LassoBench Benchmarks

The LassoBench collection provides high-dimensional hyperparameter optimization benchmarks based on Weighted Lasso regression.

More information about the functions is available in the LassoBench documentation.

Available Problems

  • bocode.LassoBench.LassoBreastCancer

  • bocode.LassoBench.LassoDiabetes

  • bocode.LassoBench.LassoDNA

  • bocode.LassoBench.LassoLeukemia

  • bocode.LassoBench.LassoRCV1

  • bocode.LassoBench.LassoSyntHard

  • bocode.LassoBench.LassoSyntHigh

  • bocode.LassoBench.LassoSyntMedium

  • bocode.LassoBench.LassoSyntSimple

Example Usage

import bocode
import torch

# Create a LassoBench problem
problem = bocode.LassoBench.LassoBreastCancer()

# Get problem information
bounds = problem.bounds
optimum_function_value = problem.optimum
optimum_input_value = problem.x_opt

# Evaluate at a point
x = torch.Tensor([[0.0] * problem.dim])
values, constraints = problem.evaluate(x)

print(f"Lasso Breast Cancer function value at origin: {values[0]}")

Output:

Lasso Breast Cancer function value at origin: tensor([-0.2626])