Techsia is pleased to offer the following internship opportunities for the academic year 2007/2008.
The general organization we foresee, for each internship, will be, depending on the candidates’ expertise:
1- state-of-the-art analysis, 2- participation in the definition of the technical specifications,
3- management of all or part of the related developments.

•  IMAGE PROCESSING :

• Automatic detection of sinusoidal contours in Borehole images

• “Textural” map clustering (unsupervised Neural Network application) using a topological self-organizing map: some algorithms have been developed; the main mission is now to connect them to the Borehole Imaging module

• K.mod ENHANCEMENTS: supervised Neural Network application

• Create a Multi-Layer Perceptron (MLP) committee by partitioning the learning space by a topological map or by using an existing partition

• Create a Multi-Layer Perceptron committee using an Adaboost like algorithm. The problematic is that the MLP in kmod is used as a non linear regression and not as a classifier. Adaboost was first developed to solve two classes problems. It has been extended to n-classes problems and some boosting techniques have been adapted to regressions

• CLASSIFICATION: Adaboost algorithm

• Integrate Multi-classes Adaboost algorithm with the MLP classifiers and the Decision Tree (results and generalisation improvements)

• FAST KERNEL DENSITY ESTIMATION

• Multi-dimensional logs density estimation for multi-well projects: Bayesian clustering, graphical density representation, normalisation

• UNCERTAINTY PROPAGATION INTO SUMMARIES WORKFLOW

• Implementation of propagation of errors in up scaled summary averages (Monte Carlo in Summaries)

• DECISION TREE

• Enhancement of the extraction and the representation of useful information for a decision tree

• RESISTIVITY/NMR/SONIC: 3D visualisation

• Visualization of array-style Resistivity/NMR/Sonic imager type of data using 3D viewer technology

• IMPLEMENTATION OF A STATISTICAL FUNCTIONALITY

• Implementation of Robust estimators: e.g. median regression, Least Absolute Regression, Method of Averages

• SUMMARIES STATISTICS

• Implementation of comprehensive statistical reporting through Summaries-type module… means, mode, median, Sdev, variances, quantiles, subject to zones/discriminants/facies etc.