Vincent Armant Research Interest

Research Interest

Research

Interest

Introduction

I am a Research Fellow at Insight Center for Data Analytics at the University College Cork, in Ireland. I'm Doctor in computer science of Paris Sud University (FRANCE). My main interest is to design complex systems and solving large problems in the area of data analysis and optimization, knowledge discovery, distributed reasoning using optimization techniques. I am currently member the of the Program Committee of the national conference JFPC 2015, and reviewer of the international conference AAAI 2015.

Data Analysis and Optimization

Actually my main focus is Data Analysis and Optimization for ride sharing systems. In [2], based on the users' behavior learnt from a history of trip schedules and a Constraint Programming model we propose a new approach for assessing the potential of ridesharing systems. We show how current ridesharing schemes can be improve up to 8o%. In [1,3] we propose different Mixed Integer Programming formulations and optimization techniques to solve more efficiently ridesharing problems.

Knowledge Discovery in Big Data

In a joint work in the area of Semantic Web, we proposed scalable algorithms and data structures to discover key properties in very large datasets of RDF triples [4]. A set of properties is a key if it uniquely identified an object. This approach is particularly useful in the context of linked open data where "same as" links have to be discovered to accurately query data and knowledge stored in geographically distributed warehouses.

Distributed Knowledge Representation and Reasoning with Privacy

During my PhD thesis, I proposed new approaches and algorithms to diagnose inherently distributed systems while preserving privacy of local knowledge. Distributed diagnosis preserving privacy aims at explaining the possible abnormal behavior of an inherently distributed system by a set of agents, each having a local view of the system. To speed up the distributed reasoning supporting the diagnosis and to tackle larger problems while preserving privacy, I have proposed new distributed algorithms based on graph decomposition [5]. I also demonstrated that any Disjunctive Normal Formal representing a system description is a compact representation that contains all the diagnoses [8,11]. Finally, I exhibited new properties and algorithms that allow a set of agents to agree on distributed optimal diagnoses [9], while preserving privacy of their local system descriptions.

Modelling Skills

Mathematical Integer Programming

Propositional formulae

Constraints

Finite State Automata

Petri Nets

Programming Skills

Java

C/C++, Pyton, R,

SQL, PLSql, MySQL, Spark,

OCaml,

HTML, PHP, Web Services, XML, XSLT, XPath

Education

Ph.D in Computer Science

University Paris-Sud

(second level of distinction)

Master degree in Computer Science

University Paris-Sud

(with honours)

Languages

French: native speaker

English: fluent

Spanish: Basic knowledge

2010 - present