Analysis of MTF degradation in the imaging of cells in a flow system

The time-delay and integration method can be used with a charge-coupled device array to acquire images of cell in a flow cytometer. The loss of high spatial frequency components in images acquired with this technique is studied by deriving the modulation transfer function (MTF) for each of the main causes of image degradation: finite sampling aperture, nonsynchronism between image movement and charge transfer rate, discrete charge motion, transfer inefficiency, and axis misalignment. The system MTF and resolution in the horizontal and vertical directions are then examined.©1994 John Wiley & Sons Inc     

Determination of material parameters from regions close to the collector using electron beam-induced current

The conventional method of extracting the minority carrier diffusion length using the electron beam-induced current (EBIC) technique requires that the electron beam be placed at region more than two diffusion lengths away from the collector. The EBIC signals obtained under this condition usually has low signal to noise ratio. In addition, the true diffusion length of the sample is initially unknown and hence it is difficult to estimate how close the beam can be placed from the collector. To overcome all these difficulties, a new method of extracting minority carrier diffusion length from the EBIC signal is proposed. It is shown that this method can be applied to EBIC signals obtained from regions close to the collector. It is also shown that the surface recombination velocity of the sample can also be obtained using this method. This theory is verified using EBIC data generated from a device simulation software.     

Closed form solution for the sensor registration problem using only position information

Sensors, mounted on the dexterous end of a robot, can be used for feedback control or calibration. When you mount a sensor on a robot it becomes necessary to find the pose (orientation and position) of the sensor relative to the robot. This is the sensor registration problem. Many researchers have provided closed-form solutions to the sensor registration problem; however, the published solutions apply only to sensors that can measure a complete pose (three positions and three orientations). Many sensors, however, can provide only position information; they cannot measure the orientation of an object. This article provides a closed-form solution to the sensor registration problem applicable when: (1) the sensor can provide only position information and (2) the robot can move along and rotate about straight lines. © 1994 John Wiley & Sons, Inc.     

Fractionation studies of palm oil by density gradient

The paper describes a method of fractionating vegetable, animal and fish oils, and in particular palm oil. The method involves addition of a medium comprising two common solvents to the semisolid oils. On centrifugation, the olein and stearin are separated by the medium in the middle. Thirteen media made up from binary combinations of nine solvents, viz. water, propylene glycol, glycerine, methanol, ethanol,n-propanol, isopropanol (IPA), acetone and butanone, are found to be effective in olein-stearin separation. However, only the water/IPA and water/methanol systems have been studied in detail. The aqueous IPA provides a higher yield of olein than water/ methanol but intersolubility between oil and medium is also greater. The fractionation process can be carried out at any suitable temperature. Fractionation of the special prime bleached (SPB) palm oil at 16 C yields an olein with a cloud point of 4.8 C. Some hybrid palm oils produce a large quantity of low cloud point olein which can be bleached readily. The process can be extended to include degumming and neutralization by using an alkaline medium for centrifugation. The olein fractions obtained have been found to be free of phosphatides and the free fatty acids reduced to as low as 0.02%. Metal-scavenging agents have also been added to the medium in an attempt to remove copper and iron. The development of this process into a continuous one has been demonstrated on the AlfaLaval LAPX 202 Separator. Fractionation of crude palm oil using a density gradient provides seven fractions of different characteristics. The iodine values vary from 37.5 to 57.4 and the unsaturated fatty acids range from 32.7% to 51.2%. Triglyceride analysis by carbon numbers shows great differences in the C₄₈ and C₅₂ constituents of the fractions. aThe volume ratio of oil to medium in each case was 1:1. The separation involved the oil and wax.     

Cellulose-binding domains: potential for purification of complex proteins

The endoglucanase CenA and the exoglucanase Cex from Cellulomonasfimi each contain a discrete cellulose-binding domain (CBD),at the amino-terminus or carboxyl-terminus respectively. Thegene fragment encoding the CBD can be fused to the gene of aprotein of interest. Using this approach hybrid proteins canbe engineered which bind reversibly to cellulose and exhibitthe biological activity of the protein partner. Alkaline phosphatase(PhoA) from Escherichia coli, and a ß-glucosidase(Abg) from an Agrobacterium sp. are dimeric proteins. The fusionpolypeptides CenA-PhoA and Abg-CBC_cex are sensitive to proteolysisat the junctions between the fusion partners. Proteolysis resultsin a mixture of homo- and heterodimers; these bind to celluloseif one or both of the monomers carry a CBD, e.g. CenA-PhoA/CenA-PhoAand CenA-PhoA/PhoA. CBD fusion polypeptides could be used inthis way to purify polypeptides which associate with the fusionpartner.     

