The A´s of Artificial Intelligence: Automation, APIs, Analytics

What a success for the organizers of this Frankfurt Meetup! Three established meetup communities and their leaders joined forces to build an exceptional event on January 24th, hosted in the “Frankfurt School of Finance”. Over 400 data scientists (or wannabes) registrated before — and it seems like all of them did actually show up. The very spacious room in this totally new business school was crowded. What did I learn during this event? What are the main aspects to have an eye on when it comes to Artificial Intelligence? I would like to describe this in two parts: first some observations about the event itself (agenda, audience & atmosphere), secondly a kind of summary where AI is heading (automation, API & analytics).

Agenda
The agenda promised a mix of keynotes, short lightning talks and startup pitches. That is the magic of an event: it has to be diverse in terms of topics and expected skill level in order to provide learning experience to huge parts of the audience. The welcome note by the host (Prof. Gergory Wheeler from Frankfurt School of Finance) set the tone of openness for the evening. Himself teaching computer science AND theoretical philosophy prototyped the expertise to follow the AI discussion in the most relevant aspects.

Then three 20 min. talks followed:
- Building a self-learning portfolio construction system and how to train it. By Jonathan Masci — Co-Founder of NNAISSENSE https://nnaisense.com/
- API for Automated Forecasting of Crypto Time Series. By Sebastian Heinz — CEO of STATWORX https://www.statworx.com
- How AI is applied to make you save smarter . By Yassin Hankir of Savedroid https://www.savedroid.de/

Normally enough to keep an community happy. Having such huge audience in the room the possibilities to directly interact was limited. The fully packed agenda only allowed 3 to 4 questions per talk.

As an extra point on the agenda these topic-centric talks were followed by three very different startup pitches
- DOMONDA — Smart finance processes for corporates (Vienna) https://domonda.com/
- INNAAS — Artificial Intelligence-Driven Innovation (Rome) http://www.innaas.com/
- Radenbrock — Genetic evolution meets capital markets (Frankfurt) http://radenbrock.com/
So you had a good chance to grasp a lot of different angles in fintech, cryptocurrency, AI, modeling methods, real world use cases for AI et.c

Audience
These talk were well appreciated by the audience. Marvin, one of the organizers, run a traditional mentimeter-session after the first part in order to ask some questions. Only one third of the audience were students and only a small perentage are long term professionals in the field of data science. Many business consultants, some recruiters, some job seekers — a very good mix of interested people. Everybody was curious, open-minded and willing to share what they know — and what they not know or didn´t understand from the presentations.

Atmosphere
Why was such a huge interested? Several things came together. There is a new hype in the air called Artificial Intellligence. Good to see, that there is a growing interest to look behind the scenes and understand what is it all about. In Frankfurt as the banking capital of Germany the blockchain / fintech hype can be added. The three startups pitched their machine learning expertise with financial use cases. And one of the longer talks had this background as well. In this community it seems to be difficult to define an USP. So “everyone” is inventing models to predict stock markets and automatically build top performing portfolios… So open source technologies were demonstrated to serve very commercial use cases.

Automation
If I summarize what I learned about Artificial Intelligence I have to start with automation. The basic idea of AI is independecy of human interaction by letting a machine do all the necessary task to solve a problem. So instead of reading newspapers and doing maths manually the robo-advisors will automatically develop a prediction for stock prices — and will put this inteo action as well. Only the automation of decisions gives a real value. Same is true for the chat bot example of INNAAS: they explained how easy it can be for a banking customer to get more insight in his private financial situation by simply asking questions — or even by choosing a very likely question proposed by an algorithm.
In technical terms the developers dream today is to create a self-learning systems which auto-tunes itself to take automatized autonomous decisions.

APIs
Another point is that not everyone can develop prediction models on his own — but might have questions which can be solved by those. So the guys from Statworks presented their approach to provide a RESTful API to a kind of generic forecasting method. Doing this developers can make use of those models by simply calling them in a cloud, providing all necessary inputs and getting back the result. Of course this reminded my to the SAS Viya concept which exactly provides these functionality. R, Python or Java programmers are able to enrich their applications with analytics very easy in this way.

Analytics
Last observation: Its about Analytics, stupid. Its quite obvious that even lines of business should educate themselves in the basic terminology. Supervised machine learning, unsupervised machine learning, constraints in different algorithms, training data, test data, accurancy and stability of models — those topics only scratch the surface. Which method fits for which problem? How do you evaluate and trust the automated decisions by a machine? Without some background in data science it might be difficult to take any business decision in a few years.

To conclude: it was worth the visit! And a lot to learn in this field.

If you want to learn more about the status of AI in European companies I suggest this study on “The Enterprise AI promise”.

This story is published in The Startup, Medium’s largest entrepreneurship publication followed by 290,182+ people.

Subscribe to receive our top stories here.