On November 28, 2019, DIGHUMLAB and Human Futures AU hosted a full-day workshop on artificial intelligence and machine learning for humanities researchers.
The workshop took place in the beautiful Richard Mortensen Stuen, and we were happy to welcome a “full house” of researchers, almost 50 participants – mainly from Aarhus University, but also Aalborg University and the Royal Danish Library.
The day was kicked off by DIGHUMLAB project lead Birte Christensen-Dalsgaard with a welcome address:
Thomas Bolander: "What is artificial intelligence – and where is it heading?"
First presenter of the day was Thomas Bolander, Professor, Department of Applied Mathematics and Computer Science at DTU, with an introduction to AI entitled “What is artificial intelligence – and where is it heading?”. Thomas’ introduction was divided into three parts (see slides below):
- Intro to AI
- Symbolic and subsymbolic AI
- Current trends and hard problems in AI.
The vast majority of participants answered “More than 160 years (potentially never)”.
Other highlights from Bolander’s talk include a discussion on the human bias/interference in AI and training of AI, a fascinating live example of deep neural network detecting a “giant panda” teddy bear on Thomas’ webcam, and finally a broad range of examples of AI:
- AI in everyday surroundings (CaptionBot image recognition, Google driverless cars, Siri on iPhone, Google Search Engine)
- AI in sci-fi (Star Wars, Wall-E, Her)
- Google DeepMind’s AlphaGo
- Microsoft Tay twitter-bot
- IBM Watson (Jeopardy World Champion)
- and many, many more.
It was a solid and all-round introduction to current and critical discussions from the field of AI as well as an inspiring and fascinating opening of the workshop as a whole.
Impressive to see so many diverse humanities researchers gathered around AI
After Thomas Bolander’s introduction and just before the lunch break, Birte Christensen-Dalsgaard facilitated a “round table” presentation where all workshop participants shared their name, affiliation and a few words on their research interests.
This way, the researchers had a better chance of finding and networking with colleagues, who share similar research interests, during the lunch break.
Upon completing the round table, Mads Rosendahl Thomsen, project lead in Human Futures AU, commented that he was both impressed to see so many diverse researchers gathered in the same room around a common AI interest – and inspired to hear about their diverse, yet overlapping research projects.
Leon Derczynski: “Sociolinguistics from data, on the back of an envelope”
The second presenter of the day was Leon Derczynski, Assistant Professor, Department of Computer Science at ITU, with the talk entitled “Sociolinguistics from data, on the back of an envelope”.
Leon’s talk was about natural language processing (NLP) – specifically how writing styles hold information about the author and the world around the author. Text can carry information on e.g. gender, age, class, demography and more.
One Danish example was the distribution of the use of the word “træls” (used more in Jutland than the rest of the country) as well as the use of “sin/sit” (used more in Zealand).
Furthermore, Leon did a “Python Notebook session” where he demonstrated live how to use NLP to identify the gender of an author.
He sampled a quote from Madonna from the BBC article “Madonna cancels shows amid ‘overwhelming pain'” (“Doing my show every night brings me so much joy and to cancel is a kind of punishment… But the pain I’m in right now is overwhelming, and I must rest and follow doctor’s orders.”) and ran it through the code. As predicted, the code estimated that the author is female.
Henrik Pedersen: “Deep learning is for everyone”
The final talk of the day was given by Henrik Pedersen, Head of Visual Computing Lab at the Alexandra Institute, and the talk was entitled “Deep learning is for everyone”.
Henrik’s talk had a main focus on computer vision. It was very inspirational and started and ended with the following key takeaways:
- Think big – ‘everything’ is possible.
- Focus on areas where you are the expert
- Everyone can learn this.
- Pair up with experts who can guide you through the process
- Hire student programmers.
During the talk, Henrik provided both interesting cases of the work they do at the Alexandra Institute and valuable resources for researchers getting started with AI and computer vision:
- Machine Learning For Kids
- Two minute papers | What a time to be alive!
- Experiments with Google – AI Experiments
- Experiments with Google – Teachable Machine
Henrik finished his talk with an open invitation for collaborations and partnerships.
Networking and "tak for i dag"
After the final talk of the day, Mads Rosendahl-Thomsen wrapped up the day with a summary and key takeaways. After the workshop, the organisers hosted an informal networking session in the adjacent Studiecaféen in Studenterhus Aarhus. Participants were invited to have a glass of wine and discuss the talks of the day.
Thanks to everyone who participated or helped out for an amazing event.
Stay tuned on DIGHUMLAB’s and Human Futures’ channels for more digital humanities events. We hope to see you again in the future!
Responsible AI with Virginia Dignum, February 25-26, 2020 in Aarhus
Virginia Dignum visits Aarhus University on February 25-26, 2020, for a conversation about Responsible AI and a roundtable seminar. Both events are hosted by Human Futures. Read more on the official event page.
Professor, Department of Applied Mathematics and Computer Science, DTU
Research in logic and artificial intelligence (AI), with particular focus on social aspects of AI. Serves on a number of commissions, expert panels and boards, primarily concerning technical and societal aspects of AI. Teacher of the year award at DTU, 2006; H. C. Ørsted Silver Medal for excellence in science communication, 2019.
Assistant Professor, Department of Computer Science, ITU
Leon researches in natural language processing. His research covers noisy and high-variation text, NLP for Nordic languages, social media processing, and information extraction. He was program co-chair for COLING 2018, co-investigator of an EC Horizon 2020 RIA, Comrades, and has co-organised SemEval shared tasks in 2013/15/16/17/19. He also has both published and spoken.
Head of Visual Computing Lab, PhD, Alexandra Institute
Over his career, Henrik has been in various academic positions, covering research and teaching in computer vision and deep learning. He still teaches the “Deep Learning for Visual Recognition” course at Department of Computer Science at Aarhus University. In his current position, Henrik heads several national research and innovation projects all with the mission of leveraging recent advances in deep learning to make computer vision technology available and accessible to Danish companies and organizations.