Using simulations of reduced precision arithmetic to design a neuro-microprocessor

This article describes some of our recent work in the development of computer architectures for efficient execution of artificial neural network algorithms. Our earlier system, the Ring Array Processor (RAP), was a multiprocessor based on commercial DSPs with a low-latency ring interconnection scheme. We have used the RAP to simulate variable precision arithmetic to guide us in the design of arithmetic units for high performance neurocomputers to be implemented with custom VLSI. The RAP system played a critical role in this study, enabling us to experiment with much larger networks than would otherwise be possible. Our study shows that back-propagation training algorithms only require moderate precision. Specifically, 16b weight values and 8b output values are sufficient to achieve training and classification results comparable to 32b floating point. Although these results were gathered for frame classification in continuous speech, we expect that they will extend to many other connectionist calculations. We have used these results as part of the design of a programmable single chip microprocessor, SPERT. The reduced precision arithmetic permits the use of multiple arithmetic units per processor. Also, reduced precision operands make more efficient use of valuable processor-memory bandwidth. For our moderate-precision fixed-point arithmetic applications, SPERT represents more than an order of magnitude reduction in cost over systems with equivalent performance that use commercial DSP chips.     

Optoelectronic charge coupled device (OECCD) using InP MIScapacitor array

An optoelectronic charge coupled device (OECCD) which is directly compatible with an incident optical signal is proposed. The nonlinear partial differential equation consisting of optical generation and recombination is solved in one dimension using the Crank-Nicolson finite-difference scheme. The charge transfer inefficiency is the main parameter considered for the calculation     

Gain measurements in 1.3 µm InGaAsP-InP double heterostructure lasers

The net gain per unit length (G) versus current (I) is measured at various temperatures for 1.3 μm InGaAsP-InP double heterostructure lasers.Gis found to vary linearly with the currentIat a given temperature. The gain bandwidth is found to decrease with decreasing temperature. The lasing photon energy decreases at 0.325 meV/K with increasing temperature. Also, the slopedG/dIat the lasing photon energies decreases with increasing temperature. This decrease is more rapid forT > sim210K. This faster decrease is consistent with the observed higher temperature dependence of threshold (low T0at high temperatures) of 1.3 μm InGaAsP lasers. A carrier loss mechanism, due to Auger recombination, also predicts thatdG/dIshould decrease much faster with increasing temperature at high temperatures. We also find that the slopedG/dIdecreases slowly with increasing temperature for a GaAs laser, which is consistent with the observed temperature dependence of threshold of these lasers.     

A Microcomputer-Based Differential Lung Ventilation System

In order to provide a versatile means of delivering differential lung ventilation (DLV), a computer-controlled system was constructed to allow a variety of ventilation protocols as well as to record and monitor relevant physiologic parameters. Two Siemens servo ventilators were modified for synchronous operation and computer control of minute volume and respiratory rate. Twenty-five parameters on the two lungs were collected every breath. Feedback control was used to adjust respiratory rate to maintain PaCO2 = 35 torr and to keep total tidal volume equal to 15 mI/kg. Three differential volume delivery protocols were established. The DLV system was evaluated in a study involving eighteen mongrel dogs (six dogs for each volume delivery protocol), each with a unilateral lung injury caused by an infusion of 0.1 N HOC through the endobronchial tube. This system has proven to be a highly effective and versatile means of providing differential ventilation as well as precise feedback control of essential physiologic parameters such as PaCO2 and tidal volume. The system handles automated data collection of all relevant physiological parameters, making graphical as well as statistical analysis extremely easy.     

Use of Remotely Sensed Data for Assessing Forest Stand Conditions in the Eastern United States

The results of three interrelated research activities conducted by Goddard scientists in support of the AgRISTARS Renewable Resources Inventory (RRI) project are summarized. The central theme of the research conducted at Goddard was the development of techniques for the detection, classification, and measurement of forest disturbances using digital, remotely sensed data. Three study areas located in Pennsylvania, North Carolina, and Maine were investigated with respect to: a) the delineation and assessment of forest damage associated with two different forest insect defoliators, and b) an assessment of the improved capabilities to be expected from Landsat Thematic Mapper (TM) data relative to Multispectral Scanner (MSS) data for delineating forest stand characteristics. Key results include the development of a statewide MSS digital data base and associated image processing techniques for accurately delineating (approximately 90 precent correct classification accuracy) insect damaged and healthy forest. Comparison of analyses using MSS and TM Simulator (TMS) data indicated that for broad land cover classes which are spectrally homogeneous, the accuracy of the classification results are similar. However, TMS data provided superior results (20 percent overall accuracy increase relative to MSS results) when detailed (Level III) forest classes were mapped. These studies also illustrated the utility of having at least one band in the visible, near infrared, and middle infrared portion of the electromagnetic spectrum for assessing specific (Level III) forest cover types.     

Effect of simulated freeze damage on soybean seed composition and functionality

Two soybean varieties (Beeson and Williams) were harvested at various maturation levels by applying paraquat directly to the growing plant. The application of this herbicide to arrest growth was used to simulate freeze damage. The free fatty content and the photometric index declined from 0.61% to 0.19% and 301.56 to 201.90, respectively, as maturation progressed for the Beeson variety. Similar trends were found with the Williams variety. However, the amount of trypsin inhibitor differed between the two varieties depending upon maturation level.     

Cray/Blitz—1984 chess champion

The chess program “Cray Blitz” is the current World Computer Chess Champion and North American Computer Chess Champion (ACM). The program has also played in human chess tournaments and is a chess master. At speed chess, where its ability to perform very accurate analysis is particularly important, it has maintained a performance rating of over 2600 for the past two years. This indicates that at speed chess, the program is one of the top players, electronic or human, in the world. It is currently running on a Cray XMP-48 computer system and has been designed around the parallelism that the XMP architecture provides.     

GaInAs PIN photodiodes grown by atmospheric-pressure MOVPE

GaInAs/InP PIN photodiodes with low dark currents and capacitances have been successfully fabricated from material grown by atmospheric-pressure MOVPE. Both Zn-diffused and grown-in p+-n homojunction material have provided yields of over 70% for devices with leakage currents less than 20 nA. This growth technique, therefore, looks particularly appropriate for a reproducible, high-yield and inexpensive method of photodetector production.     

Algorithm for multichannel LMS adaptive filtering

A modification of the LMS algorithm is presented in which the coefficients of an array of FIR filters, whose outputs are linearly coupled to an array of sensors, are adapted to minimise the mean-square sum of the error signals from these sensors. The application of the algorithm to the control of distributed parameter systems is discussed.     

Contested landscapes: territorial conflicts and the production of different ruralities in Brazil

ABSTRACT

From the perspective of Brazilian agrarian geography, the conflicts generated by land tenure disputes have as protagonists the families of the Landless Workers’ Movement (Movimento dos Trabalhadores Rurais Sem Terra), traditional landowners, and agribusiness corporations. In this article, we present a reflection on land conflicts and the spatial and territorial production of latifundia, peasant smallholders, and large-scale agribusiness through studies of the landscapes of two municipalities in the state of São Paulo. Among the various tools for analysing territorial disputes, Google Earth can be used for local, regional, and international comparative studies of how landscapes have been transformed. Informed by traditional research, the article analyses a diversity of photographs, from satellite to field shots, as evidence of landscapes that express conflicting social relations in disputes over different models of social and territorial development.