Machine Learning Students Overfit to Overfitting

Citations Over Time

Abstract

Overfitting and generalization is an important concept in Machine Learning as only models that generalize are interesting for general applications. Yet some students have trouble learning this important concept through lectures and exercises. In this paper we describe common examples of students misunderstanding overfitting, and provide recommendations for possible solutions. We cover student misconceptions about overfitting, about solutions to overfitting, and implementation mistakes that are commonly confused with overfitting issues. We expect that our paper can contribute to improving student understanding and lectures about this important topic.

Related Papers

• → Physician-Friendly Machine Learning: A Case Study with Cardiovascular Disease Risk Prediction(2019)71 cited
• → An Optimal Solution to the Overfitting and Underfitting Problem of Healthcare Machine Learning Models(2023)15 cited
• → Hybrid ensemble learning approach for generation of classification rules(2015)11 cited
• → Breakdown of Machine Learning Algorithms(2022)1 cited
• Measuring Generalization and Overfitting in Machine Learning(2019)