# MultiLayer Neural Network (ak_pynn)
A simplistic and efficient pure-python neural network library, which can be used to build , visualize and deploy deep learning ANN models. It is optimized for best performance.
- Optimized for performance
- Better Visualization
- Cross platform
## Authors
- [@ankit_kohli](https://www.github.com/ankit869)
## License
[MIT](https://choosealicense.com/licenses/mit/)
## Support
For support, email contact.ankitkohli@gmail.com
## Features
- [x] Efficient implementations of activation functions and their gradients
- [x] Sigmoid
- [x] ReLU
- [x] Leaky ReLU
- [x] Softmax
- [x] Softplus
- [x] Tanh
- [x] Elu
- [x] Linear
- [x] Efficient implementations of loss functions and their gradients
- [x] Mean squared error
- [x] Mean absolute error
- [x] Binary cross entropy
- [x] Categorical cross entropy
- [x] Several methods for weights initialization
- [x] ```'random uniform'```, ```'random normal'```
- [x] ```'Glorot Uniform'```, ```'Glorot Normal'```
- [x] ```'He Uniform'```,```'He Normal'```
- [x] Neural network optimization using
- [x] Gradient Descent (Batch/ SGD / Mini-Batch)
- [x] Momentum
- [x] Adagrad
- [x] RMSprop
- [x] Adam
- [x] Regularizations
- [x] L1 Norm
- [x] L2 Norm
- [x] L1_L2 Norm
- [x] Dropouts
- [x] Batch Normalization
- [x] Early Stopping
- [x] Validation Splits
- [x] Predict Scores
## Installation
Install the release (stable) version from PyPi
```
pip install ak-pynn
```
## Usage/Examples
Import
```python
from ak_pynn.mlp import MLP
```
Usage
```python
model = MLP()
model.add_layer(4,input_layer=True)
model.add_layer(10,activation_function='relu',batch_norm=True)
model.add_layer(10,activation_function='relu',dropouts=True)
model.add_layer(10,activation_function='relu')
model.add_layer(3,activation_function='softmax',output_layer=True)
model.compile_model(optimizer='Adam',loss_function='mse',metrics=['mse','accuracy'])
```
Output
```
( MODEL SUMMARY )
===================================================================
Layer Activation Output Shape Params
===================================================================
Input linear (None, 4) 0
-------------------------------------------------------------------
Dense relu (None, 10) 50
-------------------------------------------------------------------
BatchNormalization None (None, 10) 40
-------------------------------------------------------------------
Dense relu (None, 10) 110
-------------------------------------------------------------------
Dropout None (None, 10) 0
-------------------------------------------------------------------
Dense relu (None, 10) 110
-------------------------------------------------------------------
Output softmax (None, 3) 33
-------------------------------------------------------------------
===================================================================
Total Params - 343
Trainable Params - 323
Non-Trainable Params - 20
___________________________________________________________________
```
Visualizing model
```python
model.visualize()
```
Training the model
```python
model.fit(X_train, Y_train,epochs=200,batch_size=32,verbose=False,early_stopping=False,patience=3,validation_split=0.2)
model.predict_scores(X_test,Y_test,metrics=['accuracy','precision','macro_recall'])
plt.plot(model.history['Val_Losses'])
plt.plot(model.history['Losses'])
```
## TESTS
[@mnist_test](https://github.com/ankit869/ak_pynn/blob/main/mnist_test.ipynb)
[@iris_test](https://github.com/ankit869/ak_pynn/blob/main/iris_test.ipynb)
[@mlp_demo](https://github.com/ankit869/ak_pynn/blob/main/mlp_demo.ipynb)
## Citation
If you use this library and would like to cite it, you can use:
```
Ankit kohli, "ak-pynn: Neural Network libray", 2023. [Online]. Available: https://github.com/ankit869/ak-pynn. [Accessed: DD- Month- 20YY].
```
or:
```
@Misc{,
author = {Ankit kohli},
title = {ak-pynn: Neural Network libray},
month = May,
year = {2023},
note = {Online; accessed <today>},
url = {https://github.com/ankit869/ak-pynn},
}
```
Raw data
{
"_id": null,
"home_page": "",
"name": "ak-pynn",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "neural network,pure python,ankit_nn,machine learning,ML,deep learning,deepL,MLP,perceptron,ankit kohli,ak_pynn",
"author": "Ankit kohli",
"author_email": "<contact.ankitkohli@gmail.com>",
"download_url": "https://files.pythonhosted.org/packages/46/5f/42c8e950376e88a25406fa849d9a30651cfd70ab373b19d604633e5b09a4/ak_pynn-0.1.8.tar.gz",
"platform": null,
"description": "\r\n# MultiLayer Neural Network (ak_pynn)\r\nA simplistic and efficient pure-python neural network library, which can be used to build , visualize and deploy deep learning ANN models. It is optimized for best performance.\r\n\r\n- Optimized for performance\r\n- Better Visualization\r\n- Cross platform\r\n\r\n\r\n\r\n## Authors\r\n\r\n- [@ankit_kohli](https://www.github.com/ankit869)\r\n\r\n\r\n## License\r\n\r\n[MIT](https://choosealicense.com/licenses/mit/)\r\n\r\n\r\n## Support\r\n\r\nFor support, email contact.ankitkohli@gmail.com\r\n\r\n\r\n## Features\r\n\r\n- [x] Efficient implementations of activation functions and their gradients\r\n - [x] Sigmoid\r\n - [x] ReLU\r\n - [x] Leaky ReLU\r\n - [x] Softmax \r\n - [x] Softplus \r\n - [x] Tanh \r\n - [x] Elu \r\n - [x] Linear \r\n- [x] Efficient implementations of loss functions and their gradients\r\n - [x] Mean squared error \r\n - [x] Mean absolute error\r\n - [x] Binary cross entropy \r\n - [x] Categorical cross entropy \r\n- [x] Several methods for weights initialization\r\n - [x] ```'random uniform'```, ```'random normal'```\r\n - [x] ```'Glorot Uniform'```, ```'Glorot Normal'```\r\n - [x] ```'He Uniform'```,```'He Normal'```\r\n\r\n- [x] Neural network optimization using \r\n - [x] Gradient Descent (Batch/ SGD / Mini-Batch)\r\n - [x] Momentum\r\n - [x] Adagrad\r\n - [x] RMSprop\r\n - [x] Adam\r\n\r\n- [x] Regularizations\r\n - [x] L1 Norm\r\n - [x] L2 Norm\r\n - [x] L1_L2 Norm\r\n - [x] Dropouts\r\n\r\n- [x] Batch Normalization\r\n- [x] Early Stopping\r\n- [x] Validation Splits\r\n- [x] Predict Scores\r\n## Installation\r\nInstall the release (stable) version from PyPi\r\n```\r\npip install ak-pynn\r\n```\r\n\r\n## Usage/Examples\r\n\r\nImport\r\n```python\r\nfrom ak_pynn.mlp import MLP\r\n```\r\n\r\nUsage \r\n```python\r\nmodel = MLP()\r\nmodel.add_layer(4,input_layer=True)\r\nmodel.add_layer(10,activation_function='relu',batch_norm=True)\r\nmodel.add_layer(10,activation_function='relu',dropouts=True)\r\nmodel.add_layer(10,activation_function='relu')\r\nmodel.add_layer(3,activation_function='softmax',output_layer=True)\r\nmodel.compile_model(optimizer='Adam',loss_function='mse',metrics=['mse','accuracy'])\r\n```\r\nOutput\r\n```\r\n\r\n ( MODEL SUMMARY ) \r\n \r\n ===================================================================\r\n Layer Activation Output Shape Params \r\n ===================================================================\r\n\r\n Input linear (None, 4) 0 \r\n -------------------------------------------------------------------\r\n\r\n Dense relu (None, 10) 50 \r\n -------------------------------------------------------------------\r\n\r\n BatchNormalization None (None, 10) 40 \r\n -------------------------------------------------------------------\r\n\r\n Dense relu (None, 10) 110 \r\n -------------------------------------------------------------------\r\n\r\n