HamburgAI #4: AI powered Talks
Since almost one year, collectAI is the host of HamburgAI, an initiative of CityAI. Our fourth edition of HamburgAI took place on Wednesday, March 28, 2018 at Hamburg Mindspace. A mix of over 130 AI professionals, enthusiasts and business people welcomed our four speakers for the evening, moderated by collectAI CCO Steve Emecz.
Why invest in AI?
First was Stephan Uhrenbacher, founder (Qype, Flio and many more) and Machine Learning (ML) investor. A very nice compliment to us was that Stephan mentioned that it’s the first time in Hamburg he is holding a talk in English and it doesn’t feel weird ;). Providing insights into his experiences as an early stage investor in AI, Stephan elaborated on why Canada is such a huge market in the sector. He also pointed at the extremely highly valued AI startups that were derived due to Gartner’s Hype Cycle in 2017.
Some quotes worth remembering:
“We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.” Roy Amara
“Artificial intelligence is like the new hammer and everybody is looking for nails. Chances are that they will find them.”Ben Evans, Andreessen Horowitz
Amongst others, one interesting of his investments is ELUCID LABS. Using their technology doctors can perform a digital and non-invasive biopsy within seconds, using an AI dermatology assistant (AIDA) system.
“A computational imager probes 4000 molecular parameters of the skin to produce a digital fingerprint of the lesion. A deep learning engine then translates this fingerprint into clinical information that can help physicians make a more accurate and timely diagnosis of skin diseases.”
In-depth Data Scientists Talks at HamburgAI#4
Second was Florian Müller, Data Scientist @Tchibo on Managing Uncertainty in Time Series Forecasts using Probabilistic Programming. As part of his talk, he explained the functions of Prophet, a tool provided by the Facebook Research Lab.
Matthias Rettenmeier Chief Data Scientist @Zalando elaborated on Understanding Consumer Journeys with RNNs. The model directly relates consumer actions during brand sales. It is a flexible solution. Features are not predefined by tapping into someone’s brain, but learned by the model during training – a compute hungry process and provides predictions that yield measure of current interest in a single brand. Allover, a great way for better targeting and brand marketing.
During our first AI Challenge of the evening, Jan Erik Herrmann, ML Engineer @Lufthansa Global Business Services focused on Interpreting Contract Content. Interestingly, Erik showed their use case where 2300 documents had been scanned as PDFs with the challenge to manage imbalances amongst datasets. The audience came up with several interesting suggestion to help solve his problem. Some of the suggestions included using word embeddings, LDA for topic modelling and many more.
After HamburgAI is before HamburgAI! Next event in June
Join our upcoming event on June 27th! The last HamburgAI event was sold out. So you better hurry up to get your ticket now :). As always, the event is kindly hosted @Hamburg Mindspace. For regular updates from the community, please sign up for our Newsletter.
For those of you who haven’t attended our community gathering yet: HamburgAI brings together representatives from startups, companies and universities to discuss the current state of AI as well as future visions. More info is available on Hamburg.ai. Another reading recommendation is the JAAI article about this event.