https://github.com/peterakande/multilayerperceptron

https://github.com/peterakande/multilayerperceptron

Last synced: 7 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/peterakande/multilayerperceptron
• Owner: PeterAkande
• Created: 2020-11-09T10:41:20.000Z (about 4 years ago)
• Default Branch: main
• Last Pushed: 2021-06-24T19:39:07.000Z (over 3 years ago)
• Last Synced: 2024-11-16T06:28:25.267Z (2 months ago)
• Language: Python
• Size: 29.3 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        

# MultilayerPerceptron

Just Got up and Decided to clear the drought of repositories on my GitHub page.

It Works on the principle of Back Propagation (All MLPs Do 😂), the ouput layer is Made up of Stacked Sigmoids...Lazy to adapt it to Softmax 😪.

The network is initialized with the training set and the accompanying labels and  list, whose length connotes the amount of layers and the value of each layer index in the list denotes the amount of nodes that particular  layer should have,also a one_hot parameter ,that is by  default  set  to false...But can be Set to True if labels are already one_hot_encoded.

The first layer of the list to be passed should be the Number of features and the last item of the list should be the amount of classes.

```python

#Example usage

#Import class

from MultilayerPerceptron import MLP

...

...

#Your preferred

#Preprocessing Technique

#And Loading Of Data

Layers=[60,40,20,10]

"""The MLP

has 2 hidden Layers,len(Layers)-2.

first item of Layers,60,is the amount of

features in each instance of training data

Last item of Layers

10, is number of classes"""

n=MLP(Layers)

n.train(data,labels, epochs,batchsize,l_r,one_hot=True)

#l_r is the learning_rate

#save weights with save_weights function

n.save_weights(weights_path,bias_path)

n.load_weights(weights_path,bias_path)

"""

Load weights from disk. load_weights is

wrapped with classmethod decorator

so after weights and bias has been

saved with n.save_weights(...) ....The weights can

then be used to make predictions for a later time

Without initializing the layers...E.g 

n=MLP.load_weights(...)"""

n.predict(data)

"""

n.predict(data) can then be used to

make predictions...

😄"""

```

The utils function is mainly for preprocessing

Some functions included are:

(1). Standardization :::utils.norm(mat), it returns the transformed dataset, the std learned and the mean

,Which can then be used on the test set by using utils.norm_test(mat,std,mean)

(2). Min-Max Scaling::utils.min_max(mat),it returns the transformed dataset,the min, and the range.

,Which can then be used on the test set by using utils.min_max_test(mat,mins,_range)

###PCA and whitening should be added soon

Would write a blog post on how MLPs work, using this repo as reference, soon..