1. AI Carinthia Data Talk am 04.11.2020

Wir haben unsere Vorbereitungen abgeschlossen und starten unsere nächste Aktivität im Rahmen von „AI Carinthia“!

Am 04.11.2020 um 17:00 organisieren wir den 1. AI Carinthia Data Talk als Online-Event.

Für die Auftaktveranstaltung stehen vier Themen zur Auswahl. Im Folgenden findest du eine Beschreibung der Themen sowie die Möglichkeit an der Abstimmung teilzunehmen. Das ausgewählte Thema wird dann mit der Kalender-Einladung (in der auch der Einwahl-Link für das Online-Event zu finden sein wird) kommuniziert.

Die Themen:

1. Intellectual Property in ML
Bruno Hautzenberger (Anexia)In this Talk I will discuss the the pros and cons of recent events that show that the field of AI/ML, that is massively driven by open knowledge exchange and community driven innovation, is becoming more and more controlled by big companies that do not always keep everything as open as it was.As an example I will discuss the recent case of OpenAI’s GPT-3 language model, which is now controlled exclusively by Microsoft and what such developments mean for the AI/ML community and also companies the want to use AI/ML solutions in their business.We talk about money, being dependent on the „big players“ and how this might also affect research. 

2. Applied Computer Vision in Semiconductor Industry
Corinna Kofler & Dženana Alagić (KAI GmbH)

In our talk, we will present two examples from the semiconductor industry, where the application of computer vision methods are the key enablers for a good solution. In the first example, we will show how to automate the classification of defect images taken during the semiconductor manufacturing process by applying deep learning methods. The main challenge at this task is the highly unbalanced image set, which we tackle by applying oversampling, data augmentation, and label smoothing. The goal of the second example is to automatically extract microstructure information out of images from a polycrystalline material with the final aim to analyze material changes over time. To this end, we used classical Computer Vision techniques to identify damage patterns together with segmentation algorithms to extract grain information.

3. Common Problems with Artificial Intelligence and Big Data
Torsten Ullrich (Fraunhofer Austria Research GmbH)

More and more data is being collected and processed. Due to the media presence of big data, artificial intelligence, extensive data analysis and related success stories, there is a certain pressure on all stakeholders to collect and analyze data as well. However, this pressure also leads to errors: This talk will illustrate two common problems in data analysis that may appear in any application domain and that may develop into a cause of failure.

4. Algorithmen und ihre Gefahren
Eva Eggeling (Fraunhofer Austria – KI4LIFE)

This talk is about raising awareness of automated discrimination.

Hier geht es zur Abstimmung des Themas für das erste Online-Treffen: Mein Themenwunsch

Und hier kannst du dich für die Veranstaltung anmelden:


Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert