AI in Action - Pre-event for Rise of AI

This is a past event

81 people went

Location image of event venue

Details

A joint event with the AI in Action Berlin Meetup Group (https://www.meetup.com/AI-x-Mobility-Berlin/events/260666597/) on the eve of the Rise of AI Conference (https://riseof.ai).

This event is open to all, not just conference attendees.

Join for talks and a panel discussion with industry leaders who are applying AI to the field of mobility.

Our speakers are:
Sergey Burkov & Sophie Chauvin from Mobimeo
Jan Christoph Jahne from Deutsche Bahn
Matti Lyra from Comtravo

Talk #1
Sergey Burkov & Sophie Chauvin from Mobimeo
Topic: "ML on the edge"

Abstract:
The challenges of running ML models on mobile phones.
Discuss the pros/cons of running ML in the cloud vs on the user device.
Discuss when it really makes sense to have on-device inference.
What the limitations and pitfalls of this approach are.
Share some advice on how to optimize ML models to accommodate device specifics.
Discuss data challenges and general lessons learned.

Talk #2
Jan Christoph Jahne from Deutsche Bahn

Abstract:
“During the last decade the number of passengers in Germany’s rail environment has been steadily growing, reaching 144 million passengers in 2017. The increasing demand on capacity requires new methods for the planning and efficient management of trains and infrastructure. This project presents an approach that aims at solving future dispatching decisions by Reinforcement Learning”

Talk #3
Matti Lyra from Comtravo

Abstract:
During the last couple of years a lot of progress has been made in a number of NLP tasks. This progress is largely due to new deep learning methods that utilise ever larger data sets. Translating the research progress into new and improved products in industry is not always straightforward; there is often no direct link between an academic NLP task and some automation task in industry. In order to utilise the progress in academia, research engineers need to dissect the automation tasks into smaller pieces that can reasonably be mapped to the ones in academia. Furthermore, business requirements dictate constraints within which a machine learning system, or systems, must operate. These constraints are often missing in academic work.

In this talk Matti will use the example of an end to end automation pipeline for travel bookings to show how different modelling decisions impact the overall performance of the pipeline and talk about human-machine hybrids that together fulfill the required performance.

Sponsored by:
- Alldus (https://alldus.com)
- YND (https://ynd.co)
- Cleanride (https://www.cleanride.io)

Hosted by Omio (https://www.omio.com)