tlmnet Examples Gallery

This gallery provides practitioner-focused examples demonstrating how to leverage TlmMilpClassifier for exact ternary linear modeling, strict \(L_0\) feature cardinality budgeting, and sparse text classification.

Together, these scripts illustrate how exact Mixed-Integer Linear Programming (MILP) replaces greedy heuristic approximations, yielding globally optimal integer-weighted models compliant with Scikit-Learn 1.6+ workflows.

Gallery Highlights

• Exact L0 Feature Selection Budgeting: Demonstrates how to enforce hard integer cardinality constraints using the max_features parameter on the Breast Cancer dataset to build ultra-sparse, highly interpretable scorecards. (See: Exact L0 Sparsity Budgeting on Breast Cancer Dataset)

• Multiclass One-vs-Rest Text Classification: Demonstrates how TlmMilpClassifier integrates seamlessly with Scikit-Learn’s OneVsRestClassifier to perform exact multi-class document classification. (See: Multilabel Text Document Classification with Exact MILP One-vs-Rest)

• Runtime Scaling & Solver Performance: Quantifies branch-and-bound solve times across varying feature dimensions (\(p\)) and sample sizes (\(n\)). (See: MILP Solve Time vs. Feature & Sample Dimensions)

• Decision Boundaries & Surface Visualization: Visualizes the discrete, quantized decision surfaces created by ternary weight vectors \(w_j \in \{-1, 0, 1\}\) on 2D feature spaces. (See: Decision Boundaries & Margin Scores)

• Sparse Text Matrix Classification: Shows how to pair TfidfVectorizer natively with TlmMilpClassifier using sparse SciPy block matrices. (See: Sparse Text Classification with Vocabulary Limits)

MILP Solve Time vs. Feature & Sample Dimensions

MILP Solve Time vs. Feature & Sample Dimensions

Decision Boundaries & Margin Scores

Decision Boundaries & Margin Scores

Sparse Text Classification with Vocabulary Limits

Sparse Text Classification with Vocabulary Limits

Exact L0 Sparsity Budgeting on Breast Cancer Dataset

Exact L0 Sparsity Budgeting on Breast Cancer Dataset