Jan Pablo Burgard – Optimization Online

Mixed-Integer Quadratic Optimization and Iterative Clustering Techniques for Semi-Supervised Support Vector Machines

Published: 2023/03/22, Updated: 2023/10/02

(Mixed) Integer Nonlinear Programming, Data Science Algorithms, Data Science Applications clustering, mixed-integer quadratic optimization, Semi-Supervised Learning, support vector machines

Among the most famous algorithms for solving classification problems are support vector machines (SVMs), which find a separating hyperplane for a set of labeled data points. In some applications, however, labels are only available for a subset of points. Furthermore, this subset can be non-representative, e.g., due to self-selection in a survey. Semi-supervised SVMs tackle … Read more

Mixed-Integer Programming Techniques for the Minimum Sum-of-Squares Clustering Problem

Published: 2022/03/09, Updated: 2022/11/29

(Mixed) Integer Nonlinear Programming, Cutting Plane Approaches, Data-Mining Computational Techniques, global optimization, Minimum Sum-of-Squares Clustering, Mixed-integer nonlinear optimization

The minimum sum-of-squares clustering problem is a very important problem in data mining and machine learning with very many applications in, e.g., medicine or social sciences. However, it is known to be NP-hard in all relevant cases and to be notoriously hard to be solved to global optimality in practice. In this paper, we develop … Read more

Robustification of the k-Means Clustering Problem and Tailored Decomposition Methods: When More Conservative Means More Accurate

Published: 2020/05/17, Updated: 2022/04/27

Jan Pablo Burgard

Carina Moreira Costa

Martin Schmidt

(Mixed) Integer Nonlinear Programming, Robust Optimization alternating direction method, gamma-robustness, k-means clustering, robust optimization, strict robustness

k-means clustering is a classic method of unsupervised learning with the aim of partitioning a given number of measurements into k clusters. In many modern applications, however, this approach suffers from unstructured measurement errors because the k-means clustering result then represents a clustering of the erroneous measurements instead of retrieving the true underlying clustering structure. … Read more

Exact solution of the donor-limited nearest neighbor hot deck imputation problem

Published: 2019/08/20

Graphs and Matroids, Network Optimization, Statistics b-matching, combinatorial optimization, missing data, nearest neighbor hot deck imputation, survey data

Data quality in population surveys suffers from missing responses. We use combinatorial optimization to create a complete and coherent data set. The methods are based on the widespread nearest neighbor hot deck imputation method that replaces the missing values with observed values from a close unit, the so-called donor. As a repeated use of donors … Read more