… Notation of the form p R + (1 -p)S, where R and S are distributions, represents the distribution
that is the mixture of R and S, with proportions p and 1 -p, respectively. Equivalent models …

Background: It is unclear whether more timely cancer diagnosis brings favourable outcomes,
with much of the previous evidence, in some cancers, being equivocal. We set out to …

The nature of cancer control is changing, with an increasing emphasis, fuelled by public and
political demand, on prevention, early diagnosis, and patient experience during and after …

… at least R ≡ K/N, where K = N − M. We report results on the assumption that the rate is R. The
… If one communicates using a code of rate R then it is conventional to describe the signal to …

… in some region R of (q, p) space, with volume V, the image of R under Ts will also have …
I will use my work on Bayesian neural network models (Neal, 1996a) as an example. Such …

… I have applied mixture models with infinite numbers of components to small data sets (Neal
1992a); the infinite model can in this case be implemented with finite resources. Nevertheless…

The EM algorithm performs maximum likelihood estimation for data in which some variables
are unobserved. We present a function that resembles negative free energy and show that …

The state of the art in data compression is arithmetic coding, not the better-known Huffman
method. Arithmetic coding gives greater compression, is faster for adaptive models, and …

… As Neal mentions, the computational efficiency of our “black-box” sampler RDIP depends
on the normalization constant. Our goal was to be automatic and reasonably efficient, rather …

… to maximum likelihood learning of belief networks (Neal, 2:1992b). … to some of the
same problems (Neal, 2:1992a, 2:1992c, 2:1993a). … I discuss this point further in (Neal, 2:1993b). …