Whitepaper Unlock Complex Corporate Data With Generative AI

Related Vendors

Fraunhofer-Institut für Intelligente Analyse- und Informationssysteme IAIS

FAULHABER Drive Systems

Wuxi InfiMotion Technology Co., Ltd.

From matrices and time series to the processing of multimodal data in robotics: In the new whitepaper "Beyond Text and Image—Generative AI for Diverse Data Worlds," institutes of the Fraunhofer Alliance for Big Data and Artificial Intelligence provide an overview of new approaches to unlocking complex corporate data using GenAI.

Within a very short time, generative artificial intelligence has developed into a key technology in the digital world. From private use to applications in businesses, the possibilities are diverse—and far from exhausted. Initially made popular through text and image generation, more and more multimodal or GenAI models specialized in complex corporate data are now emerging. Whether it’s measurement data series in production, molecular structure data in pharmaceuticals and chemistry, or spreadsheets in the financial sector, generative AI increasingly enables companies to leverage more complex data formats. The new whitepaper by the Fraunhofer Big Data AI Alliance illustrates, using concrete use cases, how this data can contribute to process optimization and automation. The publication is available for free download.

Forecasts With Minimal Manual Effort

In five main chapters, Fraunhofer experts inspire companies from various industries to harness the potential of generative AI for their own often highly specialized data environments. In addition to providing an overview of technical fundamentals, the whitepaper illustrates the application of GenAI using different data structures, namely: tabular data, time series data, graphs, digital twins, and multimodal GenAI models in robotics.

GenAI Optimizes Time Series

One chapter focuses on time series: ranging from simple sensor measurements to motion trajectories, video data, and process flows. The whitepaper highlights the immense potential of context-aware modeling and analysis. "Whether in the analysis of stock prices, monitoring patient parameters, or production downtimes, time series are essential for many industries. Unlike traditional analysis methods, generative AI often provides more accurate forecasts," explains Dr. Sonja Holl-Supra, Managing Director of the Fraunhofer Big Data AI Alliance. "GenAI also opens up entirely new possibilities for time series. When data quality or quantity is lacking, GenAI can fill gaps, generate synthetic data, or extend a time series for different scenarios. The great thing about it: GenAI offers low-threshold access—regardless of individual analytical expertise."

Multimodal GenAI Models in Robotics

Another chapter, for example, focuses on the use of multimodal GenAI models in robotics, particularly in the control of machines and robots in dynamic environments. These models enable robots to link natural language with various data sources such as images, audio, and sensor data, allowing them to better perceive their surroundings and independently and efficiently execute complex tasks—autonomously, without the need for extensive manual programming. Through their context-aware understanding, they can distinguish between objects and directly apply this knowledge to their behavior. A shared multimodal embedding space ensures that different data types are merged into a unified structure, consistently evaluated, and translated into targeted actions.

Secure and Compliant Authentication in Laboratories

Secure & Contactless User Authentication at GMP/GLP