# deep-learning

**Repository Path**: refone/deep-learning

## Basic Information

- **Project Name**: deep-learning
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-08-17
- **Last Updated**: 2026-08-28

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Deep Learning 深度学习

![Python](https://img.shields.io/badge/Python-3.9+-3776AB?style=flat&logo=python&logoColor=white)
![PyTorch](https://img.shields.io/badge/PyTorch-2.x-EE4C2C?style=flat&logo=pytorch&logoColor=white)
![Jupyter](https://img.shields.io/badge/Jupyter-Notebook-F37626?style=flat&logo=jupyter&logoColor=white)

> **用 PyTorch 学习深度学习的笔记与代码示例** — 从 Tensor 基础到卷积神经网络，循序渐进。
>
> *Notes & code examples for learning deep learning with PyTorch — from tensor basics to CNN, step by step.*

## 简介 Introduction

本仓库是学习 PyTorch 深度学习过程中的笔记与代码示例合集，按学习路径组织：

This repository collects notes and code examples from my PyTorch deep learning journey, organized as a learning path:

- 🧱 **Tensor 基础** — Tensor 的创建、转换、计算、索引与变形
- 🧠 **神经网络** — 线性层、激活函数、nn 模块与模型保存
- 🏋️ **训练** — 损失函数、数据集、反向传播与完整训练流程
- ⚡ **优化** — 优化器、学习率调度、参数初始化与正则化
- 🖼️ **卷积神经网络** — 卷积、池化、经典 CNN 与实战项目

## 目录 Table of Contents

- [简介 Introduction](#简介-introduction)
- [概念导图 Concept Maps](#概念导图-concept-maps)
- [代码示例 Code Examples](#代码示例-code-examples)
- [笔记 Notes](#笔记-notes)
- [项目结构 Project Structure](#项目结构-project-structure)
- [环境要求 Requirements](#环境要求-requirements)
- [快速开始 Quick Start](#快速开始-quick-start)

## 概念导图 Concept Maps

<p align="center">
  <img src="./xmind/深度学习概念.png" alt="深度学习重要概念" width="100%"/><br/>
  <em>深度学习重要概念 / Key Concepts of Deep Learning — <a href="./xmind/深度学习概念.xmind">xmind 源文件</a></em>
</p>

<p align="center">
  <img src="./images/Back%20Propagation.png" alt="反向传播算法示意" width="100%"/><br/>
  <em>反向传播算法示意 / Backpropagation</em>
</p>

<p align="center">
  <img src="./xmind/深度学习workflow.png" alt="深度学习 workflow" width="100%"/><br/>
  <em>深度学习 workflow / Deep Learning Workflow — <a href="./xmind/深度学习workflow.xmind">xmind 源文件</a></em>
</p>

<p align="center">
  <img src="./xmind/Convolutional%20Neural%20Network卷积神经网络.png" alt="CNN 卷积神经网络" width="100%"/><br/>
  <em>CNN 卷积神经网络 / Convolutional Neural Network — <a href="./xmind/CNN.xmind">xmind 源文件</a></em>
</p>

## 代码示例 Code Examples

### 01 Tensor API · 张量操作

- [创建 Tensor (Create)](./pytorch-basic/01_tensor_api/tensor_create.ipynb)
- [类型转换 (Conversion)](./pytorch-basic/01_tensor_api/tensor_conversion.ipynb)
- [张量计算 (Calculation)](./pytorch-basic/01_tensor_api/tensor_calculation.ipynb)
- [索引与切片 (Indexing & Slicing)](./pytorch-basic/01_tensor_api/tensor_index.ipynb)
- [变形 (Reshape)](./pytorch-basic/01_tensor_api/tensor_reshape.ipynb)
- [统计 (Statistics)](./pytorch-basic/01_tensor_api/tensor_statistics.ipynb)

### 02 神经网络基础 · Neural Network Basics

- [线性层 (Linear Layer)](./pytorch-basic/02_nn_base/01_linear_test.ipynb)
- [激活函数 (Activation Functions)](./pytorch-basic/02_nn_base/02_activations.ipynb)
- [nn 模块 (nn.Module)](./pytorch-basic/02_nn_base/03_nn_test.ipynb)
- [Sequential 与模型保存 (Sequential & Save Model)](./pytorch-basic/02_nn_base/04_nn_sequential_save_model.ipynb)

