François Fouquet

Thomas Hartmann, Francois Fouquet, Yves Le Traon et al.

Intelligent software systems continuously analyze their surrounding environment and accordingly adapt their internal state. Depending on the criticality index of the situation, the system should dynamically focus or widen its analysis and reasoning scope. A naive -why have less when you can have more- approach would consist in systematically sampling the context at a very high rate and triggering the reasoning process regularly. This reasoning process would then need to mine a huge amount of data, extract a relevant view, and finally analyze this adequate view. This overall process would require some heavy resources and/or be time-consuming, conflicting with the (near) real-time response time requirements of intelligent systems. We claim that a continuous and more flexible navigation into context models, in space and in time, can significantly improve reasoning processes. This paper thus introduces a novel modeling approach together with a navigation concept, which consider time and locality as first-class properties crosscutting any model element, and enable the seamless navigation of models in this space-time continuum. In particular, we leverage a time-relative navigation (inspired by the space-time and distortion theory [7]) able to efficiently empower continuous reasoning processes. We integrate our approach into an open-source modeling framework and evaluate it on a smart grid reasoning engine for electric load prediction. We demonstrate that reasoners leveraging this distorted space-time continuum outperform the full sampling approach, and is compatible with most of (near) real-time requirements.

Andrey Boytsov, Francois Fouquet, Thomas Hartmann et al.

T-distributed stochastic neighbor embedding (tSNE) is a popular and prize-winning approach for dimensionality reduction and visualizing high-dimensional data. However, tSNE is non-parametric: once visualization is built, tSNE is not designed to incorporate additional data into existing representation. It highly limits the applicability of tSNE to the scenarios where data are added or updated over time (like dashboards or series of data snapshots). In this paper we propose, analyze and evaluate LION-tSNE (Local Interpolation with Outlier coNtrol) - a novel approach for incorporating new data into tSNE representation. LION-tSNE is based on local interpolation in the vicinity of training data, outlier detection and a special outlier mapping algorithm. We show that LION-tSNE method is robust both to outliers and to new samples from existing clusters. We also discuss multiple possible improvements for special cases. We compare LION-tSNE to a comprehensive list of possible benchmark approaches that include multiple interpolation techniques, gradient descent for new data, and neural network approximation.

Ludovic Mouline, Thomas Hartmann, François Fouquet et al.

Smart systems are characterised by their ability to analyse measured data in live and to react to changes according to expert rules. Therefore, such systems exploit appropriate data models together with actions, triggered by domain-related conditions. The challenge at hand is that smart systems usually need to process thousands of updates to detect which rules need to be triggered, often even on restricted hardware like a Raspberry Pi. Despite various approaches have been investigated to efficiently check conditions on data models, they either assume to fit into main memory or rely on high latency persistence storage systems that severely damage the reactivity of smart systems. To tackle this challenge, we propose a novel composition process, which weaves executable rules into a data model with lazy loading abilities. We quantitatively show, on a smart building case study, that our approach can handle, at low latency, big sets of rules on top of large-scale data models on restricted hardware.

Thomas Hartmann, Assaad Moawad, Francois Fouquet et al.

Gaining profound insights from collected data of today's application domains like IoT, cyber-physical systems, health care, or the financial sector is business-critical and can create the next multi-billion dollar market. However, analyzing these data and turning it into valuable insights is a huge challenge. This is often not alone due to the large volume of data but due to an incredibly high domain complexity, which makes it necessary to combine various extrapolation and prediction methods to understand the collected data. Model-driven analytics is a refinement process of raw data driven by a model reflecting deep domain understanding, connecting data, domain knowledge, and learning.

Fouquet Francois, Grégory Nain, Brice Morin et al.

The creation of Domain Specific Languages(DSL) counts as one of the main goals in the field of Model-Driven Software Engineering (MDSE). The main purpose of these DSLs is to facilitate the manipulation of domain specific concepts, by providing developers with specific tools for their domain of expertise. A natural approach to create DSLs is to reuse existing modeling standards and tools. In this area, the Eclipse Modeling Framework (EMF) has rapidly become the defacto standard in the MDSE for building Domain Specific Languages (DSL) and tools based on generative techniques. However, the use of EMF generated tools in domains like Internet of Things (IoT), Cloud Computing or Models@Runtime reaches several limitations. In this paper, we identify several properties the generated tools must comply with to be usable in other domains than desktop-based software systems. We then challenge EMF on these properties and describe our approach to overcome the limitations. Our approach, implemented in the Kevoree Modeling Framework (KMF), is finally evaluated according to the identified properties and compared to EMF.

Donia El Kateb, François Fouquet, Johann Bourcier et al.

In the last years, multi-objective evolutionary algorithms (MOEA) have been applied to different software engineering problems where many conflicting objectives have to be optimized simultaneously. In theory, evolutionary algorithms feature a nice property for runtime optimization as they can provide a solution in any execution time. In practice, based on a Darwinian inspired natural selection, these evolutionary algorithms produce many deadborn solutions whose computation results in a computational resources wastage: natural selection is naturally slow. In this paper, we reconsider this founding analogy to accelerate convergence of MOEA, by looking at modern biology studies: artificial selection has been used to achieve an anticipated specific purpose instead of only relying on crossover and natural selection (i.e., Muller et al [18] research on artificial mutation of fruits with X-Ray). Putting aside the analogy with natural selection , the present paper proposes an hyper-heuristic for MOEA algorithms named Sputnik 1 that uses artificial selective mutation to improve the convergence speed of MOEA. Sputnik leverages the past history of mutation efficiency to select the most relevant mutations to perform. We evaluate Sputnik on a cloud-reasoning engine, which drives on-demand provisioning while considering conflicting performance and cost objectives. We have conducted experiments to highlight the significant performance improvement of Sputnik in terms of resolution time.