Halo, recently I have been learning with the course of Neural Networks and Deep Learning.
That I’m confused with one notion about how the Neural Network algoritms should be explained in a mathematical way. Please help to correct my assumption for the algorithms, many thanks!
It is likely as the following:
I just wanna it to be simple for the comprehension, so it may be too cute since I’m new to this field.
Thanks!
Hello @Xiao_Le_ZHONG,
I think your summary has captured a good idea of how neural network works! I would remove “convex values” as, generally speaking, neural network is not convex, and because it is not convex, I would also refrain from saying “best optimized” but I would use the term “local minimum”. Does “deviated values” mean “derivative values”? If so, use “derivative values” instead.
I would also recommend you to use AI and make a more concise version of your summary, then adjust your summary by taking in the good stuff. If AI’s version misses out any idea you want to convey, it would be interesting to ask AI for a reason and even a revised version.
Cheers,
Raymond
Hello @Xiao_Le_ZHONG,
As explained before, I wouldn’t call it “convex function”. Also, I recommend chatbot for suggestions on further revision of the text 
For the flow chart, I would put \hat{Y} as shown below
We can have choices for hidden layer’s activation, but since you have specified log loss, you might as well want to specify sigmoid as the output layer’s activation.
Cheers,
Raymond
