Deep Learning: After ChatGPT - Self-reflecting machines || Segment Anything

Hi everyone,

please join us for our next Deep Learning meetup on June 15, 2023, hosted by FH Technikum Wien!

We will have Matthias Samwald (MedUni Wien) presenting After ChatGPT: The dawn of machines that self-reflect and self-direct and René Donner (mva.ai) presenting Meta's Segment Anything - Architecture and Use Cases.
*Note that this meetup is limited to 85 people attending. We will cap the registrations at 140 people assuming the usual show-up rate of 60%.*

***

Agenda

18:30

• Introduction by the VDLM organizers
• Welcome by Sharwin Rezagholi and Christoph Henrichs from FH Technikum Wien and Technikum Wien Academy

18:40

• After ChatGPT: The dawn of machines that self-reflect and self-direct
  Matthias Samwald, MedUni Wien
  Large language models demonstrate outstanding capabilities, but significant challenges remain regarding their ability to solve complex reasoning tasks, as well as their transparency, robustness, truthfulness and ethical alignment. In this talk, I outline how different strands of research and development—from structured reasoning, evaluation, explainable AI, systems security and ethics—will have to converge to yield powerful new machines that self-direct and self-reflect.

:: Announcements & Job Openings ::
:: Break & Networking ::

20:10

• Segment Anything - Architecture and Use Cases
  *René Donner, mva.ai*
  Meta's recent release of Segment Anything has raised the bar for zero-shot segmentation frameworks. Trained on 11 million images and 1.1 billion masks it provides high-quality segmentation masks for subsequent tasks and it can also be prompted to yield different types of segmentations. We will look into the architecture of the system, the different types of tasks it help with and how to use it in practice.

:: Networking ::

21:30 Wrap up & End