Part - 1: Backward Propagation as laid out in asgmt_backpropagation.xlsx

Here we are defining all the nodes of the neural network so we can use it to further along the way to the next step in backward propagation

Here we are calculating the partial derivatives line 2: Since E2 is not dependent on w5, we are excluding it and calculate only for E1 which is dependent on w5 line 4: we find the partial derivative E1 w.r.t a_o1 line 5: we find the partial derivative of o1 w.r.t of w5

Here we are calculating the E_Total w.r.t w5, w6, w7 and w8

Here we are calculating the E_Total w.r.t w5, w6, w7 and w8

NOTE: Rest of the steps have been marked with post-it like sticky comments in the Excel file

Part - 2:

Purpose: Session 6 assignment is to try to achieve the following:

• Reduce the parameters
• Use any or all the techniques about CNN layers, loss functions etc.,
• Achieve 99.3% accuracy

Based on MNIST dataset

Create a simple Convolutional Neural Network model and predict

Project Setup:

Clone the project as shown below:-

$ git clone [email protected]:pankaja0285/era_v1_session6_pankaja.git
$ cd era_v1_session6_pankaja

About the file structure
|__asgmt_backpropagation.xlsx
|__README.md
|__S6.ipynb

NOTE: List of libraries required: torch and torchsummary, tqdm for progress bar, which are installed using requirements.txt

One of 2 ways to run the S6.ipynb notebook:

1. Using Anaconda prompt - Run as an administrator start jupyter notebook from the folder era_v1_session5_pankaja and run it off of your localhost
   NOTE: Without Admin privileges, the installs will not be correct and further import libraries will fail.
   

jupyter notebook

2. Upload the notebook folder era_v1_session6_pankaja to google colab at colab.google.com and run it on colab
   

NOTE: Follow along the S6.ipynb - notebook cells and run sequentially to see the outputs.

Notebook execution sequence of sections is as follows:

File used: S6.ipynb

1. Following the section Model - 1 for all the set up of params and training and plotting the accuracies and losses
2. Following the section Model - 2 for all the set up of params and training and plotting the accuracies and losses

Contributing:

For any questions, bug(even typos) and/or features requests do not hesitate to contact me or open an issue!