Industrial AI and AutoML - 2 days "AI in the forest"
Nov 1, 2023
auto_awesome
Prof. Dr. Frank Hutter and Prof. Dr. Marco Huber discuss democratizing machine learning, benefits of AutoML, limitations and solutions for Time Series data, exploring AutoML tools and AI augmented R&D, and future plans for upcoming events.
AutoML simplifies the development process and provides new ideas for data scientists, while also bridging the gap between domain experts and traditional data scientists.
TAP-PFN shows potential for processes and production planning, with a need for more tailored solutions for time series data in the industrial sector.
Deep dives
AutoML as a tool for inspiration and new approaches
During the podcast, the concept of AutoML was discussed, focusing on its potential as a tool for inspiration and the generation of new approaches in data science. The participants explored how AutoML can be used to simplify the development process, standardize and document development processes, and provide new ideas for data scientists. The discussion also highlighted the importance of making AutoML more accessible to decision-makers and non-technical users, allowing them to experiment and play with models. The potential for AutoML to bridge the gap between domain experts and traditional data scientists was also emphasized.
TAP-PFN for tabular data sets and limitations
Another topic discussed in the podcast was TAP-PFN, a trained transformer capable of supervised classification on small tabular data sets. The participants acknowledged the limitations of TAP-PFN, particularly its current limitations in dealing with large data sets. However, there was interest in exploring how TAP-PFN could be used for processes and production planning, and the exchange of ideas between the participants highlighted the potential for further development in this area. The conversation also touched on the need for more solutions and tools specifically tailored to time series data in the industrial sector.
AI Augmented R&D and patent research
The podcast also featured a discussion on AI Augmented R&D, a tool developed by Professor Dennis Kavalucci from Strasbourg, which combines large language models with patent research capabilities. The participants expressed interest in this tool, as it offers a deep learning-based approach to analyzing patents, scientific papers, and identifying potential opportunities and challenges in the engineering and design engineering fields. The tool was seen as a valuable resource for individuals involved in patent research, providing a streamlined way to navigate complex patent landscapes and generate new ideas for research and development.
Together with 30 decision-makers from industry, we spent two days discussing AutoML and TabPFN with Prof. Dr. Frank Hutter and his team and Prof. Dr. Marco Huber and discovered quite a few new approaches.
Thanks for listening. We welcome suggestions for topics, criticism and a few stars on Apple, Spotify and Co.