https://arxiv.org/pdf/2212.07432v1.pdf

A counterfactual explanation is a type of explanation for a model's prediction that tells you the smallest changes that would need to be made to your input data to change the model's prediction.

SVMs are advantageous because they work well with high dimensional data, have theoretical guarantees on stability and sample complexity, tend to generalize well (due to low Rademacher Complexity), and they find decision boundaries (hyperplanes) that avoid overfitting (by having a low 2 norm and a wide margin). SVMs seem to perform relatively well even when there isnt alot of data.

• The wide-margin property of Support Vector Machines (SVMs) refers to their ability to not only find a hyperplane (decision boundary) that separates classes, but to do so in a way that maximizes the distance (or "margin") between the hyperplane and the closest data points from each class. These closest points are known as "support vectors", hence the name of the method.
• This generalizes really well, because this wide buffer zone allows separation of the highest degree
• Influence of Support Vectors: In SVMs, the position of the hyperplane is determined only by the support vectors (the data points closest to the hyperplane). Other points, even if they are misclassified or removed, won't affect the position of the hyperplane as long as they are not support vectors. This can make SVMs robust to outliers.

there is little to no information on how to use the wide-margin property to change the labels of instances with undesirable predictions and how this can be used to enhance interpretability

<aside> 💡 features are basically explanations in classification models

</aside>

actionability as giving a recommendation of the easiest way for a user to change an undesirable model prediction, based on a set of rules.

The case study is with diabetes dataset. The way factors interact is not understood. Controlling all factors are unrealistic. The goal is to tweak the least amount of factors the least amount.

<aside> 💡 SVMs help with some critical decisions in medical and financial fields. Therefore, interpretability and explanation must be high. This paper addresses that.

</aside>