Skip to content

Commit

Permalink
📝Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
jswanglp committed Jun 26, 2019
1 parent 244fe62 commit e42dbc4
Showing 1 changed file with 5 additions and 4 deletions.
9 changes: 5 additions & 4 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -18,19 +18,20 @@ matplotlib 2.0.2
- [code](./codes/NN.py)
- [jupyter notebook](./notebooks/NN.ipynb)

**小波神经网络 (Wavelet Neural Network, WNN)** 程序实现包含 `1` 个小波隐藏层的小波神经网络,激活函数为小波函数 [morlet 函数](https://www.mathworks.com/help/wavelet/ref/morlet.html)
**小波神经网络 (Wavelet Neural Network, WNN)** 程序实现包含 `1` 个小波隐藏层的小波神经网络,激活函数为小波函数 [morlet 函数](https://www.mathworks.com/help/wavelet/ref/morlet.html)(左下图),右下图所示的则为 `morlet` 小波的导函数。

<div align='center'>
<img src="./images/morlet_wavelet.png" alt="morlet_wavelet.png" height="200" width="250">
<img src="./images/morlet_wavelet_and_its_derivative.png" alt="morlet_wavelet.png" height="200" width="500">
</div>

包含单个隐层的小波神经网络的能力与双隐层的普通神经网络相当。更多关于小波激活函数的小波神经网络在这个 [repo](https://github.com/jswanglp/MyML/blob/master/notebooks(colab)/Neural_network_models/Supervised_learning_models/WNN.ipynb),运行结果的笔记保存至 `jupyter notebook` 文件。

- [code](./codes/Wavelet_NN.py)
- [jupyter notebook](./notebooks/Wavelet_NN.ipynb)

当隐层的小波函数为 `POLYWOG3` 小波函数<a href='#fn2' name='fn2b'><sup>[2]</sup></a>时
当隐层的小波函数为 `POLYWOG3` 小波函数<a href='#fn2' name='fn2b'><sup>[2]</sup></a>时(左下图),右下图所示的则为 `POLYWOG3` 小波的导函数。
<div align='center'>
<img src="./images/POLYWOG3_wavelet.png" alt="POLYWOG_wavelet.png" height="200" width="250">
<img src="./images/polywog3_wavelet_and_its_derivative.png" alt="POLYWOG_wavelet.png" height="200" width="500">
</div>

网络收敛速度明显快于普通的神经网络,精度在经过 `40` 次迭代之后达到了含双隐层的普通神经网络需要 `200` 次迭代才能达到的结果<a href='#fn3' name='fn3b'><sup>[3]</sup></a>。代码、运行结果的笔记保存至 `py``jupyter notebook` 文件。
Expand Down

0 comments on commit e42dbc4

Please sign in to comment.