A team has developed a new method that facilitates and improves predictions of tabular data, especially for small data sets with fewer than 10,000 data points. The new AI model TabPFN is trained on synthetically generated data before it is used and thus learns to evaluate possible causal relationships and use them for predictions.
↧