Connectivity and performance tradeoffs in the cascade correlationlearning architecture

The cascade correlation is a very flexible, efficient and fast algorithm for supervised learning. It incrementally builds the network by adding hidden units one at a time, until the desired input/output mapping is achieved. It connects all the previously installed units to the new unit being added. Consequently, each new unit in effect adds a new layer and the fan-in of the hidden and output units keeps on increasing as more units get added. The resulting structure could be hard to implement in VLSI, because the connections are irregular and the fan-in is unbounded. Moreover, the depth or the propagation delay through the resulting network is directly proportional to the number of units and can be excessive. We have modified the algorithm to generate networks with restricted fan-in and small depth (propagation delay) by controlling the connectivity. Our results reveal that there is a tradeoff between connectivity and other performance attributes like depth, total number of independent parameters, and learning time.     

Vasodilation with adenosine or sodium nitroprusside after coronary artery bypass surgery: a comparative study on myocardial blood flow and metabolism

The effects of adenosine and sodium nitroprusside (SNP) on central hemodynamics and myocardial blood flow and metabolism were investigated postoperatively after elective coronary artery bypass (CABG) surgery in ten sedated and mechanically ventilated patients in the intensive care unit. During three consecutive 15-min periods, SNP (0.8 +/- 0.1 micrograms.kg-1 x min-1), adenosine (88.9 +/- 13.3 micrograms.kg-1 x min-1), and then again SNP (0.7 +/- 0.1 micrograms.kg-1 x min-1) were infused to control postoperative hypertension at a mean arterial pressure of approximately 80 mm Hg. Systemic and pulmonary hemodynamics and global (coronary sinus flow, CSF) as well as regional (great cardiac vein flow, GCVF) myocardial blood flow and metabolic variables were measured. During adenosine infusion, in comparison to SNP, heart rate was unchanged, stroke volume index and cardiac index increased (24% and 32%, respectively), and the systemic vascular resistance index decreased (-26%). Mean pulmonary arterial pressure (24%) as well as pulmonary capillary wedge pressure (27%) and central venous pressure (18%) were higher with adenosine compared to SNP. Adenosine also increased CSF and GCVF (108% and 103%, respectively) without altering the CSF/GCVF flow ratio compared to SNP. Furthermore, adenosine increased the coronary oxygen content (51%) and decreased the arterio-great cardiac vein oxygen content difference (-48%) without changing regional myocardial oxygen consumption, indicating a more pronounced hyperkinetic myocardial circulation compared to SNP. In addition, adenosine infusion decreased arterial PO2 (-11%) and increased the intrapulmonary shunt fraction (57%). The PR interval time of the electrocardiogram was prolonged (12%) and the ST segment was more depressed during adenosine infusion compared to SNP.(ABSTRACT TRUNCATED AT 250 WORDS)     

A study of phase equilibria of nitrogen for constructing a medium value pressure scale

Translated from Izmeritel'naya Tekhnika, No. 3, pp. 40–42, March, 1993.     

Some features of the chemical transformation of a highly exothermic ore concentrate-based mixture in a combustion wave

Use of inexpensive ore concentrates as initial components in SHS metallurgy [1–3] brings about some specific features in the process and the properties of the materials obtained. Ore reagents contain a considerable amount of impurities (Al₂O₃, SiO₂, Fe₂O₃) that decrease the energetics of SHS systems and the temperature and the rate of combustion and create the limits of combustion and phase separation. In addition, the impurities participate in chemical reactions and enter the desired product. In many cases the presence of silicon, iron, and aluminum in a cast product does not prevent use of this product for practical purposes.Institute of Metallurgy, Academy of Sciences of the Republic of Georgia, Tbilisi. Institute of Structural Macrokinetics, Russian Academy of Sciences, Chernogolovka. Translated from Inzhenerno-Fizicheskii Zhurnal, Vol. 65, No. 5, pp. 613–616, November, 1993.     

Synthesis and analysis of transformer-isolated converters

A systematic and straightforward procedure is developed for the synthesis and analysis of transformer-isolated power converters. The procedure can be used to determine the ranges of duty-ratio over which the transformer-isolated power converters of a given class can be operated without transformer saturation. The procedure can also be used to study the dependence of the power converter switch stresses on duty-ratios. This information is useful in the selection of the transformer-isolated power converter most suitable for a given application and in the design of this power converter with minimum switch stresses, high power density, and low cost