Dropout None (None, 10) 0 \r\n -------------------------------------------------------------------\r\n\r\n Dense relu (None, 10) 110 \r\n -------------------------------------------------------------------\r\n\r\n Output softmax (None, 3) 33 \r\n -------------------------------------------------------------------\r\n\r\n ===================================================================\r\n\r\n Total Params - 343\r\n Trainable Params - 323\r\n Non-Trainable Params - 20\r\n ___________________________________________________________________\r\n \r\n```\r\nVisualizing model\r\n```python\r\nmodel.visualize()\r\n```\r\n\r\n\r\n\r\nTraining the model\r\n```python\r\nmodel.fit(X_train, Y_train,epochs=200,batch_size=32,verbose=False,early_stopping=False,patience=3,validation_split=0.2)\r\nmodel.predict_scores(X_test,Y_test,metrics=['accuracy','precision','macro_recall'])\r\nplt.plot(model.history['Val_Losses'])\r\nplt.plot(model.history['Losses'])\r\n\r\n```\r\n## TESTS\r\n\r\n\r\n[@mnist_test](https://github.com/ankit869/ak_pynn/blob/main/mnist_test.ipynb)\r\n\r\n[@iris_test](https://github.com/ankit869/ak_pynn/blob/main/iris_test.ipynb)\r\n\r\n[@mlp_demo](https://github.com/ankit869/ak_pynn/blob/main/mlp_demo.ipynb)\r\n## Citation\r\nIf you use this library and would like to cite it, you can use:\r\n```\r\nAnkit kohli, \"ak-pynn: Neural Network libray\", 2023. [Online]. Available: https://github.com/ankit869/ak-pynn. [Accessed: DD- Month- 20YY].\r\n```\r\nor:\r\n```\r\n@Misc{,\r\n author = {Ankit kohli},\r\n title = {ak-pynn: Neural Network libray},\r\n month = May,\r\n year = {2023},\r\n note = {Online; accessed <today>},\r\n url = {https://github.com/ankit869/ak-pynn},\r\n}\r\n```\r\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "A simplistic and efficient pure-python neural network library that allows to build multilayer neural network with ease.",
"version": "0.1.8",
"project_urls": null,
"split_keywords": [
"neural network",
"pure python",
"ankit_nn",
"machine learning",
"ml",
"deep learning",
"deepl",
"mlp",
"perceptron",
"ankit kohli",
"ak_pynn"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "4e08ccd4ee4289ebb27fdc1770077df5adec52cdd63f0a225814a779e0a3109a",
"md5": "95ab3a9a64385bf66724745d11bd8db7",
"sha256": "31b7bf8332ec94e686cc4b2c61bea840b446f9c3c7801fbb6f25a731a0496db5"
},
"downloads": -1,
"filename": "ak_pynn-0.1.8-py3-none-any.whl",
"has_sig": false,
"md5_digest": "95ab3a9a64385bf66724745d11bd8db7",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 15122,
"upload_time": "2023-05-28T19:44:46",
"upload_time_iso_8601": "2023-05-28T19:44:46.185856Z",
"url": "https://files.pythonhosted.org/packages/4e/08/ccd4ee4289ebb27fdc1770077df5adec52cdd63f0a225814a779e0a3109a/ak_pynn-0.1.8-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "465f42c8e950376e88a25406fa849d9a30651cfd70ab373b19d604633e5b09a4",
"md5": "67e8bddb2705a4462aacd7309d0c5053",
"sha256": "9b0b618102724a5f21b44decd092f7b10a52a886e8d18f5d6db656000add2619"
},
"downloads": -1,
"filename": "ak_pynn-0.1.8.tar.gz",
"has_sig": false,
"md5_digest": "67e8bddb2705a4462aacd7309d0c5053",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 16670,
"upload_time": "2023-05-28T19:44:49",
"upload_time_iso_8601": "2023-05-28T19:44:49.343777Z",
"url": "https://files.pythonhosted.org/packages/46/5f/42c8e950376e88a25406fa849d9a30651cfd70ab373b19d604633e5b09a4/ak_pynn-0.1.8.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-05-28 19:44:49",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "ak-pynn"
}
JSON