User talk:Mpennin/sandbox

Overall, I feel that this article was a bit rushed and could use some touching up. In some areas, it is not clear as to how the paragraphs relate to Hinton. It's great that you mentioned several things of which Hinton is known for, but try to talk more about how he came up with those ideas rather than primarily explaining what those things are.

I would recommend decreasing the font of Geoffrey Hinton right at the beginning. Also, more citations should be added throughout the article. A picture of Geoffery Hinton in the description box would be nice too just to match all his contributions to a face, if possible.

You mentioned his date of birth twice, I recommend keeping the date of birth in your "Early Life" paragraph only.

Under "Career," reword the sentence to: Hinton devised many theories and models in an attempt to explain and understand how the brain works. He began this process in 1983 with the Boltzmann Machine.

Under "The Boltzmann Machine," hyperlink stochastic, binary, cost function, visible-visible or hidden-hidden connections and reword search problems, and learning problems to search problems and learning problems.

Under "Backpropagation," hyperlink artificial neural network (ANN).

Under "Contrastive Divergence," perhaps a definition would help in understanding the connection between that and RBM and also hyperlink contrastive divergence.

Also, hyperlink geometric transformations, Alex Krizhevsky, and Ilya Sutskever.

Rpau23 (talk) 19:57, 3 April 2013 (UTC)

Peer Review

Overall, this was a good article. I feel like you got the general message across and kept it short and to the point. There were a few minor spelling and punctuation errors that I corrected. Generally, I would suggest adding more hyperlinks because in many spots there are important terms, which should be highlighted to the reader. As for the hyperlinks that are already present, it would be helpful if you explained some of the concepts beyond simply linking them to adjacent pages (makes it easier for the reader-- they don't have to jump from page to page unless they are interested in reading or delving further). I would also recommend inputting at least one or two images so that your article is not just a solid block of text. I had some criticisms/praise for each section of your article so I will present them below in a manner structured to your article.

Geoffrey Hinton

Please decrease the font for Geoffrey Hinton (in concordance with what was stated by the previous editor). There were a few minor spelling/punctuation errors, which I corrected. I would like to commend you on your use of a table here. It was very useful and informative to the reader and presented a general summary of your individual in a visually pleasing manner. The only suggestion I have here is to recommend perhaps inputting an image of Geoffrey Hinton in your table. I would also recommend adding more citations in this section linking to research related to the Boltzmann machine, backpropagation theory and the Hemlholtz machine. Good use of hyperlinks.

Biography

I would get rid of the part where you say "although he disliked it" in reference to his early education. This is speculation and not necessarily empirically validated evidence. It's not important to your reader's understanding of the text and does not present an objective frame of view. I would add a hyperlink to the word "Psychology". Also, do we know why he became interested in Psychology? Did he have an undergraduate supervisor, which inspired him? How do we know this? Build on this a little more, otherwise simply state facts. Could you also please explain "relaxation methods" more? To a reader who does not know of Geoffrey Hinton's work this is an abstract concept which needs to be operationalized.

Career

In this section, I would recommend further explaining parallel computing architecture. How did this lead to work on various AI machines? Please explain this further. I feel as if you are stating facts without actually backing them up.

Theories

This section is much too technical. You have a difficult topic and need to break things down in a more comprehensible manner. Please make sure you are defining concepts; don't assume that everyone knows what you are talking about (especially since Wikipedia is targeted to a general public). For example: define stochastic (perhaps in its place use a term like "balanced" or hyperlink at least), define hidden units (you can also use an illustration to understand this concept). Hyperlink: neural computations, cognitive psychology, vectors, cost function. Check your hyperlink for NETtalk because it does not seem to be working. If you do not hyperlink at least do your best to define these terms. