nexgml 0.1.2

NexGML — Next Generation Machine Learning (educational ML utilities)

Nexarians - The NexGML Core Repository

Installation

pip install nexgml

🔬 Core Philosophy: Transparent, Fast, and Modular

NexGML is a custom Machine Learning utility built for educational and research purposes, emphasizing code transparency and high performance.

Key Features & Technology Stack

• JIT Acceleration: Core math operations (nexgml.helper.amo and nexgml.helper.indexing) are optimized using Numba for near C-speed performance.
• Advanced Optimization: Includes modern solvers like Adam and AdamW in classifiers.
• Modular Helpers: Separates complex logic into focused helper modules (AMO, ForTree, Indexing) for easy customization.
• Sparse Data Ready: Full support for scipy.sparse matrices (CSR/CSC) for memory efficiency.

---

💻 Available Modules & Quick Start

1. Classifiers (The Models)

The primary model is the Gradient Supported Intense Classifier (GSIC).

from nexgml.gradient_supported import IntenseClassifier
import numpy as np

# Load data X, y...

model = IntenseClassifier(
    optimizer='adamw', 
    lr_scheduler='plateau', 
    batch_size=32, 
    penalty='elasticnet'
)
model.fit(X_train, y_train)

print(f"Final Training Loss: {model.loss_history[-1]:.6f}")

2. Helper Modules (Performance Backbone)

These modules contain the high-speed math used internally.

Module	Purpose	Example Usage
nexgml.helper.amo	Advanced Math Operations. Activation/Loss functions (Softmax, CCE, RMSE).	AMO.softmax(logits)
nexgml.helper.amo.ForTree	Tree Criteria. Impurity measures (Gini, Entropy, Friedman MSE).	ForTree.gini_impurity(labels)
nexgml.helper.Indexing	Data Utilities. One-hot encoding, smart feature slicing (standard_indexing).	Indexing.standard_indexing(n_features, 'sqrt')

📝 Documentation & Exploration

This repository is dedicated to experimentation, learning, and personal research, primarily in the following fields:

• 🤖 Artificial Intelligence and Machine Learning
• 💻 Python development and performance optimization
• 📖 Technical documentation and concept notes

⚠️ This project is intended for exploration and learning purposes only.

If you find this repo helpful or interesting, feel free to fork, star, or open a pull request.
This is a learning space—no pressure, just passion! 😄