Having taken some courses from David and others at UCL recently, I'm a big fan of this.
The Bayesian modelling perspective I think is very useful if you're interested in machine learning as more than just a collection of clever algorithms and optimisation techniques to throw at a problem and see what sticks. (Not that this isn't useful sometimes...)
It provided a lot of motivation and unifying intuition for me anyway. The elegance of having a nice statistical model doesn't come for free though, there are some tricky computational issues associated with inference in many Bayesian models. The book covers them in some depth and seems quite a useful reference into the state of the art as well as a nice introduction to the area.
W/r/t computing gnarly integrals, interested parties might appreciate pymc, a python package that implements markov chain monte carlo methods to estimate them
Assuming this is legitimately released (seems to be), authors who write and release these books for free are heroes for those of us not currently undergrads at Stanford etc.
I'm sure, that because of TeX and stuff, it's popular among lots of people to publish their free (or non-free) e-books as PDFs. Still, because of small screen reading devices, it would be great if they could publish an epub also. Or source. Or anything convertible to epub/mobi.
This is one of the best resources for learning about Bayesian ML methods if you need a gentle introduction. The only other book I found which was similarly clear and well thought-out is Christopher Bishop's "Pattern Recognition and Machine Learning".
2) @reader5000 It's legit he links to it from his homepage http://web4.cs.ucl.ac.uk/staff/D.Barber/pmwiki/pmwiki.php?n=...