Postdoc Application: Hybrid Knowledge Discovery with Application to Biomedical Data

Postdoc Subject. 

Hybrid Knowledge Discovery with Application to Biomedical Data (GeenAge Project, LUE 2018) 

many thanks to spread this postdoc offer (more details in pdf attached file). 
Sorry for multiple postings. 


Proposed by Marie-Dominique Devignes and Amedeo Napoli, 
LORIA / Inria Nancy Grand Est, 
BP 239, 54506 Vandoeuvre les Nancy, 
(Marie-Dominique.Devignes ; Amedeo.Napoli@loria.fr) 


Context, positioning and objectives of the proposal. 

Human agents have the remarkable capability to learn a large variety 
of concepts, often with very few examples, whereas current 
state-of-the-art data mining algorithms require hundreds or thousands 
of data points and struggle with problems such as ambiguity, validity, 
overfitting etc. Another characteristic of human agents is the 
ability to acquire knowledge about the world and to draw subsequent 
inferences. Following this line, we are interested in investigating 
the combination of numeric and symbolic data mining and as well the 
use of domain knowledge for improving the performances and 
capabilities of different data mining approaches. Indeed, data mining 
methods can be either symbolic or numerical, and applying one or the 
other to a given dataset does not provide the same output. Combining 
numeric and symbolic data mining methods remains a tasks still poorly 
investigated. 

Accordingly, the objectives of this postdoc research work are to study 
how Knowledge Discovery (KD) should be carried out, given the data at 
hand and a collection of data mining methods. In this way, we 
consider that the knowledge discovery process is iterative, 
interactive, supported by graphical tools, and dependent on various 
dimensions related to data, domain knowledge and the target task 
(problem-solving). We intend to study the potential and the 
characteristics of a such an hybrid KD process, and establish an 
operational and reusable methodology. Hybrid means that symbolic and 
numerical methods, as well as supervised and non supervised methods, 
can be combined for mining complex and possibly large data. 

This study takes place within the GeenAge Research Project (GeenAge is 
one of the so-called ``IMPACT Projects'' funded by ``Lorraine 
Université d'Excellence'' LUE). The ambition of the GeenAge project is to 
design new strategies of diagnosis and management of healthy and 
pathological ageing by targeting the functional consequences of the 
interplay between genes, epigenome and environment. Our teams are 
particularly involved in the Axis ``New transcriptomics avenues in 
diagnosis and therapy of age-related diseases''. In this axis, 
biologists are characterizing circulating non coding RNAs in the blood 
of patients from a longitudinal cohort, the Stanislas cohort. 

Thus, the mission of the post-doctoral fellow will be to design a 
framework for hybrid knowledge discovery and to implement the related 
algorithms in the context of the GeenAge project. Interactions with 
biologists will help to identify the potential domain knowledge to be 
reused. Visualization and data mining tools will be deployed in this 
hybrid and knowledge-based approach. 


The Organization of the Research Work and the Details for Application. 

The research work will be organized around three main tasks. 
- Task 1. ``Hybrid Knowledge Discovery''. 
- Task 2. ``An Integrated System for Hybrid Knowledge Discovery''. 
- Task 3. ``Publications''. 

Details for Application. 

Project Teams: Capsid and Orpailleur (LORIA/Inria Nancy Grand Est) 

Supervisors: Marie-Dominique Devignes (CR CNRS, Marie-Dominique.Devignes@loria.fr ; https://members.loria.fr/MDDevignes/). 

Amedeo Napoli (DR CNRS, Amedeo.Napoli@loria.fr ; https://members.loria.fr/ANapoli/). 

Keywords: knowledge discovery, mining of complex data, pattern mining, 
numerical data mining, meta-mining, biomedical data. 

Skills and profile of the candidate: A PhD Thesis in Computer Science 
or in Applied Mathematics. Prior research work on knowledge discovery, 
data mining and or machine learning will be highly 
appreciated. Additional knowledge about biomedical data and processes 
will be welcome too. 

Job location, terms and duration: 
This two-year position is based in LORIA/Inria Nancy Grand Est Lab in Nancy (Capsid and Orpailleur teams). 
Applicants will be interviewed by a commission in October 2018. 
The duration of the postdoc is 24 months with a planned starting date in November 2018 (this date is flexible). 

How to apply: 
Applicants are requested to submit the following elements: 
- A full and pudated CV including the list of publications of the applicant. 
- A motivation letter related to the position. 
- Recommendation letters. 
- Academic transcripts (if possible). 

The Deadline for applications is September 16th, 2018. 

Applications are only accepted through email. 
All documents must be sent to marie-dominique.devignes@loria.fr and amedeo.napoli@loria.fr.