Biomedical subjects

D J MacKay

Publications and source records attributed to D J MacKay.

3 recordsLinked to original sources

PubMed2001

Exact sampling from nonattractive distributions using summary states.

Propp and Wilson's method of coupling from the past allows one to efficiently generate exact samples from attractive statistical distributions (e.g., the ferromagnetic Ising model). This method may be generalized to nonattractive distributions by the use of summary states, as first described by Huber. Using this method, we present exact samples from a frustrated antiferromagnetic triangular Ising model and the antiferromagnetic q=3 Potts model. We discuss the advantages and limitations of the method of summary states for practical sampling, paying particular attention to the slowing down of the algorithm at low temperature. In particular, we show that such slowing down can occur in the absence of a physical phase transition.

Journal Article↗

PubMed1998

A recurrent neural network for modelling dynamical systems.

We introduce a recurrent network architecture for modelling a general class of dynamical systems. The network is intended for modelling real-world processes in which empirical measurements of the external and state variables are obtained at discrete time points. The model can learn from multiple temporal patterns, which may evolve on different timescales and be sampled at non-uniform time intervals. We demonstrate the application of the model to a synthetic problem in which target data are only provided at the final time step. Despite the sparseness of the training data, the network is able not only to make good predictions at the final time step for temporal processes unseen in training, but also to reproduce the sequence of the state variables at earlier times. Moreover, we show how the network can infer the existence and role of state variables for which no target information is provided. The ability of the model to cope with sparse data is likely to be useful in a number of applications, including, in particular, the modelling of metal forging.

Models, Theoretical↗

PubMed1986

Growth hormone release in man induced by galanin, a new hypothalamic peptide.

Galanin, a 29-aminoacid neuropeptide, was infused for 60 min into healthy volunteers at 7.8 pmol/kg/min (n = 4) or 33.2 pmol/kg/min (n = 6). During the infusion there was no change in heart rate or blood pressure and the only symptoms were a transitory bitter taste and slight hypersalivation. Plasma growth hormone levels rose during the high-dose galanin infusion from 2.8 +/- 0.8 mU/l to a mean peak of 48.5 +/- 19.8 mU/l; prolactin levels rose from 176 +/- 33 mU/l to 274 +/- 33 mU/l. A significant rise in growth hormone also occurred with the low-dose infusion (2.5 +/- 1.1 mU/l to a mean peak of 23.5 +/- 6.6 mU/l). There was no change in cortisol, thyroid-stimulating hormone, follicle-stimulating hormone, or luteinising hormone at either dose. 20 min after the start of the infusion a 25 g glucose bolus was given intravenously. Galanin reduced glucose clearance without significantly affecting plasma insulin concentrations. Pancreatic polypeptide levels were suppressed by the galanin infusion but levels of glucagon and gastric inhibitory peptide were unchanged.

Adult↗