The role of biases in on-line learning of two-layer networks

The influence of biases on the learning dynamics of a two-layer neural network, a normalized soft-committee machine, is studied for on-line gradient descent learning. Within a statistical mechanics framework, numerical studies show that the inclusion of adjustable biases dramatically alters the learning dynamics found previously. The symmetric phase which has often been predominant in the original model all but disappears for a non-degenerate bias task. The extended model furthermore exhibits a much richer dynamical behavior, e.g. attractive suboptimal symmetric phases even for realizable cases and noiseless data.

Hebbian and homeostatic plasticity mechanisms in regular spiking and intrinsic bursting cells of cortical layer 5

Layer 5 contains the major projection neurons of the neocortex and is composed of two major cell types: regular spiking (RS) cells, which have cortico-cortical projections, and intrinsic bursting cells (IB), which have subcortical projections. Little is known about the plasticity processes and specifically the molecular mechanisms by which these two cell classes develop and maintain their unique integrative properties. In this study, we find that RS and IB cells show fundementally different experience-dependent plasticity processes and integrate Hebbian and homeostatic components of plasticity differently. Both RS and IB cells showed TNFα-dependent homeostatic plasticity in response to sensory deprivation, but IB cells were capable of a much faster synaptic depression and homeostatic rebound than RS cells. Only IB cells showed input-specific potentiation that depended on CaMKII autophosphorylation. Our findings demonstrate that plasticity mechanisms are not uniform within the neocortex, even within a cortical layer, but are specialized within subcircuits.

Estimation of functional connectivity from electromagnetic signals and the amount of empirical data required

An increasing number of neuroimaging studies are concerned with the identification of interactions or statistical dependencies between brain areas. Dependencies between the activities of different brain regions can be quantified with functional connectivity measures such as the cross-correlation coefficient. An important factor limiting the accuracy of such measures is the amount of empirical data available. For event-related protocols, the amount of data also affects the temporal resolution of the analysis. We use analytical expressions to calculate the amount of empirical data needed to establish whether a certain level of dependency is significant when the time series are autocorrelated, as is the case for biological signals. These analytical results are then contrasted with estimates from simulations based on real data recorded with magnetoencephalography during a resting-state paradigm and during the presentation of visual stimuli. Results indicate that, for broadband signals, 50-100 s of data is required to detect a true underlying cross-correlations coefficient of 0.05. This corresponds to a resolution of a few hundred milliseconds for typical event-related recordings. The required time window increases for narrow band signals as frequency decreases. For instance, approximately 3 times as much data is necessary for signals in the alpha band. Important implications can be derived for the design and interpretation of experiments to characterize weak interactions, which are potentially important for brain processing.

Bioactive glass sol-gel foam scaffolds:evolution of nanoporosity during processing and in situ monitoring of apatite layer formation using small- and wide-angle X-ray scattering

Recent work has highlighted the potential of sol-gel-derived calcium silicate glasses for the regeneration or replacement of damaged bone tissue. The work presented herein provides new insight into the processing of bioactive calcia-silica sol-gel foams, and the reaction mechanisms associated with them when immersed in vitro in a simulated body fluid (SBF). Small-angle X-ray scattering and wide-angle X-ray scattering (diffraction) have been used to study the stabilization of these foams via heat treatment, with analogous in situ time-resolved data being gathered for a foam immersed in SBF. During thermal processing, pore sizes have been identified in the range of 16.5-62.0 nm and are only present once foams have been heated to 400 degrees C and above. Calcium nitrate crystallites were present until foams were heated to 600 degrees C; the crystallite size varied from 75 to 145 nm and increased in size with heat treatment up to 300 degrees C, then decreased in size down to 95 rim at 400 degrees C. The in situ time-resolved data show that the average pore diameter decreases as a function of immersion time in SBF, as calcium phosphates grow on the glass surfaces. Over the same time, Bragg peaks indicative of tricalcium phosphate were evident after only 1-h immersion time, and later, hydroxycarbonate apatite was also seen. The hydroxycarbonate apatite appears to have preferred orientation in the (h,k,0) direction.