FULL NAME + PROFESSIONAL DESIGNATION
Nadja Herger, PhD
Deep Learning and NLP
How did you get into this area of work? (Was there a specific moment where you said this is what I want to do?)
I started my career as a climate scientist, analyzing and visualizing high-dimensional outputs from global climate models in order to better understand the unprecedented effect humans have on our planet. This is where my passion for data started. I find making sense of our messy world through the discovery of patterns, and telling data-driven stories, very rewarding. After completing my PhD in Australia, I joined Thomson Reuters Labs a bit over a year ago in the hope of being able to play with large and diverse datasets. I am happy to report that I was not disappointed — I have been able to work on a range of challenging projects across different domains. Earlier this year, I joined the Deep Learning team, whose aim it is to stay on top of the latest advances in Deep Learning, recognize opportunities for Thomson Reuters, and work on initial feasibility studies. I am excited to be part of a team that is leveraging a technology, which has the potential to transform businesses worldwide.
What kind of projects do you work on?
Deep Learning has now reached a stage of maturity where it can solve tasks that were previously considered uniquely human. This includes natural language understanding and generation, which are core tasks within Thomson Reuters. As part of the Deep Learning team, I have been involved in extracting information from long, unstructured legal documents in order to speed up editorial processes and increase our coverage.
What is one interesting fact you’ve come across while performing your work?
I recently learned from a colleague that our US case law corpus is larger than the massive corpora, which state-of-the-art language models like BERT (Google) and GPT2 (OpenAI) have been trained on. I find that quite impressive. This puts Thomson Reuters in a unique position to develop Deep Learning models that are trained to understand the nuances in legal- and tax-specific language.
How would you apply AI/ML to improve some aspect of everyday life?
AI has the potential to transform our world. I believe the best way to improve people’s lives with AI is to make it more widely available for everyone. It is therefore essential to democratize access to this technology at an early stage. We should focus on the educational aspect to enable the use of AI for social good, such as fighting climate change and advancing health research.