TINTOlib: A Python Library for Transforming Tabular Data into Synthetic Images for Deep Neural Networks - Examples

Release Title:

TINTOlib v1.0.6.1: Official Release Supporting Tabular-to-Image Transformations for Machine Learning

Release Description:

We are pleased to announce the first official release of TINTOlib, a Python library designed for transforming tabular data into synthetic images, facilitating the integration of tabular data with state-of-the-art vision-based machine learning models. This release accompanies the research article currently under review, which validates the library's methods and highlights its contributions to advancing hybrid neural network architectures.

Key Features:

• Comprehensive Transformation Methods: Implements cutting-edge techniques such as IGTD, SuperTML, REFINED, and others for converting tabular data into synthetic images.
• Hybrid Neural Network Support: Optimized for architectures combining Convolutional Neural Networks (CNNs) or Vision Transformers (ViTs) with Multi-Layer Perceptrons (MLPs), delivering superior results for regression and classification tasks.
• Cross-Platform Compatibility: Compatible with Linux, macOS, and Windows, supporting Python 3.7 and higher.
• Easy Integration: Seamless support for popular frameworks like TensorFlow and PyTorch, enabling effortless experimentation and deployment.

Included Materials:

• Datasets: Preprocessed datasets for regression and classification tasks.
• Notebooks: Practical examples for applying TINTOlib in machine learning pipelines, covering TensorFlow and PyTorch implementations.
• Presentations: Detailed visual material explaining TINTOlib's functionality and its application in hybrid architectures.
• Logs: Comprehensive experimental results for regression and classification tasks, including detailed metrics for all evaluated models.

Installation:

pip install TINTOlib

For methods with additional requirements, such as REFINED or SuperTML, refer to the detailed installation instructions provided in the README.

Additional Resources:

• Official Documentation
• GitHub Repository.
• TINTOlib Crash Course.

This release represents a significant step forward in leveraging synthetic image generation for advanced machine learning, offering a robust framework validated through rigorous experimentation and scientific review.