XAI-Pro News letter October 2023

A collaboration project about Explainable AI

A part of the research environment on applied AI (AFAIR) at the School of Engineering, Jönköping University

October 2023

About the project

Objectives

The project investigates how AI-based products and services can be designed, using explainable approaches, to facilitate adjustments in response to customer demands.


Research Questions

Can we uncover the inner workings of neural networks, and if so, how? How can expert knowledge be formalized into a demonstration for an IRL agent to provide a useful solution? And how can the solution from the IRL agent provide insights to educate and train humans?

Insights from ”The Black Box”

Mid-time in our explainable AI project we have successfully initiated an inspiring collaboration with our industrial partners. We have also welcomed our two PhD students who have commenced their research. The time is right to start communicating our objectives and progress to our stakeholders. From now on, this newsletter will reach you each quarter, sometimes more frequent if there is a lot going on. We are grateful for the opportunity to develop together with you all!

Maria, Florian, Marcus & Max

Hi there…

Marcus Gullstrand and Max Pettersson, our PhD Students who participated in the Swedish AI Society’s yearly conference (SAIS) in Karlskrona this summer, presenting their PhD-projects within XAI-Pro to the Swedish AI-community.

What trends did you identify?

Marcus: ”Within deep learning there’s an obvious trend focusing on semantic concepts. Meaning that the model gets to take part of established knowledge as a part of its decision making.”

Max: ”The combination of qualitative and quantitative methods for machine learning, i.e the usage of domain rules and ontologies together with data driven machine learning, is interesting.”

UP NEXT

We are currently defining practical use cases and experiments with Husqvarna and SAAB for the fall. The work with SAAB is reinforced with our SAAB colleagues Oskar Persson and Samuel Svensson who will join the team. Experiments will be conducted during the fall.

Representatives from our team will participate at the AI for Humanity and Society convent in Malmö the 14th of November. We will take part in the WASP-HS Workshop ”Knowledge Elicitation: Data Collect and Knowledge Modeling for Human-Aware Systems”.

Researchers

Contact

Maria Riveiro

XAI-Pro is one of several projects within the frame for the research profile AFAIR in the research and educational environment SPARK at Jönköping University’s School of Engineering.

Read more at ju.se/spark.


Financed by

Industrial Partners

2023-11-23