Last January, for the second time in a row, Artificial Intelligence experts met at our Munich offices in cooperation with Munich Datageeks Over 150 AI researchers and interested parties gathered in our CIB modul-E to take part in this exciting event. In addition to highly interesting specialist presentations, there was plenty of opportunity for intensive exchange and networking.
The presentations: Insights into innovative AI research
Our Senior Data Scientist and expert in automatic document processing, Konrad Grosser, gave the first presentation on "Massively Multimodal Input Management in CIB flow". In his talk, he showed how CIB uses sophisticated feature engineering to train a massively multimodal document classification network. Konrad introduced the use of AI expert features and explained how they can be adapted to complex customer requirements, even when only limited amounts of annotated data are available. Finally, he showed how our classification network seamlessly integrates with the CIB flow business process engine and, together with our OCR and data extraction tools, provides a comprehensive end-to-end solution for document processing.
Konrad Grosser:
"Documents are rich in linguistic and visual information and play a critical role in many business processes. Automatic understanding of such documents is an evolving area of research that requires robust and advanced multimodal approaches.
Document classification is An important aspect, which requires detailed, contextual features. These range from task-specific to "general-purpose" features.
Our guest speaker and AI researcher Adithya Mohan was also able to convince the expert audience with his presentation "Adversarial Attacks in Deep Reinforcement Learning: A Call for Robust Defenses".
Deep reinforcement learning (DRL) has shown remarkable potential in various fields such as robotics, autonomous driving, and gaming. However, its vulnerability to attack poses a significant challenge to its reliability in real-world applications. Attackers can exploit weaknesses in DRL models, resulting in suboptimal or even unsafe decisions. Adithya Mohan showed how attackers target DRL systems and uncover critical vulnerabilities that affect their performance and reliability.
The talk was well received by developers, researchers, and AI enthusiasts, and inspired the audience to come up with new ideas for developing robust DRL frameworks that can withstand adversarial threats.
At the end: CIB quiz and “networking”
The CIB quiz was held at the end of the event. As usual, this quiz included questions on the topic of artificial intelligence.
As is usual at all CIB events, there was no shortage of socializing in addition to the professional exchange.
As usual, the Munich Datageeks January Edition provided an excellent platform for knowledge exchange and collaboration within the AI community.
We look forward to the next event!
We are hiring!
CIB is constantly looking for talented experts and researchers from all fields to join the team.
Send us your spontaneous application and become part of our history!
Let´s CIB!
