Metadata-Version: 2.4
Name: blinear
Version: 0.3.2
Summary: Bilinear Neural Network Library
Author: Nguyen Minh Tuan
Author-email: Nguyen Minh Tuan <minhtuan@ptit.edu.vn>
License: MIT
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: tensorflow
Requires-Dist: numpy
Dynamic: author
Dynamic: license-file
Dynamic: requires-python

# blinear

Bilinear Neural Network Library for Deep Learning and Biomedical Signal Classification.

`blinear` provides custom bilinear neural network layers and hybrid CNN–bilinear architectures built with TensorFlow/Keras. The library focuses on nonlinear feature interaction modeling using Hirota-inspired bilinear operators for signal processing, sequence learning, and classification tasks.

---

## Features

- Hirota Bilinear Layer
- Convolutional Hirota Bilinear Layer
- OGD-based Bilinear Classification Layer
- CNN + Bilinear Hybrid Models
- Double Bilinear Architectures
- Binary and Multiclass Classification Support
- TensorFlow / Keras Compatible

---

## Installation

```bash
pip install blinear
```

## 1. Hirota CNN Model

CNN feature extraction followed by a Hirota bilinear interaction block.

### Usage

```python

from blinear import BilinearLayer

```

## 2. ConvHirota Bilinear Model

End-to-end convolutional bilinear feature extraction using ConvHirotaBilinear.

### Usage

```python
from blinear import ConvBilinear
```

## 3. Online Gradient Descent (OGD) Layer Guide

The `OGDLayer` in `blinear` is a deep bilinear representation layer designed for multiclass classification tasks.

```python
from blinear import OGDLayer
```

Requirements
------------

- Python >= 3.9
- TensorFlow >= 2.x
- NumPy

---

# License

MIT License

---

# Author

Developed for nonlinear bilinear deep learning research and biomedical AI applications.
