UMR CNRS 7253

Site Tools

en:research

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
Last revisionBoth sides next revision
en:research [2012/10/26 11:30] – sdestercen:research [2015/05/22 09:37] – sdesterc
Line 5: Line 5:
 ====== Practical uncertainty representations ====== ====== Practical uncertainty representations ======
  
-Work (mainly) benefiting from collaborations with D. Dubois, M. Troffaes, E. Miranda, L. Utkin, E. Chojancki and E. Quaeghebeur+Work (mainly) benefiting from collaborations and discussions with D. Dubois, M. Troffaes, E. Miranda, L. Utkin, E. Chojnacki, E. Quaeghebeur and I. Sanchez
  
 There exist many practical representations in imprecise probability theories, including possibility distributions, belief functions, imprecise probability assignments, pari-mutuel models, imprecise cumulative distributions (p-boxes), clouds, ... There exist many practical representations in imprecise probability theories, including possibility distributions, belief functions, imprecise probability assignments, pari-mutuel models, imprecise cumulative distributions (p-boxes), clouds, ...
Line 11: Line 11:
 Aside establishing new relations between the properties of several models (i.e. clouds, p-boxes, possibility distributions), we have proposed a model called generalized p-box that models uncertainty by probabilistic bounds over collection of nested sets. Such models appear naturally in elicitation procedures or statistical confidence structures, and first results indicates that generalized p-boxes may be an interesting non-parametric model to handle multivariate problems and/or to handle bipolar information.  Aside establishing new relations between the properties of several models (i.e. clouds, p-boxes, possibility distributions), we have proposed a model called generalized p-box that models uncertainty by probabilistic bounds over collection of nested sets. Such models appear naturally in elicitation procedures or statistical confidence structures, and first results indicates that generalized p-boxes may be an interesting non-parametric model to handle multivariate problems and/or to handle bipolar information. 
  
-Our current focus is to link this model with other models (such as info-gap theory) that use the idea of nested sets to propose solutions to statistical problems, and to work out what are the practical usefulness and limitations of the model.  +Our latest research on the topic include the specification of p-boxes limitations, as well as the investigation of some specific cases of comparative probabilities (where only singletons probabilities are qualitatively compared).
  
 ====== Information fusion and combination ====== ====== Information fusion and combination ======
  
-Work (mainly) benefiting from collaborations with D. Dubois, P. Buche, B. Charnomordic, E. Chojnacki, R. Thomopoulos, F. Sais, T. Burger and F. Pichon+Work (mainly) benefiting from collaborations and discussions with D. Dubois, P. Buche, B. Charnomordic, E. Chojnacki, R. Thomopoulos, F. Sais, T. Burger and F. Pichon
  
 Merging information from multiple sources is a recurring problem in modern systems. Common problems encountered by such merging are to cope with dependent and conflicting sources and to take account of sources characteristics (reliability, propensity to lie, precision, conflict level, ...). Merging information from multiple sources is a recurring problem in modern systems. Common problems encountered by such merging are to cope with dependent and conflicting sources and to take account of sources characteristics (reliability, propensity to lie, precision, conflict level, ...).
Line 22: Line 21:
 For practical purposes, we have proposed to use the notion of maximal coherent subsets as a way to deal with conflict among sources, and have applied it to the problem of estimating source reliability form meta-information.  For practical purposes, we have proposed to use the notion of maximal coherent subsets as a way to deal with conflict among sources, and have applied it to the problem of estimating source reliability form meta-information. 
  
-Our current focus in this area concerns the characterization of inconsistency and relaibility degrees resulting from assumptions about the source of information or from the combination of the different pieces of information.+Our current focus in this area concerns the characterization of inconsistency and reliability degrees resulting either from assumptions about the source of information or from the combination of the different pieces of information. Such degrees can then be used to guide the fusion process. 
 + 
  
 ====== Uncertainty propagation and (in)dependence modelling ====== ====== Uncertainty propagation and (in)dependence modelling ======
  
-Work (mainly) benefiting from collaborations with D. Dubois, G. De Cooman E. Chojnacki, J. Baccou, T. Burger, M. Sallak and F. Aguirre+Work (mainly) benefiting from collaborations and discussions with D. Dubois, G. De Cooman, E. Chojnacki, J. Baccou, T. Burger, M. Sallak, M.C.M. Troffaes, F. Coolen, S. Ferson, F. Aguirre and I. Sanchez
  
