Speakers Corner LIVE
The first fully live video event in the Barrel. Recorded and broadcasted by Joakim Jardenberg from Get AI. New times calls for new measures. Today we could offer 5 presentations from a variety of areas within machine learning. This was also our first physical event since the Corona outbreak, and in the middle of the summer. But the interest for tis format was so high that we already have speakers to fill another similar event later this year.
Markus Borg started out speaking about testing of AI applications and how they differ from testing regular applications. We are certainly curious to hear more as time goes and what becomes of his initiative AIQ Meta-Testbed.
Our second speaker, Isabella Gagner talked about her master's thesis focused on calculating the number of people in a room based on sound alone. Here we heard many lessons learned from struggling to collect training data and keeping the memory of the model at a minimum to be able to run it in the local device, on the edge...
... Edge AI was something that our third speaker David Fu had more to tell us about. Many aspects of compute need to be done in the edges, on small devices. Mainly for to reasons, the first being the latency issue of processing data in the cloud for real time system and second to ensure personal data does not end up on servers with the risk of conflicting with GDPR. Then we learned more about image recognition using the neural network YOLO.
In the last of the technical presentations Jacob Tyrberg presented his journey on extracting text information from old product specifications at Alfa Laval, to make it available and searchable in a database. This was a case with a surprisingly good result with a very simple model on a small sample of hand labelled data. And just so you know, as a machine learning expert you might end up hand labelling data by drawing bounding boxes for three days.
Finally Malin Larsson from AI Sweden shared how they work on trying to connect businesses with machine learning experts, to be an engine that build ecosystems that will accelerate innovation within AI in Sweden. To cross pollinate solutions from academia with partners from industry to be able to push the front line of AI technology and usage of these technologies within businesses and the public sector in Sweden.
See the entire video at https://youtu.be/sb2G0cj8SIs
- sound recognition, image recognition, Edge AI, Text analysis, AI Sweden