Parallel, self-organizing, hierarchical neural networks. II

For pt.I see IEEE Trans. Neural Networks, vol.1, p.167-78 (1990). Parallel, self-organizing, hierarchical neural networks (PSHNNs) involve a number of stages with error detection at the end of each stage, i.e., rejection of error-causing vectors, which are then fed into the next stage after a nonlinear transformation. The stages operate in parallel during testing. Statistical properties and the mechanisms of vector rejection of the PSHNN are discussed in comparison to the maximum likelihood method and the backpropagation network. The PSHNN is highly fault tolerant and robust against errors in the weight values due to the adjustment of the error detection bounds to compensate errors in the weight values. These properties are exploited to develop architectures for programmable implementations in which the programmable parts are reduced to on-off or bipolar switching operations for bulk computations and attenuators for pointwise operations     

The greenhouse effect; Impact upon and the role to be played by construction

The authors highlight several possible ways in which the construction industry could lower present emissions of the greenhouse gas, carbon dioxide. It is an excellent discussion document and is recommended reading.     

Interfacing systems LOCA (ISLOCA) component pressure capacity methodology and typical plant results

A proposed methodology and sample results from several plant investigations are presented for evaluating the expected pressure capacity of nuclear power plant components which could potentially be subjected to Interfacing Systems LOCA (ISLOCA) conditions. Included in this evaluation are tanks, heat exchangers, filters, pumps, valves and flanged connections for systems designed for relatively low-pressure, low temperature conditions, but which could possibly be exposed to reactor coolant conditions, either through human error or isolation valve failures. The probabilities of failure, as a function of internal pressure, are evaluated as well as the variabilities associated with them. Leak rates or leak areas are estimated for the controlling modes of failure. Pressure capacities for the pipes and vessels are evaluated using limit-state analyses for the various failure modes considered. The capacities are dependent on several factors, including the material properties, modeling assumptions, and the postulated failure criteria. The failure modes for gasketed-flange connections, valves and pumps do not lend themselves to evaluation by conventional structural mechanics techniques and evaluation must rely primarily on the results from ongoing gasket research test programs and available vendor information and test data.     

Synthesis and characterization of poly(ethylene aspartate)-metal complexes

Poly(ethylene aspartate) [PEA] was synthesized by the melt condensation of D,L-aspartic acid and ethylene glycol. PEA containing pendent amino and carbonyl groups in its repeating chain was used as the polymeric ligand for complexation with transition metal ions, viz. Co(II), Ni(II), Cu(II), Mn(II), Zn(II), Cd(II), Ca(II), Mg(II), Pb(II) and Hg(II). Complexation was found to be most effective in DMSO. The resulting polyester-metal complexes were solid coloured materials which have been characterized by IR spectroscopy, elemental analysis and magnetic susceptibility measurements. The thermal stability of the polyester-metal complexes was investigated by thermogravimetric analysis (TGA). On the basis of the physico-chemical studies, an oxygen and nitrogen coordinated structure for the polyester-metal complexes is proposed.     

Mathematical analysis of dynamic channel selection in indoor mobilewireless communication systems

Dynamic channel selection (DCS) is an algorithm for flexible resource sharing in mobile radio systems. The digital enhanced cordless telecommunications (DECT) standard implements a version of DCS based on time as well as frequency multiplexing. In this paper, mathematical models are developed to evaluate the probabilities of channel availability, desired carrier power, and the carrier-to-interference ratio (CIR) for a constant traffic load. These results can be used to compute the call setup blocking probability. The models reported in this study are based on a decentralized DCS according to the DECT standard specifications. The results show that blocking due to availability of a channel (resource blocking), not interference blocking, generally is the most important factor in overall call blocking. Furthermore, it is seen that resource blocking is sensitive to the implementation of portable sets. Also, with homogeneous static traffic, this type of DCS ran be as good as fixed channel allocation (FCA) with respect to resource blocking     

Diarrhea in young children associated with Escherichia coli non-O157 organisms that produce Shiga-like toxin

The aim of this study was to evaluate the way in which short-term protection declines and is eventually lost in preconditioning and to determine the efficacy of a second preconditioning at various reperfusion intervals. Male rabbits were divided into six groups. Forty-five minutes (sustained) ischemia followed by 120 minutes reperfusion was applied 60, 65, 70, 75, and 80 minutes after a 5 minute preconditioning (groups A, B, C, D, and E) and in a control group (F) after no preconditioning. The infarct to risk ratio (I/R) was 38.3 +/- 3.5% in group A, 46.0 +/- 7.8% in B, 61.6 +/- 9.7% in C, 68.1 +/- 4.2% in D, 64.5 +/- 7.8% in E, and 61.0 +/- 7.7% in F. Group A had a smaller I/R compared with groups C, D, E, and F (p < 0.05). In another series, groups G, H, and I were exposed to two 5-minute preconditioning stimuli, separated, respectively, by 45, 60, and 75 minutes of reperfusion; 10 minutes after the last preconditioning, the animals were exposed to 45-minutes ischemia and 120 minutes reperfusion. Groups A and D (with the smaller and higher I/R ratio) were also incorporated into this protocol in order to compare the effect of the additional preconditioning with the single one. The I/R ratio was 25.4 +/- 8.5% in group G, 22.8 +/- 7.0% in group H, and 14.7 +/- 4.0% in group I (p = NS). Group D showed a higher I/R compared with groups G, A, and H (p < 0.01), and group I had a smaller I/R compared with groups A (p < 0.01) and D (p < 0.001). Cardioprotection after a first preconditioning declines gradually and is eventually lost. An additional preconditioning is always effective, and the longer the interval from the first preconditioning, the more potent is the effect.