Course talk:CPSC522/Artificial Neural Network

Hi Ke,

It is glad to read your page, you gave a detailed explanation of ANN's architecture and how it works. Only have a few suggestions for you to refer:

1. It would be nice if you give the pseudo-code of the ANN, it will help me to better understand how the ANN works.

2. The citation is not in the correct form, the location it appears in your page should be indicated, maybe you could refer to the Wikipedia page, it use the same method to citing as UBC wiki.

Best regards,

Jiahong Chen

Hi Ke, Thanks for the write up on Neural Net. It provided many insights, however I have few suggestions, which may help in clarifying few things better:

• The examples you used to describe neural nets, contain directed edges. However it is not explicitly not made clear, what these directions imply
• You have given few examples of activation functions, but there use is not made clear. Providing a comparison of those function and in what context should which one be used will help readers to understand them better
• Neural nets while being powerful by providing a choice of different activation function, are also prone to over fitting. A small section on that trade of and best practices used in that setting may be added to help realizing the full potential of neural net over other models.
• Looks like the figure number 1.6 is missing
• The last section of where neural nets are used does not provide much insight. A one line description against those application areas can mitigate this
• I dont see any other pages linked both internal and external. So adding few of them will help interested readers to learn more about those
• The builds on and more general than sections are not edited. In case you feel like nothing to add there, then you may remove the texts.

best, Prithu Banerjee

Hi Prithu, thanks for your critique and suggestions. I have added Figue 1.6 and some links to Wikipedia, and edited Builds on and More general than sections.

Hi Jiahong, thanks for your critique and suggestions. However, for your first suggestion, I would say this page does not involve any specific algorithm, so I do not think I need to give any pseudo-code. For your second suggestion, I would say all the contents of this page comes from the book I cited.

Hi Ke,

First let me thank you for contributing in our course Wiki. I believe the Content part is really good and certainly has a lot of bright points. But I found a few parts of the page lack some of the things that are expected:
1) There are no links to other Wiki pages.
2) In fact there are no links and pointers at all (It really helps the users to let them have the option of clicking on some of the terminologies in order to let them read about the things they do not know).
3) The "Abstract" section needs more development I would say (Abstract is such a critical part of a page and really helps the users to get an idea about the content in a short period of time).
4) The "Builds on" section is also not developed (really helps the readers to understand the related areas).
5) There is only 1 reference in your bibliography section. You should reference more I believe, as in academic world a huge chunk of an article's credit is for its references.
6) There is no "More General Than" section. It is nice if you can add this part as well.
7) In the last part of the page, you mention the areas that NNs are used, but you do not give any links or pointers so that the reader can be directed to good sources for those areas.
I believe since you have done a great job in developing the content, with fixing these issues you can have a fantastic Wiki page.
Cheers!

Hi Mehrdad, thanks for your critique and suggestions. I have edited Builds on and More general than sections, added some links to Wikipedia in this page, and developed Abstract section.

Excellent work! Here are my scores, with comments below. Let me know if you want to discuss any of the points.

Scale of 1 to 5, where 1 = strongly disagree and 5 = strongly agree:

---

• (5) The topic is relevant for the course.
• (4.5) The writing is clear and the English is good.
• (5) The page is written at an appropriate level for CPSC 522 students (where the students have diverse backgrounds).
• (5) The formalism (definitions, mathematics) was well chosen to make the page easier to understand.
• (5) The abstract is a concise and clear summary.
• (5) There were appropriate (original) examples that helped make the topic clear.
• (3) There was appropriate use of (pseudo-) code.
• (5) It had a good coverage of representations, semantics, inference and learning (as appropriate for the topic).
• (5) It is correct.
• (5) It was neither too short nor too long for the topic.
• (5) It was an appropriate unit for a page (it shouldn't be split into different topics or merged with another page).
• (1) It links to appropriate other pages in the wiki.
• (1) The references and links to external pages are well chosen.
• (5) I would recommend this page to someone who wanted to find out about the topic.
• (4.5) This page should be highlighted as an exemplary page for others to emulate.

Comments:

---

• There are a few typos/grammar issues, but nothing that a quick proofread wouldn't fix.
• It can be useful to highlight important terms in the text using bold or italic effects.
• Some of the variables, such as neuron Y, are written using math tags in some places and in plain text in others.
• I believe we were supposed to use math tags instead of images for the formulas.
• I'd love to see some links in the text, both internal (to other parts of your page or to other 522 wiki pages) and external (to Wikipedia or other online sources).
• The page is fairly light on math/code, but you explain the functionality very well.

Abstract:

• Emphases are placed, not focused.
• The "Builds on" and "More general than" sections still consist only of the instructions; be sure to put some links in there.

Introduction:

• Contextual and historical motivation. Very nice.

What is an artificial neural network?:

• Nicely done!
• The sigmoid function doesn't need to be specified there, and the fact that it uses x instead of y can be a bit confusing. I'd suggest removing the formula and instead putting an internal link to your Common Activation Functions section, since you cover it there.

How are neural networks used?:

• Nicely done here as well!
• The Competitive Layer section appears to have no content.
• I'm not sure why I need to know what the derivative of the hyperbolic tangent is.

Where are neural nets being used?:

• Links would be especially awesome here.

Hi Jordon, Thanks for your critique and suggestions. I have added some links to Builds on and More general than sections, and changed the word focused to placed.