Dear AI Enthusiasts,
Good news! The October AI Meetup is coming up in few weeks. Thanks to our sponsor TripleSense. We will be testing Online and in-person [Only 30 people will be allowed] method this time.
NOTE: if you are interested in joining in-person meetup please RSVP now [corona vaccination certificate or negative test required]. We will have drinks and perhaps snacks but no proper food this time (due to corona restrictions we are limited with safety measures).
Let's talk about the main course of the evening.
We have two speakers for this event. Our first speaker of the evening will be Dr. Andreea Lazar from Ernst Strüngmann Institute, Frankfurt. She has extensive research experience in applied brain science specifically focusing on computational capabilities of small unit of the brain in comparison with artificial self-organizing recurrent networks. So, we will have a lot of model specific technical as well as possible future application related ideas from her talk [abstract below]. Then, our next speaker will be Martin Lellep, a Physicist from Philipps-Universität Marburg (jointly working with The University of Edinburgh). His research interests is mainly focused in Fluid Dynamics, Complex Flow and machine leaning applications in dynamic systems. But here, he has a very different topic to share. From his personal observations to detailed analysis of empty NextBike stations, he has a data driven, predictive modeling based business idea to showcase [abstract below].
Hold on, we are not done yet. We are bringing a live demo by Dan Fitzpatrick on AI stuffs. It will be a surprise; so not disclosing it right now. Therefore, stay tuned and join us on the event day.
P.S: Sign in to Zoom; No sign-up required
Zoom credentials for online joining will be provided on the day of event.
Agenda for the meetup:
18:00 – 18:15; Reception
18:15 – 18:20; A welcome note from organizers
18:20 – 18:50; Speaker I: Dr. Andreea Lazar, Ernst Strüngmann Institute
Title: The unreasonable effectiveness of recurrent cortical networks.
18:50 - 19:10; Demo of cool AI stuff by Dan Fitzpatrick
19:10 – 19:40; Speaker II: Martin Lellep, School of Physics and Astronomy, The University of Edinburgh
Title: Nextbike in Marburg: An Analysis.
19:40 – 19:50; Announcements (open positions, demos, etc.)
19:50 – Socializing
Location: Zoom platform + TripleSense GmbH [in-person location]
Recurrent neural networks are powerful computational tools that exploit long-term dependencies in the input when calculating the output. Here we take a look at the computational capabilities of cortical microcircuits in the visual system and compare them to artificial self-organizing recurrent networks. We show that self-supervised learning of an image-set alters the internal dynamics of both biological and artificial recurrent networks, in a similar manner, resulting in an optimized encoding of stimulus identity and stimulus structure.
Imagine you are running late for your bus and decide to grab a bike-sharing bike to get there in time. More often than not I found myself standing at an empty station only to miss my bus.
Here, I present you my data-driven approach to avoid walking to empty bike-sharing stations using data I scraped myself from Nextbike in the city of Marburg. Additionally, my hobby analysis worth >600,000 data points goes far beyond the question of empty stations by calculating many more real-life statistics. Finally, it concludes with a few predictive models that might inspire future business ventures.
Claim the event and start manage its content.I am the organizer