Analytic center cutting plane method for multiple kernel learning

Multiple Kernel Learning (MKL) is a popular generalization of kernel methods which allows the practitioner to optimize over convex combinations of kernels. We observe that many recent MKL solutions can be cast in the framework of oracle based optimization, and show that they vary in terms of query point generation. The popularity of such methods is because the oracle can fortuitously be implemented as a support vector machine. Motivated by the success of centering approaches in interior point methods, we propose a new approach to optimize the MKL objective based on the analytic center cutting plane method (accpm). Our experimental results show that accpm outperforms state of the art in terms of rate of convergence and robustness. Further analysis sheds some light as to why MKL may not always improve classification accuracy over naive solutions.     

Feature selection via sensitivity analysis of SVM probabilistic outputs

Feature selection is an important aspect of solving data-mining and machine-learning problems. This paper proposes a feature-selection method for the Support Vector Machine (SVM) learning. Like most feature-selection methods, the proposed method ranks all features in decreasing order of importance so that more relevant features can be identified. It uses a novel criterion based on the probabilistic outputs of SVM. This criterion, termed Feature-based Sensitivity of Posterior Probabilities (FSPP), evaluates the importance of a specific feature by computing the aggregate value, over the feature space, of the absolute difference of the probabilistic outputs of SVM with and without the feature. The exact form of this criterion is not easily computable and approximation is needed. Four approximations, FSPP1-FSPP4, are proposed for this purpose. The first two approximations evaluate the criterion by randomly permuting the values of the feature among samples of the training data. They differ in their choices of the mapping function from standard SVM output to its probabilistic output: FSPP1 uses a simple threshold function while FSPP2 uses a sigmoid function. The second two directly approximate the criterion but differ in the smoothness assumptions of criterion with respect to the features. The performance of these approximations, used in an overall feature-selection scheme, is then evaluated on various artificial problems and real-world problems, including datasets from the recent Neural Information Processing Systems (NIPS) feature selection competition. FSPP1-3 show good performance consistently with FSPP2 being the best overall by a slight margin. The performance of FSPP2 is competitive with some of the best performing feature-selection methods in the literature on the datasets that we have tested. Its associated computations are modest and hence it is suitable as a feature-selection method for SVM applications. Editor: Risto Miikkulainen.     

Experimental dynamic characterizations and modelling of disk vibrations for HDDs

Currently, the rotational speed of spindle motors in HDDs (Hard-Disk Drives) are increasing to improve high data throughput and decrease rotational latency for ultra-high data transfer rates. However, the disk platters are excited to vibrate at their natural frequencies due to higher air-flow excitation as well as eccentricities and imbalances in the disk-spindle assembly. These factors contribute directly to TMR (Track Mis-Registration) which limits achievable high recording density essential for future mobile HDDs. In this paper, the natural mode shapes of an annular disk mounted on a spindle motor used in current HDDs are characterized using FEM (Finite Element Methods) analysis and verified with SLDV (Scanning Laser Doppler Vibrometer) measurements. The identified vibration frequencies and amplitudes of the disk ODS (Operating Deflection Shapes) at corresponding disk mode shapes are modelled as repeatable disturbance components for servo compensation in HDDs. Our experimental results show that the SLDV measurements are accurate in capturing static disk mode shapes without the need for intricate air-flow aero-elastic models, and the proposed disk ODS vibration model correlates well with experimental measurements from a LDV.     

Distributed model predictive control of dynamically decoupled systems with coupled cost

This paper considers the distributed model predictive control (DMPC) of systems with interacting subsystems having decoupled dynamics and constraints but coupled costs. An easily-verifiable constraint is introduced to ensure asymptotic stability of the overall system in the absence of disturbance. The constraint introduced has a parameter which allows for the performance of the DMPC system to approach that controlled by a centralized model predictive controller. When the subsystems are linear and additive disturbance is present, the added constraint ensures the state of each subsystem converges to its respective minimal disturbance invariant set. The approach is demonstrated via several numerical examples.