Browsing by Author "Acuna, G"

Now showing 1 - 4 of 4
  • Thumbnail Image
    Item
    A neural network estimator for total biomass of filamentous fungi growing on two dimensional solid substrate
    (KLUWER ACADEMIC PUBL, 1998) Acuna, G; Giral, R; Agosin, E; Jorquera, H; Perez Correa, R; Ferret, E; Molin, P; Thibault, J
    A neural network dynamic model is proposed for the on-line estimation of total biomass during filamentous fungi cultures on two dimensional solid substrate. The neural network provides an accurate and robust estimation of biomass from macroscopic measurements of the colony radius evolution. Experiments were performed on Gibberella fujikuroi growing on Petri dishes under different conditions of temperature and water activity.
  • Thumbnail Image
    Item
    Forecasting ozone daily maximum levels at Santiago, Chile
    (PERGAMON-ELSEVIER SCIENCE LTD, 1998) Jorquera, H; Perez, R; Cipriano, A; Espejo, A; Letelier, MV; Acuna, G
    In major urban areas, air pollution impact on health is serious enough to include it in the group of meteorological variables that are forecast daily. This work focusses on the comparison of different forecasting systems for daily maximum ozone levels at Santiago, Chile. The modelling tools used for these systems were linear time series, artificial neural networks and fuzzy models. The structure of the forecasting model was derived from basic principles and it includes a combination of persistence and daily maximum air temperature as input variables. Assessment of the models is based on two indices: their ability to forecast well an episode, and their tendency to forecast an episode that did not occur at the end (a false positive). All the models tried in this work showed good forecasting performance, with 70-95% of successful forecasts at two monitor sites: Downtown (moderate impacts) and Eastern (downwind, highest impacts). The number of false positives was not negligible, but this may be improved by expressing the forecast in broad classes:low, average, high, very high impacts; the fuzzy model was the most reliable forecast, with the lowest number of false positives among the different models evaluated. The quality of the results and the dynamics of ozone formation suggest the use of a forecast to warn people about excessive exposure during episodic days at Santiago. (C) 1998 Elsevier Science Ltd. All rights reserved.
  • Thumbnail Image
    Item
    Macroscopic growth of filamentous fungi on solid substrate explained by a microscopic approach
    (JOHN WILEY & SONS INC, 1999) Ferret, E; Simeon, JH; Molin, P; Jorquera, H; Acuna, G; Giral, R
    A quantitative model predicting biomass growth an solid media has been developed. The model takes into account steric interactions between hyphae and tips at the microscopic level (competition for substrate and tip-hypha collisions), These interactions effect a slowing down of the hyphal, population-averaged extension rate and are responsible, at the microscopic level, for the distribution of tip orientations observed at the colony border. At the macroscopic level, a limiting value of the colony radial extension rate is attained. A mathematical model that combines hyphal branching, tip diffusion, and biomass growth was proposed to explain such behavior. Experiments using Gibberella fujikuroi were performed to validate the model; good agreement between experiments and simulations was achieved. Most parameters can be measured by simple image analysis on the peripheral growth zone, and they have clear physical meaning; that is, they correspond to properties of single, leading hyphae. The model can be used to describe two-dimensional (2D) solid media fermentation experiments under varying culture conditions; the model can also be extended to consider growth in three-dimensional (3D), complex geometry substrates. (C) 1999 John Wiley & Sons, Inc.
  • Thumbnail Image
    Item
    Neural network method for failure detection with skewed class distribution
    (BRITISH INST NON-DESTRUCTIVE TESTING, 2004) Carvajal, K; Chacon, M; Mery, D; Acuna, G
    The automatic detection of flaws through non-destructive testing uses pattern recognition methodology with binary classification. In this problem a decision is made about whether or not an initially segmented hypothetical flaw in an image is in fact of law. Neural classifiers are one among a number of different classifiers used in the recognition of patterns. Unfortunately, in real automatic flaw detection problems there are a reduced number of flaws in comparison with the large number of non-flatus. This seriously limits the application of classification techniques such is artificial neural networks elite to the imbalance between classes. This work presents a new methodology for efficient training with imbalances in classes. The premise of the present work is that if there are sufficient cases of the smaller class, then it is possible to reduce the Size of the larger class by using the correlation between cases of this latter class, with a minimum information loss. It is then possible to create it training set for a neural model that allows good classification. To test this hypothesis a problem of great interest to the automotive industry is used, which is the radioscopic inspection of cast aluminium pieces. The experiments resulted in perfect classification of 22936 hypothetical flaws, of which only 60 were real flat-vs and the rest were false alarms.