Mining Condensed Set Of Patterns From Time Series Data For Explainable Prediction

Universities and Institutes of France
January 31, 2023
Offerd Salary:Negotiation
Working address:N/A
Contract Type:Temporary
Working Time:Full time
Working type:N/A
Job Ref.:N/A
  • Organisation/Company: INSA Strasbourg
  • Research Field: Computer science › Programming Mathematics › Mathematical analysis Technology › Information technology
  • Researcher Profile: First Stage Researcher (R1) Recognised Researcher (R2)
  • Application Deadline: 31/01/2023 23:00 - Europe/Athens
  • Location: France › Strasbourg
  • Type Of Contract: Temporary
  • Job Status: Full-time
  • Hours Per Week: 35
  • Offer Starting Date: 01/02/2023
  • Project in a few words:

    French and German manufacturing companies are known for their high-quality production and their

    orientation towards smart factories. Quality assurance and control of complex production systems is

    a major challenge and is further exacerbated by the shortage of skilled labour in this domain. Quality

    problems must be detected and eliminated quickly. When a quality problem is detected, it is

    necessary to quickly understand the multiple possible causes for it (which can sometimes be

    contradictory to each other) in order to propose the most appropriate corrective actions to return

    the manufacturing process to its normal operating mode.

    The XQuality project is researching hybrid and explainable AI approaches to help manufacturing

    companies implement intelligent and automated quality assurance. The project combines data-based

    machine learning, semantic technologies and expert knowledge to monitor and explain product and

    process quality targets in a company. The goal is to develop an AI-based system that will assist the

    staff in identifying the main causes of quality issues as early as possible, to achieve reliability

    engineering in the domain of manufacturing, thanks to the new quality assurance models.

    PostDoc Subject:

    XQuality will develop new mining algorithms that analyze time series data and generate a condensed

    set of rich patterns called chronicles (Sellami et al. 2020). Unfortunately, as highlighted in literature,

    the number of chronicles is still high and their mining is time expensive. Therefore, we aim to reduce

    the set of generated patterns by optimising several quality set metrics such as discriminance,

    confidence to cite a few (Sahoo et al 2015) aiming toward more explainable patterns.

    From a methodological point of view, we aim to propose a new mining algorithm to mine rich and

    sequential patterns called chronicles. A new condensed set based on the closure definition of

    patterns is proposed. Additionally, machine troubleshooting documents will be mined using a CRF

    based technique to find corrective actions to each quality loss situation. The idea is to associate

    patterns to situations and corrective actions. This quality of this set is evaluated on a predictive task

    to assess the accuracy of quality loss prediction. we extend the work of (Sellami et al 2020) by

    integrating several constraints in the chronicle mining algorithm. These quality constraints are

    related to either the domain knowledge or to quality rule metrics such as discriminance, confidence.

    Two directions could be undertaken: either by declarative programing or optimization process (Guyet

    et al 2017). The found condensed set of chronicles is applied on a prediction task of quality loss

    prediction to ensure quality loss monitoring.

    Additional comments

    Startup scheduled: Date of the start could be planned with the selected candidate after

    January 1st 2023

    Required Research Experiences
  • Computer science

  • 1 - 4

    Offer Requirements
  • Computer science: PhD or equivalent

  • ENGLISH: Excellent


    The job requires a strong pattern mining and mathematical background. A

    previous experience on data mining, constraint programming/uncertainty modelling are highly

    recommended. Having a PhD contribution pattern mining and explainable AI on time series is highly

    appreciated. Additionally, previous experience on Industry 4.0 and quality control is highly


    Specific Requirements

    Knowledge in time series pattern mining is expected from any application.

    Applicants should show the adequacy of their profile to the project description. Having published in

    top ranked data mining conference is highly appreciated.

    Contact Information
  • Organisation/Company: INSA Strasbourg
  • Organisation Type: Public Research Institution
  • Website: https: // www.
  • E-Mail:
  • Country: France
  • City: Strasbourg
  • State/Province: Grand Est
  • Postal Code: 67000
  • Street: 24 boulevard de la Victoire
  • From this employer

    Recent blogs

    Recent news