What Is Backpropagation?
👉🏻 Get Our Complete Data Science Training at 57% OFF: https://bit.ly/3vTadn8
Download Our Free Data Science Career Guide: https://bit.ly/3GX9k3p
Welcome to this 365 Data Science Tutorial Video where we will be talking about backpropagation. The most intuitive lesson, yet the hardest to grasp in mathematical terms. But before that, let us start from the other side of the coin with forward propagation.
Forward propagation is the process of pushing inputs through the net. At the end of each epoch, the obtained outputs are compared to targets to form the errors. Then we backpropagate, through partial derivatives and change each parameter, so errors at the next epoch are minimized. For the next minimal example, the backpropagation consisted of a single step: aligning the weights, given the errors we obtained. Here is where it gets a little tricky.
When we have a deep net, we must update all the weights related to the input layer and the hidden layer. In the example we give in the video, we have a picture with 270 weights. Updating all 270 weights is a big deal. But wait. We also introduced activation functions. This means we have to update the weights accordingly, considering the used non-linearities and their derivatives.
Finally, to update the weights we must compare the outputs to the targets. This is done for each layer. But we have no targets for the hidden units. We do not know the errors, so how do we update the weights? This is where backpropagation comes into play. We must derive the appropriate updates as if we had targets. Now, the way academics solved this issue is through errors. The main point is that we can trace the contribution of each unit, hidden or not, to the error of the output.
Stay till the end of the video tutorial to see an illustration of the backpropagation concept.
.
► Consider hitting the SUBSCRIBE button if you LIKE the content: / @365datasciencetutorials
► VISIT our newly gamified platform: https://bit.ly/3Zl6CMe
🤝 Connect with us on LinkedIn: https://www.linkedin.com/school/365datasci...
Follow us on Instagram: https://www.instagram.com/365datascience/
365 Data Science is an online educational career website that offers the incredible opportunity to find your way into the data science world no matter your previous knowledge and experience. This is why we have dedicated this channel to those who are completely new and are curious to explore the wonderful world of data science. Once you have built a basic theoretical knowledge you can sign up to our comprehensive curriculum where we have prepared numerous courses that suit the needs of aspiring BI Analysts, Data Analysts and Data Scientists.
#365datasciencetutorials #backpropagation #deeplearning