1

I am new in Deep learning. And was implementing linear regression NN from scratch

I have seen

y = X*W +b

so i believe bias is for per node. Like layer haveing 4 node then number of biases will be 4. Per according to this formula it saying bias is for per layer. And all the nodes have that same bias Is that correct ?

Rajan Lagah
  • 111
  • 3
  • Related: https://stats.stackexchange.com/questions/178835/bias-input-in-neural-network and https://stats.stackexchange.com/questions/447528/how-to-understand-the-weights-and-bias etc: https://stats.stackexchange.com/search?q=%5Bneural-networks%5D+bias+ – Sycorax Jul 20 '20 at 17:02
  • @Sycorax No. bcoze in this link https://medium.com/@lankinen/fast-ai-lesson-8-notes-part-2-v3-8965a6532f51. You can see he is making length 10 array of biases which is number of hidden layers – Rajan Lagah Jul 20 '20 at 18:01
  • I'll say it again, I don't think you're using "layer" in the same way that everyone else does. The post is creating a **single layer** with **10 nodes.** In other words, the reason you're confused appears to be because you're not using terminology correctly. – Sycorax Jul 20 '20 at 18:03
  • 1
    @Sycorax Okayyyyyyyyy that is correct – Rajan Lagah Jul 20 '20 at 18:17

0 Answers0