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     

Far-infrared pump-probe measurement of the lifetime of the 2p−1 shallow donor level in n-GaAs

The low temperature lifetime of electrons excited in the 2p^?1 donor level of n-GaAs has been studied in a far-infrared pump-probe experiment. The measurement has been carried out using a pulsed far-infrared molecular gas laser working at a wavelength of 292µm, with the sample in a magnetic field of 5.1 T, resonant with the 1s_o?2p^?1 transition. Two FIR pulses are sliced from one FIR-laser pulse by means of optical switching techniques using two Q-switched Nd:YAG lasers. The first pulse is used to saturate the transition, while the second pulse probes the return of the population in the excited state towards thermal equilibrium as a function of the time delay after the excitation pulse. The value of 350±50 ns found for the lifetime falls in line with CW saturation results on materials with other doping concentrations.     

Evaluation of the quality of compacting water-saturated sands by sounding

Various methods of sounding were investigated for evaluating the quality of compacting water-saturated fine and medium sands. A comparative analysis of the data obtained with the data of direct soil sampling was carried out. The given recommendations can be used when preparing the new edition of the building codes for engineering surveys for construction. Translated from Osnovaniya, Fundamenty i Mekhanika Gruntov, No. 4, pp. 17–20, July–August, 1994     

Data management and analysis for high-throughput DNA sequencingprojects

The rapid advances in molecular biology have begun to shift many of the bottlenecks in genome research from the laboratory to the data analysis facility. The pace at which this has occurred creates a situation in which software development always has to catch up with the flow of data. Since such large-scale processes were not anticipated, the analysis infrastructure has not been fully established. Furthermore, most systems that have been built were designed by the biologists who collected the data. More recently, computer scientists, mathematicians, and engineers have taken an interest in this problem. This has had a positive effect, since it has created a tight synergy between the informatics and the biology. Several principles affected the design of the system developed at TIGR. Each of the sample preparation, sequencing, and analysis steps had to be managed, scheduled, and tracked. This information had to be made readily available to those who needed it for carrying out their tasks. Different skill levels of the users had to be taken into account. The degree of human intervention at each step had to be evaluated and built into the design. A mixed processing environment of Macintosh and Unix platforms had to be integrated. Most importantly, the system had to save time, reduce error, and ensure uniformity of the analysis and quality of the results. In the authors' experience, the tools they have built work well because of their early decisions as to which systems to use for development. The authors settled on a robust relational database management system (Sybase) and a portable development environment (C, C++)     

High speed p-type SiGe modulation-doped field-effect transistors

We report on the fabrication and characterization of high-speed p-type modulation-doped field-effect transistors (MODFETs) with 0.7-μm and 1-μm gate-lengths having unity current-gain cut-off frequencies (f_T) of 9.5 GHz and 5.3 GHz, respectively. The devices were fabricated on a high hole mobility SiGe heterostructure grown by ultra-high-vacuum chemical vapor deposition (UHV-CVD). The dc maximum extrinsic transconductance (g_m) is 105 mS/mm (205 mS/mm) at room temperature (77 K) for the 0.7-μm gate length devices. The fabricated devices show good pinch-off characteristics and have a very low gate leakage current of a few μA/mm at room temperature and a few nA/mm at 77 K     

Calculation of the acoustical characteristics of gas transport equipment of the unit type in the design stage

Translated from Khimicheskoe i Neftyanoe Mashinostroenie, No. 11, pp. 28–30, November, 1993.     

Cement versus cementless fixation in total knee arthroplasty

A prospectively studied group of 55 uncemented Press Fit Condylar total knee arthroplasties was compared retrospectively with a matched group of 51 cemented Press Fit Condylar total knee arthroplasties at a mean of 10 years after operation. For the cemented group, the pain and function scores improved from 32 and 45 preoperatively to 95 and 77, respectively. For the uncemented group the scores improved from 33 and 50 preoperatively to 93 and 60, respectively. There were 10 revisions in the uncemented group for femoral or tibial aseptic loosening or osteolysis compared with two revisions in the cemented group. Exclusive of problems related to patellar metal backing, survival to revision for aseptic failure or radiographic loosening was 72% in the uncemented group and 94% in the cemented group at 10 years. A significantly higher revision rate was found in the uncemented compared with cemented total knee arthroplasty of the Press Fit Condylar design.     

Image coding by multistep, adaptive flux interpolation

The authors describe and discuss the new technique, multistep adaptive flux interpolation (MAFI), and its application to image data for coding. When applied to an image, MAFI produces an output which is also in an image form, but which has a more uniform feature density and a greatly reduced size. MAFI warps the input image by removing those rows and columns which contain a majority of redundant pixels. The side information required for reconstruction is minimal, and the image can be further compressed using conventional coders, making the compression ratio even higher. Because of its warped nature, the MAFI output's statistics are also more consistent with the properties assumed by block-based discrete cosine transform (DCT) methods     

Manufacture of films displaying liquid-crystalline structure

Reported is an attempt to attain orientation of macromolecules during extrusion of films of cellulose derivatives that led to the liquid-crystalline state of a polymer and resulted in enhanced film strength.