Source-to-Source Instrumentation for the Optimization of an Automatic Reading System

Recording address traces that occur during a program execution is a significant technique for computer performance analysis. This paper describes a software method for address tracing via the instrumentation of C based languages. All program transformations are performed at the language level. This approach, which differs from the usual methods, allows portable and flexible program instrumentation. This tool has been developed to make easier the memory optimization of LIREChèques, an automatic bank check reading system. Two applications of the tool are clearly identified: (i) data cache use optimization, (ii) dynamic memory use optimization.     

Detection of chemicals by a reporter immunoassay: application to fluoride

This report describes a concept in which an immunoassay is used indirectly to quantify a nonantigenic very low molecular weight compound participating in a chemical reaction with a haptenic reporter. The detection limit of each reagent is, therefore, governed only by the affinity of the antibodies toward the reporter. Fluoride was used as a model, and silylated estradiol was used as a reporter. Upon silylation with N-O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) or N-O-bis(dimethylterbutylsilyl) trifluoroacetamide (MTBSTFA), estradiol is no longer recognized by antibodies specific to estradiol. After reaction with hydrofluoric acid (HF) or fluoride salts (KF, CsF, NaF), its immunoreactivity is restored, and native estradiol is formed and is detected by immunoassay. The level of synthesized estradiol is dependent on the concentration of fluoride. A fluoride detection limit of 0.3 microg/L (15 nM) is obtained. Potential interference with other acids has been eliminated by choosing the silyl group (trimethylsilyl vs tert-butyldimethylsilyl) and by selecting optimal reaction conditions for the desilylation. The method has been applied to the detection of fluoride salts in natural waters (range 0.28-9.0 mg/L) and in an atmosphere artificially contaminated with HF between 8 and 160 microg/m(3) in the parts-per-billion range. This indirect immunoassay combines simplicity and high sensitivity and, therefore, can be used in field monitoring. Finally, the extension of the concept to other chemicals is discussed.     

A single amino acid substitution in the antimicrobial defense protein cecropin B is associated with diminished degradation by leaf intercellular fluid

Degradation is one of several factors that may affect the level of accumulation of transgene products in plants. In plants engineered to secrete antimicrobial proteins to the intercellular compartment of leaves, the degenerative activity of proteases residing in leaf intercellular fluid (IF) could be critical to achieving the expected transgene function. We synthesized a structural analogue (MB39) of the antibacterial protein cecropin B and compared the susceptibility of both proteins to degradation in vitro by IF extracted from leaves of various crops. The half-life of the two proteins in the various IF extracts ranged from 3 min to 25.5 h, with the analogue MB39 displaying the longer half-life in IF from nine of 10 species. Overall, the half-life of MB39 averaged 2.9 times greater than that of cecropin B. Analysis of the peptides produced by endopeptidase activity in potato iF indicated that the 5.7-fold lower degradation rate of MB39 was associated with the substitution of valine for methionine at residue 11 of cecropin B. These findings point to the possibility of tailoring antimicrobial protein genes to reduce the rate of protein degradation in a particular target crop.     

One class random forests

One class classification is a binary classification task for which only one class of samples is available for learning. In some preliminary works, we have proposed One Class Random Forests (OCRF), a method based on a random forest algorithm and an original outlier generation procedure that makes use of classifier ensemble randomization principles. In this paper, we propose an extensive study of the behavior of OCRF, that includes experiments on various UCI public datasets and comparison to reference one class namely, Gaussian density models, Parzen estimators, Gaussian mixture models and One Class SVMs—with statistical significance. Our aim is to show that the randomization principles embedded in a random forest algorithm make the outlier generation process more efficient, and allow in particular to break the curse of dimensionality. One Class Random Forests are shown to perform well in comparison to other methods, and in particular to maintain stable performance in higher dimension, while the other algorithms may fail.     

A categorization system for handwritten documents

This paper presents a complete system able to categorize handwritten documents, i.e. to classify documents according to their topic. The categorization approach is based on the detection of some discriminative keywords prior to the use of the well-known tf-idf representation for document categorization. Two keyword extraction strategies are explored. The first one proceeds to the recognition of the whole document. However, the performance of this strategy strongly decreases when the lexicon size increases. The second strategy only extracts the discriminative keywords in the handwritten documents. This information extraction strategy relies on the integration of a rejection model (or anti-lexicon model) in the recognition system. Experiments have been carried out on an unconstrained handwritten document database coming from an industrial application concerning the processing of incoming mails. Results show that the discriminative keyword extraction system leads to better recall/precision tradeoffs than the full recognition strategy. The keyword extraction strategy also outperforms the full recognition strategy for the categorization task.