Steigungsprüfungen

Steigungsprüfungen, also known as gradient tests or slope tests, are a type of assessment used to evaluate the performance of a machine learning model, particularly in classification tasks. They are designed to check how sensitive a model's predictions are to small changes in the input features. The core idea is to perturb the input data slightly and observe the resulting change in the model's output, often the predicted class or probability.

In a typical Steigungsprüfung, a small amount of noise or a minor modification is added to the input features of a data sample. The model then makes a prediction on this altered sample. The difference between the prediction on the original sample and the prediction on the perturbed sample indicates the model's stability. A model that is highly sensitive to small changes might exhibit drastic shifts in its predictions, suggesting it may be overfitting to the training data or is not robust. Conversely, a model that shows little change in its predictions under small perturbations is considered more stable and potentially more reliable.

These tests are valuable for diagnosing issues such as overfitting, understanding the model's decision boundaries, and ensuring that the model's predictions are not overly influenced by spurious correlations in the data. While not a primary evaluation metric like accuracy or F1-score, Steigungsprüfungen provide a complementary perspective on model behavior, especially in safety-critical applications where robustness is paramount. They can help identify areas where a model might be unreliable in real-world scenarios where data is rarely perfectly clean.