AI for Business Meetup Berlin No. 3

In collaboration with Sophie Hundertmark, we at OMQ are looking forward to be involved in the planning and contributions for the third AI meetup in Berlin.

Here is the agenda (the presentations will be held in English):

18.00 – Welcome und Get Together

18.30 – Explainable machine understanding, and examples for clinical applications

Transformer models are the state-of-the-art in Question Answering and many other Natural Language Processing (NLP) tasks. However, what kind of information is stored within the models and how the information is used is still mostly hidden, so called "Black Box" problem. In this talk, Alexander Löser will introduce his work on explaining transformer networks on complex question/answering tasks (CIKM 2019), as well as his work on neural text mining (TACL 2019, TheWebConf 2018). Finally, he will outline his 'bets' in the area of neural text mining.

Presenter Info:

Alexander Löser advises platform companies, such as IBM, Zalando, eBay, MunichRe, on setting up data science organizations and for designing data products. As a full professor of Beuth-University of Applied Sciences Berlin, Alexander heads the Data Science Research Center and initiated the English language master’s program in Data Science. Founded in 1971, the university is one of the largest universities of applied sciences in Germany (13,000 students, 300 professors) and hosts one of the leading groups for neural text mining in Germany. His previous positions had been with HP Research Lab Bristol, IBM Almaden Research Center, SAP SE Research Division and the Institute of Technology Berlin. Since 2018, Alexander is also advising the Federal Ministry of Science and Education in the national AI-flagship program, is executive board member of the Einstein Center of Digital Future Berlin and mentors several AI-startups.

19.00 – Transformers in action: applying transformer models to understand customer interactions

Natural Language Understanding has been progressed nicely with the arrival of large pre-trained neural network models, such as Google's BERT or Facebook's XLM. However, applying those models in products is not trivial due to various issues. In this talk, Gil Noh visit issues and challenges that need to be overcome to use state-of-the-art transformer models on production, with OMQ's experience of applying the models to process customer support text of all European languages.

Presenter Info:

Gil Noh is a machine learning engineer at OMQ. He got his PhD in 2011 from KNU, South Korea, on the topic applying machine learning methods for determining text relatedness. After graduation, he moved to Germany and worked as researcher in Heidelberg University and in NEC Labs Europe on multiple EU projects to help applying machine learning methods for solving real world problems. Since 2016, he is working at OMQ GmbH as lead developer of OMQ's language understanding system.

19.30 – Snacks, Drinks, Networking

20.30 – End of event

We look forward to meeting you!

Your OMQ Team and Sophie Hundertmark

If you wish to get to know us ahead of the event, why not take a look at our blog, where you will find current information about AI in customer service: https://www.omq.ai/blog/