vopy.datasets

Datasets for maximization problems.

References:

[Liao2008]

Liao, Li, Yang, Zhang, Li. Multiobjective optimization for crash safety design of vehicles using stepwise regression model. Structural and Multidisciplinary Optimization, 2008.

[Tanabe2020] (1,2)

Tanabe, Ishibuchi. An easy-to-use real-world multi-objective optimization problem suite. Applied Soft Computing, 2020.

[Zuluaga2012]

Zuluaga, Milder, Püschel. Computer generation of streaming sorting networks. Design Automation Conference, 2012.

[Grari2024]

Grari, Laugel, Hashimoto, Lamprier, Detyniecki. On the Fairness ROAD: Robust Optimization for Adversarial Debiasing. International Conference on Learning Representations, 2024.

class vopy.datasets.dataset.Dataset

Abstract base class for datasets that handles min-max scaling of input and standardization of output. Any class inheriting from this class should implement the following attributes:

  • _in_dim: int

  • _out_dim: int

  • _cardinality: int

class vopy.datasets.dataset.DiscBrake

Disc brake optimization balancing mass and stopping time. Based on [Tanabe2020]. Both objectives are negated.

  • _in_dim = 4

  • _out_dim = 2

  • _cardinality = 128

class vopy.datasets.dataset.Fairness

Dataset for optimizing fairness in machine learning. Dataset consists of possible hyperparameter selections of ROAD approach on a neural network architecture. The reward vector represents the trade-off between global disparate impact and global accuracy. See [Grari2024]. Disparate impact is negated.

  • _in_dim = 2

  • _out_dim = 2

  • _cardinality = 200

class vopy.datasets.dataset.SNW

Dataset for optimizing sorting network configurations in computational hardware design. The reward vector represents the trade-off between throughput and hardware area. The area is negated to maximize it. See [Zuluaga2012].

  • _in_dim = 3

  • _out_dim = 2

  • _cardinality = 206

class vopy.datasets.dataset.Test

A miniature DiscBrake dataset variant for using in testing.

  • _in_dim = 4

  • _out_dim = 2

  • _cardinality = 32

class vopy.datasets.dataset.VehicleSafety

Vehicle structure optimization dataset for enhancing crashworthiness. The reward vector includes weight, acceleration, and toe-board intrusion. See [Liao2008] and [Tanabe2020].

  • _in_dim = 5

  • _out_dim = 3

  • _cardinality = 500

vopy.datasets.dataset.get_dataset_instance(dataset_name: str) Dataset

Returns an instance of the dataset class corresponding to the given dataset name. If the dataset name is not recognized, a ValueError is raised.

Parameters:

dataset_name (str) – Name of the dataset class to be instantiated.

Returns:

Instance of the dataset class.

Return type:

Dataset