ball bearings

Three industrial doctoral students will present their research at AI conference

At this year’s workshop of the Swedish Artificial Intelligence Society, three doctoral students from the industry graduate school Data Intensive Applications will take part. They will review their research in unsupervised defect detection in manufacturing, machine learning models in forest industry, and identifying legitimate quality issues.

On June 12–13, SAIS 2023, the 35th annual workshop of the Swedish Artificial Intelligence Society (SAIS) will take place in Karlskrona, Sweden. From the industry graduate school Data Intensive Applications (DIA9 at Linnaeus University, three doctoral students will present their research.

“We encouraged some of our industrial doctoral students to submit papers and three were selected to present. We plan to invite all our DIA doctoral students to attend the workshop, since it’s a great opportunity to network and meet other researchers”, says Diana Unander, coordinator of DIA.

A case for unsupervised defect detection in manufacturing

Felix Viberg, industrial doctoral student at SKF, will discuss two distinct approaches that can be employed for automatic visual inspections in the manufacturing industry.

“Currently, the most widely used method for solving this problem is through the use of supervised machine learning. We will be comparing this method to an alternative, unsupervised approach based on normalizing flows”, says Felix Viberg.

Improving supervised machine learning models in forest industry with generated data

Dag Björnberg is an industrial doctoral student at Softwerk.

“I’m exploring the feasibility of using generated image data as a supplementary resource to real images in a variety of learning tasks. Specifically, as a proof of concept, the generated data is being employed in various classification tasks that are relevant to the forestry industry. This includes, for instance, counting the number of annual rings present in a log end image.”

Towards better product quality: Identifying legitimate quality issues through NLP & machine learning techniques

The third participating doctoral student is Rakhshanda Jabeen at Electrolux. She will discuss her research on using natural language processing (NLP) and machine learning (ML) to identify genuine quality issues.

“When a customer reports a quality issue via a service call, it is crucial to retrospectively determine whether the call was a legitimate quality issue or not. By analysing textual data from both customers and service technicians, I have compared various approaches in order to identify the most effective method for classifying calls.”

More information