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.

What problem does this paper attempt to address?

This paper attempts to solve some time - consuming tasks in machine learning (ML) research and applications, such as model architecture prototyping, feature selection, and dataset preparation. Specifically, these problems include: 1. **Model Architecture Prototyping**: As the complexity of model architectures and the scale of datasets increase, manually designing and adjusting neural network structures becomes increasingly time - consuming and difficult. 2. **Feature Selection**: Selecting features that are most helpful to model performance from a large number of features is a complex and important task, especially in the case of high - dimensional data. 3. **Dataset Preparation**: It includes steps such as data splitting, subsampling, and visualization, and these tasks are especially cumbersome when dealing with large - scale datasets. To solve the above problems, the paper introduces a Python library named **Deep Fast Machine Learning Utils (DFMLU)**. DFMLU provides a series of tools to automate and enhance these processes, thereby improving the efficiency of the machine - learning workflow. Specific functions include: - **Dense Neural Network Search**: Automatically design dense neural networks through the Principal Component Cascade Dense Neural Architecture Search (PCCDNAS) method. - **Advanced Feature Selection**: Provides multiple feature selection methods, such as Adaptive Variance Threshold (AVT), Rank Aggregated Feature Selection (RAFS), and Chained Feature Selection (ChainedFS). - **Data Management Tools**: Includes a Dataset Splitter and a Data Sub Sampler, which are used to simplify dataset splitting and subsampling. - **Training Result Visualization**: Provides the functions of plotting validation curves (plot_history_curves) and generating confusion matrices (plot_confusion_matrix) to help diagnose model performance. Through these tools, DFMLU aims to simplify and accelerate the development and optimization process of machine - learning models, enabling researchers and developers to conduct experiments and debugging more efficiently.

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

Deep Fast Vision: A Python Library for Accelerated Deep Transfer Learning Vision Prototyping

Machine Learning Made Easy (MLme): a comprehensive toolkit for machine learning–driven data analysis

Up to 100x Faster Data-Free Knowledge Distillation

STREAMLINE: A Simple, Transparent, End-To-End Automated Machine Learning Pipeline Facilitating Data Analysis and Algorithm Comparison

Deep PLS: A Lightweight Deep Learning Model for Interpretable and Efficient Data Analytics

Julearn: an easy-to-use library for leakage-free evaluation and inspection of ML models

A Data-Centric Optimization Framework for Machine Learning

DLSIA: Deep Learning for Scientific Image Analysis

Fast Yet Effective Machine Unlearning

nuts-flow/ml: data pre-processing for deep learning

Automated Deep Learning Optimization via DSL-Based Source Code Transformation

DistML.js: Installation-free Distributed Deep Learning Framework for Web Browsers

Deepfacelab: Integrated, Flexible and Extensible Face-Swapping Framework.

DLBench: a comprehensive experimental evaluation of deep learning frameworks

Deep Learning, Machine Learning -- Digital Signal and Image Processing: From Theory to Application

DriveML: An R Package for Driverless Machine Learning

MLOps: Automatic, Zero-Touch and Reusable Machine Learning Training and Serving Pipelines

Deep Factorized Metric Learning

Deep Learning and Machine Learning -- Python Data Structures and Mathematics Fundamental: From Theory to Practice