### 03 神经网络训练 · Neural Network Training

- 损失函数 / Loss Functions
  - [回归损失 (Regression Loss)](./pytorch-basic/03_nn_train/01_loss_function/01_regression_loss_function.ipynb)
  - [BCE 损失 (BCE Loss)](./pytorch-basic/03_nn_train/01_loss_function/02_bce_loss.ipynb)
  - [交叉熵损失 (Cross Entropy Loss)](./pytorch-basic/03_nn_train/01_loss_function/03_cross_entropy_loss.ipynb)
- 数据集 / Datasets
  - [Dataset](./pytorch-basic/03_nn_train/02_dataset/01_dataset.ipynb)
  - [DataLoader](./pytorch-basic/03_nn_train/02_dataset/02_dataloader.ipynb)
- [反向传播 (Backpropagation)](./pytorch-basic/03_nn_train/03_backward.ipynb)
- [训练流程 (Training Pipeline)](./pytorch-basic/03_nn_train/04_nn_train.ipynb)
- [线性回归实例 (Linear Regression)](./pytorch-basic/03_nn_train/05_linear_regression_sample.ipynb)
- 🕹️ [手写数字识别 — 训练 (Digit Recognizer · Train)](./pytorch-basic/03_nn_train/06_digit_recognizer_sample.py)
- 🕹️ [手写数字识别 — 测试 (Digit Recognizer · Test)](./pytorch-basic/03_nn_train/07_digit_recognizer_test.py)

### 04 学习优化 · Optimization

- 优化器 / Optimizers
  - [动量与 SGD (Momentum & SGD)](./pytorch-basic/04_optimize/01_optimizers/01_momentums.py)
  - [StepLR 学习率调度 (Step LR Schedule)](./pytorch-basic/04_optimize/01_optimizers/02_step_lr.py)
  - [MultiStepLR 学习率调度 (MultiStep LR Schedule)](./pytorch-basic/04_optimize/01_optimizers/03_multistep.py)
  - [ExponentialLR 学习率调度 (Exponential LR Schedule)](./pytorch-basic/04_optimize/01_optimizers/04_exp_lr.py)
  - [Adagrad](./pytorch-basic/04_optimize/01_optimizers/05_adagrad.py)
  - [RMSprop](./pytorch-basic/04_optimize/01_optimizers/06_rmsprop.py)
  - [Adam](./pytorch-basic/04_optimize/01_optimizers/07_adam.py)
- [参数初始化 (Parameter Initialization)](./pytorch-basic/04_optimize/02_params_init.ipynb)
- [正则化 (Regularization)](./pytorch-basic/04_optimize/03_regularization.ipynb)
- 🕹️ [房价预测 (House Price Prediction)](./pytorch-basic/04_optimize/04_house_prices.py)

### 05 卷积神经网络 · CNN

- [卷积层 (Convolution)](./pytorch-basic/05_cnn/01_conv.ipynb)
- [池化层 (Pooling)](./pytorch-basic/05_cnn/02_pooling.ipynb)
- [经典 CNN 网络 (Classic CNNs)](./pytorch-basic/05_cnn/03_famous_cnn.ipynb)
- 🕹️ [手写数字识别 (Digit Recognizer · CNN)](./pytorch-basic/05_cnn/04_digit_recognizer_cnn.py)
- [服装分类 (Fashion Classification)](./pytorch-basic/05_cnn/05_fashion_classification.py)
- 🕹️ [服装分类器 — 训练 (Fashion Classifier · Train, AlexNet)](./pytorch-basic/05_cnn/06_fashion_V2_train.py)
- 🕹️ [服装分类器 — 测试 (Fashion Classifier · Test, AlexNet)](./pytorch-basic/05_cnn/07_fashion_V2_test.ipynb)

## 笔记 Notes

- [PyTorch 核心 API 笔记](./notes/PyTorch核心API笔记.md) — PyTorch Core API Notes
- [激活函数详解](./notes/激活函数详解.md) — Activation Functions Explained

## 项目结构 Project Structure

```text
.
├── data/            # 数据集:digit-recognizer、房价预测、fashion-mnist
├── graphic/         # 网络结构图绘制脚本
├── images/          # 可视化图片:激活函数曲线、优化器对比等
├── model/           # 训练好的模型权重
├── notes/           # 学习笔记
├── pytorch-basic/   # 代码示例(见「代码示例」章节)
│   ├── 01_tensor_api/
│   ├── 02_nn_base/
│   ├── 03_nn_train/
│   ├── 04_optimize/
│   └── 05_cnn/
└── xmind/           # 概念导图(.xmind 源文件 + 导出图片)
```

## 环境要求 Requirements

- Python 3.9+
- PyTorch 2.x(CPU / GPU 均可)
- NumPy · pandas · Matplotlib · scikit-learn · Pillow · Jupyter

```bash
pip install torch numpy pandas matplotlib scikit-learn pillow jupyter
```

> 💡 `networkx` 仅在 [graphic/nn_net_graph.py](graphic/nn_net_graph.py) 中用于绘制网络结构图，按需安装。

## 快速开始 Quick Start

```bash
git clone <你的仓库地址>
cd deep-learning
pip install torch numpy pandas matplotlib scikit-learn pillow jupyter

# 训练一个三层神经网络识别手写数字
python pytorch-basic/03_nn_train/06_digit_recognizer_sample.py
```

> 📝 `.ipynb` 文件用 `jupyter notebook` 打开浏览；`.py` 示例直接用 `python` 运行。
> 部分示例依赖外部数据集(如 Kaggle 的 digit-recognizer `train.csv`、fashion-mnist)，需提前下载并放入 `data/` 目录。