Deep Fast Machine Learning Utils - A Python Library for Streamlined Machine Learning Prototyping

Aug 2, 2026

Upload by Eden

Machine learning research and application often require time-intensive efforts in model architecture prototyping, feature selection, and dataset preparation. Deep Fast Machine Learning Utils (DFMLU) is introduced as a Python library that automates and improves parts of this workflow. Compatible with TensorFlow, Keras, and scikit-learn, it supports dense neural network search via PCCDNAS, advanced feature selection, and tools for dataset management and visualization of training outcomes, accompanied by Python examples.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Introduction

Methods

Principal Component Cascade Dense Neural Architecture Search (PCCDNAS)

Implementation Example

arXiv :2409 .09537v1 [ cs .LG] 14 Sep 2024 Deep Fast Machine Learning Utils: A Python Library for Streamlined Machine Learning Prototyping Fabi Prezja 1,2,* 1 University of Jyvskyl, Faculty of Information Technology, Jyvskyl, Finland 2 Finnish Artificial Intelligence Research Network, Jyvskyl, Finland * [corresponding.faprezja@jyu.fi](corresponding.faprezja@jyu.fi) ABSTRACT Machine learning (ML) research and application often involve time-consuming steps such as model architecture prototyping, feature selection, and dataset preparation. To support these tasks, we introduce the Deep Fast Machine Learning Utils (DFMLU) library, which provides tools designed to automate and enhance aspects of these processes. Compatible with frameworks like TensorFlow, Keras, and Scikit-learn, DFMLU offers functionalities that support model development and data handling. The library includes methods for dense neural network search, advanced feature selection, and utilities for data management and visualization of training outcomes. This manuscript presents an overview of DFMLU’s functionalities, providing Python examples for each tool. Introduction Machine Learning (ML) has steadily evolved, creating numerous tools and frameworks that support model development and deployment. Within the field of artificial intelligence, deep learning 1 stands as a significant subset of ML that typically focuses on deep neural networks and big data. Deep architectures can enable learning complex patterns and representations from large data, allowing for developments in areas such as Biomedical computer vision2–14, and natural language processing 15–27. Well-established libraries, including PyTorch7 , TensorFlow28 , Keras29 , and Scikit-learn30 , offer dependable platforms for building a variety of ML and deep learning models. As model architectures increase in complexity and datasets grow, tasks such as feature selection, architecture design, and data preparation may become more resource-intensive. The Deep Fast Machine Learning Utils (DFMLU) library is a Python-based utility collection designed to complement existing ML libraries by offering additional tools that simplify aspects of these tasks. DFMLU emphasizes efficiency in the ML workflow by partly automating processes such as dense neural network search, advanced feature selection, dataset management, and performance tracking. Methods This section outlines the DFMLU library’s key tools, focusing on model architecture search and advanced feature selection. Principal Component Cascade Dense Neural Architecture Search (PCCDNAS) PCCDNAS provides an automated method for designing dense neural networks. This approach uses PCA (Principal Component Analysis), which systematically sets the number of neurons in each network layer. After applying PCA to the initial data, the neuron count for the first layer is determined based on the principal component counts (PCs) for a given variance threshold. Subsequently, the cascade mechanism ensures that the activations from a trained layer undergo PCA again. This process, in turn, determines the neuron count for the subsequent layers using the same principal component variance threshold criteria. PCCDNAS may reduce the need for manual dense architecture tuning in this context. While a more comprehensive description of the PCCDNAS method will be presented in future research, it can be viewed as a specialized case of additive auto-encoding31 , while both works were developed independently. PCCDNAS Pseudo-Algorithm: 1 2 3 4 5 6 7 \# Pseudo-Algorithm for PCCDNAS : 1 . Initialize: -Create an empty neural network model . -Create an empty list to store the number of neurons for each layer . 2 . Data Initialization: 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | -Accept training data and labels .
-Center or normalize the data if required .

3 . Initialize Model Search: | | | | --- | ---

FAQs

What problem does the DFMLU library address in machine learning workflows?

DFMLU targets time-consuming steps such as model architecture prototyping, feature selection, and dataset preparation by providing utilities that automate and enhance these processes.

How does PCCDNAS design dense neural networks?

PCCDNAS uses PCA with a variance threshold to set neuron counts per layer, then applies a cascade mechanism where activations from a trained layer are again analyzed with PCA to determine subsequent layers.

Which frameworks is DFMLU compatible with?

DFMLU is presented as compatible with TensorFlow, Keras, and scikit-learn, supporting model development and data handling.

Aug 2, 2026

Upload by Eden