Post-Doc in Explanable AI and Machine Learning - 18 months contract

ABOUT INSTITUT MINES-TELECOM BUSINESS SCHOOL

Institut Mines-Telecom Business School is a public business school, following the French Grandes Ecoles tradition, and is part of IMT (Mines-Telecom Institute)—the primary French technology institute dedicated to higher education, research, and innovation in engineering and management. Established in 1979, the School is under the purview of the Ministry of Economy and is renowned for its proficiency in management, innovation, entrepreneurship, ethics, and information and communication technologies.

Situated on the same campus with the engineering school, Telecom SudParis, the School fosters a unique synergy in France. Additionally, it houses an on-campus business incubator for young, innovative companies. The institution is committed to cultivating future managers and entrepreneurs who are responsible, innovative, globally aware, and equipped to navigate their organizations through the significant transitions of tomorrow's society.

Routinely ranking among the top 10 management schools in France for post-graduation salary levels, Institut Mines-Telecom Business School offers a comprehensive program range, spanning from undergraduate to doctorate, all in line with international standards. The School holds accreditation from AACSB and AMBA, with courses delivered in both French and English. Furthermore, for those seeking to learn or enhance their proficiency in French, the School provides courses in French as a foreign language.

More information on Institut Mines-Telecom Business School can be found at https://www.imt-bs.eu/ .

MISSIONS

Many procedures currently rely on predictions made by an Artificial Intelligence (AI) model, serving as a decision support element for human operators. To improve this type of collaboration between humans and machines, it may be beneficial to provide the human operator with information about the overall behavior of the predictor, as well as enrich the predictions with elements intended to explain them. For example, this means providing information to clarify, for a given input, the prediction of a deep learning model, such as a Vision Transformer: why does this model arrive at a particular classification result for a given image? Explainable AI techniques, or XAI, have been proposed and studied in the literature. These techniques allow us to assess the overall behavior of trained models and explain specific predictions based on the inputs (through local interpretability methods). These techniques rely on assumptions regarding the understanding of predictive models' processing and the importance given to the information provided to them.

What is the impact of XAI techniques on the collaboration between humans and machines? The proposed post-doctoral research aims to provide answers to this question. Indeed, in many cases, AI models are complex systems that process enormous amounts of data and perform calculations that are difficult to intuitively interpret. The objective of explainability is to provide users with information about the decision-making logic of the model, so that they can understand and trust the AI's results. If understanding an AI's conclusions can influence the trust placed in it, can the collaboration between humans and machines evolve towards a relationship of mediation, delegation, or even substitution? These are the questions we aim to explore in this study.

ACTIVITIES

You will contribute to:

• The evaluation of variations in the results produced by XAI methods in specific study contexts (e.g., image classification tasks, etc.).
• The evaluation of the impact of XAI methods on Human-Machine collaboration in simplified decision-making contexts:
• Evaluation of the human operator's performance in executing a task, in different contexts: alone, with the help of a predictive model for which decisions are not explained or explained using an XAI technique.
• Evaluation of Human-Machine collaboration: delegation, substitution, or mediation.
• Evaluation of potential biases induced by the use of AI techniques.

You will participate in defining:

• The study contexts (e.g., games, image classification, etc.) and the testing protocols to be implemented.
• The selection and implementation of predictive models and XAI techniques.
• The organization of experiments with cohorts of human operators.
• The evaluation and valorization of results.

Depending on your profile, certain aspects will be explored in greater depth than others, and additional aspects may be considered regarding the identification of XAI techniques that could address the limitations of existing techniques identified in the literature or through the tests performed. We are open to candidates’ proposals for contributions according to their areas of interest.

The project will be carried out in close collaboration with IMT Mines Alès, particularly with Jacky Montmain, Sébastien Harispe, and Andon Tchechmedjiev (CERIS laboratory). Travel or even stays in Alès may be arranged if needed.

Level of training and / or experience required:

• PhD in economics, management, or computer science on a topic related to explainable AI and/or machine learning less 3 years

Essential skills, knowledge and experience:

• XAI techniques: knowledge of the main XAI methods and tools in the field. Skills can be enhanced during the mission, but familiarity with these aspects is desirable.

Advantageous skills, knowledge and experience:

• An important part of the mission is the evaluation of human-machine collaborations, specifically evaluating the impact of AI and XAI models on human decision-making, thus prior experience working with human cohorts would be appreciated.
• Proficiency in Machine Learning techniques.

Abilities and skills:

• Work in a multidisciplinary team
• Rigor
• Listening skills and interpersonal qualities
• Ability to be proactive

APPLICATION PROCEDURE

• Deadline for applications: November, 08th 2024
• Nature of the contract: 18 months contract
  
• Location of the position : Evry-Courcouronnes (France)
• The positions offered for recruitment are open to all with, upon request, accommodations for candidates with disabilities
• Working conditions : teleworking possible, on-site restaurant and cafeteria, accessibility by public transport (with employer's contribution) or close to main roads and staff association
• Contact person : Nicolas SOULIE ; 01 60 76 46 85