay. They might let her out on Friday, but today's ultrasound shows enough fluid for the baby's lungs to properly develop. I'm still spending almost all my time at the hospital, but it looks like it will all turn out. Thank you, everyone, for your thoughts, prayers and support.

Posted 2005/06/06 20:09 (Mon) | TrackBack

I'm glad to hear this... Let's hope the rest of the pregnacy doesn't give any problems anymore...

Posted by: Peter Dirix at June 7, 2005 11:08

Here's hoping.

Posted by: David Weman at June 7, 2005 18:03

Great news. I'll be thinking of you.

Posted by: John Emerson at June 8, 2005 2:56

I'll be thinking of you and Kiera, Scott. I haven't been following your blog lately, and today I rode the rollercoaster in a very short time. I'm still in shock, and I can only imagine what you must have gone through. I hope the good news keep coming.

Posted by: pedro at June 8, 2005 20:52

I'm very glad to hear things are looking better for the 2 (3?) of you!

I don't know the status of your PhD proposal; I notice you've taken it off line, and disabled comments. If you're still working on it, and if I remember your description correctly, the talk I attended today might interest you.

The talk was by Deepayan Chakrabarti of CMU (http://www-2.cs.cmu.edu/~deepay/) about his work on data mining in large graphs. He uses a minimum-description-length approach to graph clustering.

One of his examples (starting at slide 22 in http://www-2.cs.cmu.edu/~deepay/mywww/presentations/CrossAssociations-2004.ppt) sounded potentially relevant to some of the things you've written about: he ran his algorithm on a bipartite graph of documents and words, where each document is connected to the words that occur in it. His sample data was a combination of 3 corpi, one of medical journal articles, one of CS information retrieval?) articles, and one of aerodynamics articles. The clustering correctly classified most of the articles by subject, but the interesting part was the clustering of words. Basically, each cluster of words had a distinct distribution among the clusters of documents; some clusters of words appeared mostly in one cluster of documents (medical terms, aerodynamics terms, etc.), and other clusters appeared fairly uniformly among all the documents. One humorous point was that the word "assumptions" fell into the cluster of words appearing mostly in aerodynamics documents!

Posted by: Jeremy Leader at June 10, 2005 21:10
Post a comment

Remember personal info?