Doubt regarding the forward propagation lecture


Here what are the 25 (and 15) networks in the single network layer. Like in the demand prediction they were the various factors like price, material, marketing etc. What do these 25 values represent here. I mean there is only 1 property which is color intensity of the digit so why do we need 25.

Hey @PARTH_SARATHI_DIXIT1,

The 25 and 15 here represents the number of hidden units of a single layer.
I’d recommend you to enroll in the “Deep Learning Specialization” to know more about this. However, you can ask whatever you want and i will try to answer all your questions but whenever you want to go in depth i recommend you to enroll in the specialization.

Happy Learning!
Regards,
Jamal

For more clarification @PARTH_SARATHI_DIXIT1,

The purpose of having multiple hidden units (25 or 15 in this case) is to allow the neural network to capture different aspects or representations of the input data. Each hidden unit can learn to detect specific patterns or combinations of pixel intensities that are relevant for the task at hand. By having multiple hidden units, the network can learn a more diverse set of features and potentially improve its ability to recognize digits accurately.

1 Like