Counting Koalas from space: Integrating remote sensing, GIS and traditional ecological survey techniques

Proceedings of the 11th Australasian Remote Sensing and Photogrammetry Conference

Gold nanoparticles decorated with oligo(ethylene glycol) thiols:enhanced Hofmeister effects in colloid−protein mixtures

Oligo(ethylene glycol) (OEG) thiol self-assembled monolayer (SAM) decorated gold nanoparticles (AuNPs) have potential applications in bionanotechnology due to their unique property of preventing the nonspecific absorption of protein on the colloidal surface. For colloid-protein mixtures, a previous study (Zhang et al. J. Phys. Chem. A 2007, 111, 12229) has shown that the OEG SAM-coated AuNPs become unstable upon addition of proteins (BSA) above a critical concentration, c*. This has been explained as a depletion effect in the two-component system. Adding salt (NaCl) can reduce the value of c*; that is, reduce the stability of the mixture. In the present work, we study the influence of the nature of the added salt on the stability of this two-component colloid-protein system. It is shown that the addition of various salts does not change the stability of either protein or colloid in solution in the experimental conditions of this work, except that sodium sulfate can destabilize the colloidal solutions. In the binary mixtures, however, the stability of colloid-protein mixtures shows significant dependence on the nature of the salt: chaotropic salts (NaSCN, NaClO4, NaNO3, MgCl2) stabilize the system with increasing salt concentration, while kosmotropic salts (NaCl, Na2SO4, NH4Cl) lead to the aggregation of colloids with increasing salt concentration. These observations indicate that the Hofmeister effect can be enhanced in two-component systems; that is, the modification of the colloidal interface by ions changes significantly the effective depletive interaction via proteins. Real time SAXS measurements confirm in all cases that the aggregates are in an amorphous state.

Finite-size effects in on-line learning of multilayer neural networks

We complement recent advances in thermodynamic limit analyses of mean on-line gradient descent learning dynamics in multi-layer networks by calculating fluctuations possessed by finite dimensional systems. Fluctuations from the mean dynamics are largest at the onset of specialisation as student hidden unit weight vectors begin to imitate specific teacher vectors, increasing with the degree of symmetry of the initial conditions. In light of this, we include a term to stimulate asymmetry in the learning process, which typically also leads to a significant decrease in training time.