Join Christoph Molnar and Timo Freiesleben, co-authors of 'Supervised Machine Learning for Science,' as they dive deep into practical machine learning applications in research. They discuss the significance of tailoring evaluation metrics to enhance model performance and the pivotal role of domain knowledge in data collection. The duo also highlights strategies for measuring causality and improving robustness against distribution shifts. Finally, they tackle the challenges of reproducibility in science versus machine learning, offering insightful solutions.
41:58
forum Ask episode
web_stories AI Snips
view_agenda Chapters
menu_book Books
auto_awesome Transcript
info_circle Episode notes
volunteer_activism ADVICE
Choose Metrics that Reflect Goals
Choose evaluation metrics carefully as they direct all downstream modeling choices.
Design metrics to incorporate domain knowledge, like cost-specific weights, for more relevant models.
volunteer_activism ADVICE
Embed Domain Knowledge Effectively
Use data augmentation to incorporate domain knowledge by expanding datasets with known transformations.
Alternatively, encode domain knowledge as inductive biases directly into model architecture.
insights INSIGHT
Domain Knowledge is Crucial
Domain knowledge is often undervalued in data science but remains crucial for meaningful predictions.
Ignoring domain expertise causes issues in model reliability and interpretability.
Get the Snipd Podcast app to discover more snips from this episode
Part 2 of this series could have easily been renamed "AI for science: The expert’s guide to practical machine learning.” We continue our discussion with Christoph Molnar and Timo Freiesleben to look at how scientists can apply supervised machine learning techniques from the previous episode into their research.
The AI Fundamentalists 
