Interview with Tania Jacob, Staff member in the »Financial Mathematics« department

»I Realized That I Am Just More Interested in Machine Learning and Data Science!«

Interview / September 28, 2023

Every year, the Sparkassen Stiftung Kaiserslautern awards several young scientists at the University of Kaiserslautern-Landau (RPTU) prizes for outstanding achievements. In 2023, one of these coveted prizes went to Tania Jacob from our department »Financial Mathematics« for her master's thesis on early anomaly detection in manufacturing processes of components. The special thing about it: After several years as a software developer in Australia, the 41-year-old decided to reorient her career, including a new master's degree program. In 2020, she came to the Fraunhofer Institute for Industrial Mathematics ITWM and studied at the RPTU. We met her and talked to her about her master's thesis and the Sparkasse Prize, her retraining, and her work at Fraunhofer ITWM:

Congratulations! You won this year’s Sparkassenstiftung prize with your master’s thesis. How does it feel?

I was very happy to receive the prize, especially after putting so much effort into my master thesis. I also think it’s great that the work of my supervisors, Prof. Dr. Marius Kloft, and two of his doctoral students, Phillip Liznerski and Saurahb Varshneya, has also been recognized with this. We worked well together and had many interesting discussions, so I think this award is also a recognition of the high standards of their work.

How exactly does such an award ceremony work? Do you submit your work yourself or are you nominated?

You are usually suggested by someone in your department – in my case, it was the managing director of the computer science department, Dr. habil. Bernd Schürmann. When I received the email from my supervisor about the nomination, I was on a six-month trip to Australia! I was surprised but very happy about it.

Your work is about detecting anomalies in so-called time series as early as possible. What exactly are these time series and what does »anomaly« mean in this context?

For my work, I developed a model for detecting anomalies in time series data. Time series are data that change over time. Here at Fraunhofer ITWM, we often work with financial time series, for example, the values of stocks on the stock exchange, which are observed over months or years. In my master’s thesis, I did research with time series in manufacturing or industrial 3D printing of components. While such an element is printed, a camera records the complete process and sends them to the model, which then processes these images into a time series. The idea is that as soon as an anomaly is detected in a sequence, the machine is automatically stopped to prevent further production of the component. This is mainly to save time and expensive materials.

© Fraunhofer ITWM
Interview with Tania Jacob.

Could your results also be used in other sectors of the economy, for example in power plants? What advantages would there be?

Although I focus my work on component manufacturing, there are many other potential applications. My method works with high-dimensional data, i.e. image or video data. Any process that is observed by cameras would therefore come into question. I can therefore imagine that my model could also be applied to data-driven processes of power plants. Early detection of anomalies in data-driven processes could possibly prevent accidents altogether or at least reduce the severity of the accident.

You mentioned that Fraunhofer ITWM is doing research on financial time series. Could your results be applied to that as well?

Yes, we could possibly apply similar methods in the financial sector. However, there is definitely a need for further research, because financial time series are usually not high-dimensional, so they are not image or video data. Therefore, my model would need to be adapted to a new type of data before it could be used successfully. Then we would also be able to detect anomalies early in the financial sector, for example price changes in the stock market.

How did you actually find your way to our institute and since when have you been at Fraunhofer ITWM?

I came to Fraunhofer ITWM in 2020. I had previously worked as a software developer for a few years, but then realized that I was more interested in Machine Learning and Data Science. Therefore, I decided to completely reorient myself professionally once again. Since I love pursuing my own ideas, conducting experiments, and discovering new possibilities, I was keen to get into research. I then simply looked for companies that could offer me these opportunities, and Fraunhofer ITWM was, of course, a good fit.

Did you then work full-time with us while you were studying?

I started working here full-time but switched to part-time during my master’s degree. The Fraunhofer ITWM fully supported me in this step and was super flexible.

And what are your plans for the future with us? Are you aiming for a doctorate?

Currently, I am glad that I have my master’s degree and am not thinking about a doctorate. Due to my rather late professional reorientation, I’m not sure how meaningful a doctorate would be now, but I don’t rule it out for the future.

Symbolbild Federated Learning Framework FACT — © freepik
Federated Learning is about training an AI model, through locally generated data. This can improve accuracy, but at the same time ensure data protection.

What projects are you currently working on?

At the moment I am working on two different projects. The first is an industrial project with our customer, who is a premium automotive manufacturer. This is about anomaly detection in after-sales data and the analysis of damage or error logs of vehicles. I’m also working on a research project in the area of Federated Learning, which is a Machine Learning approach that enables model training across decentralized devices while keeping data localized, allowing the model to learn from data without it leaving its original source. I am also preparing a workshop for it, that should provide education for the data science community, as well as help us to win new cooperation partners or customers in this area.

That sounds like a lot of work! What do you do in your free time to switch off?

I love hiking in the woods, which is why I like it so much here in this region!

Press Release

Interview with Tania Jacob, Staff member in the »Financial Mathematics« department

»I Realized That I Am Just More Interested in Machine Learning and Data Science!«

Congratulations! You won this year’s Sparkassenstiftung prize with your master’s thesis. How does it feel?

How exactly does such an award ceremony work? Do you submit your work yourself or are you nominated?

Further Information

Your work is about detecting anomalies in so-called time series as early as possible. What exactly are these time series and what does »anomaly« mean in this context?

Could your results also be used in other sectors of the economy, for example in power plants? What advantages would there be?

You mentioned that Fraunhofer ITWM is doing research on financial time series. Could your results be applied to that as well?

How did you actually find your way to our institute and since when have you been at Fraunhofer ITWM?

Did you then work full-time with us while you were studying?

And what are your plans for the future with us? Are you aiming for a doctorate?

What projects are you currently working on?

That sounds like a lot of work! What do you do in your free time to switch off?

Contact Press / Media

M.Sc. Tania Jacob

Contact Press / Media

Anika Sedlmeier

Contact Press / Media

Mona Steinmeier