What is data centric AI and why is it important for industrial AI?
Apr 12, 2023
auto_awesome
The podcast explores topics like data quality, robustness in AI, and the data centric approach. It also covers discussions on Amazon's dataset, interviews with AI experts like LeCun and Andrew Ng, and the importance of data in industrial AI solutions.
Iterative approach to building training datasets is crucial for enhancing industrial AI solutions.
Spotlight tool by Renomics simplifies data curation by detecting data leakage efficiently.
Focus on tangible use cases and gradual advancements in data quality are essential in evolving AI landscape.
Deep dives
Boosting Industrial AI Through Data Quality Improvement
Improving data quality is crucial for enhancing industrial AI solutions. Stefan Zübelach discusses the importance of a systematic and iterative approach to building great training datasets, leveraging insights from trained models. He emphasizes the significance of data inspection, tagging, and the use of self-supervised learning to extract valuable information.
Spotlight Solution for Multimodal Data Curation
Spotlight, an open-core data curation tool by Renomics, simplifies loading data and configuring views for inspection. With a focus on multi-modal data, the tool offers quick configuration and the ability to build interaction templates, known as recipes, to streamline data curation processes efficiently.
Addressing Data Leakage in Industrial AI
The spotlight solution helps detect data leakage, a common problem in machine learning, where testing data features samples similar to those in training data. By extracting embeddings and comparing datasets, users can flag and inspect potential leakage points using the interactive visualization capabilities of Spotlight.
Use Case in Acoustic Analysis for Engineering
Renomics delves into AI-assisted engineering applications, such as acoustic analysis for condition monitoring and test data analysis in automotive industries. By employing the Industrial AI canvas, users can leverage data curation strategies to improve model performance, providing real-time insights into machine operations.
Navigating the AI Technology Landscape in Engineering
Stefan Zübelach discusses the evolving AI landscape, highlighting the potential impact of technologies like chat GPT and generative AI. While acknowledging the transformative possibilities, caution is advised against over-hyping and ensuring a focus on tangible use cases and gradual advancements in data quality and model refinement.
Peter Seeberg met Stefan Suwelack from Renumics at the Festo IO conference and talked with him about data quality, more robustness and their data centric AI approach.
The podcast is growing and we want to keep growing. That's why our German-language podcast is now available in English. We are happy about new listeners.