Hi, I am curious why we have multiple examples. Can’t this be one big and long example? Are there any mathematical reasons behind this? If there are multiple examples, the size of the row needs to match.
Thanks!
What do you mean by multiple examples? Please mention the week and assignment number or a video link you are talking about.
1 Like
I am in w3, and the video I am referring to is Vectorizing Across Multiple Examples. There were multiple x training set.
Hi @YoTaro_Watanabe ,
Not sure what your question is about. The function of a neural network is to figure out the weight and bias parameters through learning by examples fed to the network. The more examples we can get, the better the estimation is going to be. Vectorization simply has an advantage in terms of efficiency and speed by utilising the hardware’s vectorization functionality.
Thanks Kic, I understand the importance and advantages of more examples. My question is, why does it need to be in several columns in a matrix?
Hi @YoTaro_Watanabe ,
By organizing it in columns or rows (depending on implementation) in a matrix would clearly show what each example consists of.
If you model is a cat classifier using a neural network, then the training and testing dataset would be images of cat in various poses, background, lighting etc. So how would you present this dataset to your model?
Ah! Understood now. Thank you!
Hello YoTaro Watanabe,
Mentor Kic has provided a very good explanation to your query by giving an example on cat classifier using a neural network. The only thing we can replace here is the sample, not example.
Example is something where we solve a case using an object, but samples are something that are applied as inputs to get best of the estimations.