July 25th at 3:00pm
AI Showcases
Moderated by Marc Kolbe
In this session, companies will present AI showcases from their respective fields.
Dr. Alexandre Sawczuk da Silva, Fraunhofer Institute for Cognitive Systems IKS
"Bringing Generative AI to Manufacturing"
When we hear the term 'generative AI', many of us automatically picture the automated creation of images, videos, and texts. There are multiple approaches that fall under the generative AI umbrella, as well as many scenarios in industrial automation where they can be used. In this talk we show you how Large Language Models and evolutionary computation, two different generative approaches, can be used to make manufacturing more flexible. We explore ideas such as an assistant for understanding the state of the shop floor, an optimizer for scheduling small production batches, and a code generator for dynamically changing production controls. The examples show that the work in this area can unlock solutions that lower costs and increase efficiency.
Dr. Gereon Weiß, Fraunhofer Institute for Cognitive Systems IKS
"safe.trAIn - Safe AI using the example of driverless regional trains"
Fully automated rail transport with the highest possible degree of automation is a crucial building block for a climate-neutral and attractive transport mix. According to the state of the art, conventional automation technology will not be sufficient for fully automated rail operations. Artificial intelligence, on the other hand, holds great potential for this. The aim of the safe.trAIn project is therefore to lay the foundations for the safe use of AI for driverless rail vehicles in order to make regional rail transport more efficient and sustainable. In contrast to existing driverless rail transport, which is used exclusively in closed and controlled environments, regional transport is a more open environment in which obstacles such as people in the track or fallen trees must be detected safely and reliably. To this end, the safe.trAIn project is developing test standards and methods for the use of AI in the automation of rail transport. The suitability of the test standards is also being investigated using an application example, the driverless regional train.
Leonid Mill, Mira Vision
"MIRA Vision - Photo-realistic synthetic data for biomedical image analysis in microscopy"
The field of biomedical image analysis has been revolutionized by the application of artificial intelligence (AI), particularly in the domain of microscopy and digital pathology. AI has the potential to greatly enhance the analysis and interpretation of microscopic images, leading to advancements in disease diagnosis, drug discovery, and biological research. However, to fully unleash the potential of AI-based image analysis and make it available throughout the diverse applications within microscopy, several challenges have to be overcome – from data acquisition to regulatory aspects. In this context, one of the most important challenges is the availability and acquisition of representative, high-quality and well-annotated datasets for training AI models. Obtaining such datasets is in general a very time-consuming, costly and error-prone process which requires profound expert knowledge to ensure accurate annotations.
In this talk, we demonstrate our approach to this fundamental problem: an innovation that allows us to significantly decrease AI-development time and costs by creating parametric, interpretable, and photo-realistic synthetic AI training data, without using and having the disadvantages of image-based generative AI. We show the potential of our technology for various applications ranging from digital pathology, infectious diseases to cancer research. To this end, we present the MIRA AI Platform, a web-based solution for intuitive, collaborative and AI-assisted image analysis for researchers and pathologists.
Dr. Alexandra Mikityuk, Staex
"Trusted networks for AI"
In this talk, we will explore the concept of web-trusted networks for AI, focusing on creating secure, decentralized environments for AI systems to operate. We'll discuss the integration of blockchain technology to enhance data integrity, privacy, and trust within AI networks. The session will cover key use cases, including the role of decentralized systems in preventing data tampering and ensuring reliable AI outputs. Attendees will gain insights into how these trusted networks can revolutionize AI deployment across various industries, promoting transparency and accountability.
Flora Geske, SUMM AI
54% of people read below a 6th-grade level and are overwhelmed by complicated texts. Thus, they need easy-to-read language to navigate their daily lives. That's why the team of SUMM AI has developed the "Google Translate" for easy language: The very first AI-powered tool that automatically translates any complicated text into easy language. Easy language is a defined style of language rooted in the accessibility space. It uses short sentences, simple choice of words, and additional explanations to make text accessible and easy to understand for everyone - particularly for people who are excluded from our complicated everyday language due to learning difficulties, educational disadvantages, or because they learn English as a foreign language. Thus, SUMM AI supports public institutions and companies to provide accessible content and reach a new target audience with the click of one button! More Info: www.summ-ai.com