 How to propagate uncertainty analysis in various models is an important issue that may face several difficulties. Most of my research in this domain has concerned the propagation of uncertainty model through deterministic functions with methods combining Monte-Carlo simulation and interval analysis, with an industrial risk-assessment purpose.  How to propagate uncertainty analysis in various models is an important issue that may face several difficulties. Most of my research in this domain has concerned the propagation of uncertainty model through deterministic functions with methods combining Monte-Carlo simulation and interval analysis, with an industrial risk-assessment purpose. 
  
-Related problems are how to model independence to obtain joint models (and how to compute with such latter models), or how to simulate a given imprecise probabilistic model. +Related problems are how to model independence to obtain tractable joint models (and how to compute with such latter models), or how to simulate a given imprecise probabilistic model. 
  
-Some of our recent research also deals with the problem of how to efficiently evaluate the reliability when the component reliabities are uncertain and modelled by imprecise probabilistic knowledge (more particularly belief functions). +Some of our recent research also deals with the problem of how to efficiently evaluate the reliability when the component reliabilities are uncertain and modelled by imprecise probabilistic knowledge (more speicifically belief functions). 
  
-====== Classification problems ====== 
  
-Work (mainly) benefiting from collaborations with B. Quost, T. Denoeux, B. Ben Yaghlane, N. Sutton-Charani, E. Hullermeier, A. Antoinucci and G. Corani+ 
 +====== Learning problems ====== 
 + 
 +Work (mainly) benefiting from collaborations and discussions with B. Quost, T. Denoeux, B. Ben Yaghlane, N. Sutton-Charani, G. Yang, M. Masson, E. Hüllermeier, A. Antonucci, G. Corani, M. Poss and N. Ben Abdallah
  
 Outside of extending some classical classifiers (k-NN methods, Naïve networks) to imprecise probabilistic settings, our work currently focuses on the combination of classifiers, to address both the usual multi-classification problem, as well as more complex problems such as label ranking and multilabel classification. Outside of extending some classical classifiers (k-NN methods, Naïve networks) to imprecise probabilistic settings, our work currently focuses on the combination of classifiers, to address both the usual multi-classification problem, as well as more complex problems such as label ranking and multilabel classification.
  
-One of our current favorite field of investigation is the so-called binary decomposition, where complex problems are decomposed in several binary ones (facilitating the learning but increasing the number of models to learn). +One of our current favorite field of investigation is the so-called binary decomposition, where complex problems are decomposed in several binary ones (facilitating the learning but increasing the number of models to learn). In the future, we plan to focus more on active learning topics, where imprecise probabilistic methods can have an important role to play, due to their ability to identify cases where information is missing. 
  
  
 ====== Applications ====== ====== Applications ======
  
-Work (mainly) benefiting from collaborations with P. Buche, B. Charnomordic, O. Strauss, V. Guillard, E. Chojnacki+Work (mainly) benefiting from collaborations and discussions with P. Buche, B. Charnomordic, O. Strauss, V. Guillard, E. Chojnacki, M. Sallak, I. Thouvenin
  
 We have applied ideas coming from imprecise probability theories and more generally concerning uncertainty handling to a number of frameworks, including: We have applied ideas coming from imprecise probability theories and more generally concerning uncertainty handling to a number of frameworks, including:
 +
   * Flexible querying in data bases (P. Buche, V. Guillard)   * Flexible querying in data bases (P. Buche, V. Guillard)
-  * Signal filtering with kernels (O. Strauss)+  * Signal filtering with kernels (O. Strauss, F. Comby)
   * Knowledge Engineering (B. Charnomordic, R. Thomopoulos)   * Knowledge Engineering (B. Charnomordic, R. Thomopoulos)
-  * Risk analysis and robust design (E. Chojancki, V. Guillard)+  * Risk analysis and robust design (E. Chojancki, V. Guillard, M. Sallak) 
 +  * Process modelling (C. Baudrit) 
 +  * Virtual training (I. Thouvenin)

Page Tools

User Tools

Mentions légales