Oct 29, 2009, 9:05:32 PM10/29/09
Never one to refuse a soapbox / platform / pulpit, I regret that I
might not be able to attend. My daughter needs a ride to the airport.
Until Wednesday night, I thought she needed the ride Thursday morning,
but once again Dad was wrong: it's Friday. It's early, so I'll try to
get back for book club. But no promises.
I'm nearly finished with Martyn Amos' _Genesys Machines_, a book that
deserves much wider readership. When we last met, I had just read
about Adelman's seminal work on using DNA to solve a Hamiltonian path
problem. I was enthusiastic about molecular computing's potential for
tackling NP-complete problems. Since then the book has covered a
historical turn. Genetic machines run into the same issues as silicon-
based machines when face with combinatorial explosion. The remainder
of the book explains where molecular computing has gone since the
early days and where it's going today.
On the machine learning front, I coded naive Bayes in Ruby, following
Segaran's examples in Python. I've been looking for a chance to rehone
my Ruby skills; this chapter was a good opportunity, because unlike
many of the others it did not rely on external libraries. To that
extent, it was a success. However, the focus on language may have been
too great a distraction; I still don't have good insights to the
calculations. ("The purpose of computing is insight, not numbers."
Hamming, I believe.) More work to do on that front...
See you tomorrow if I can.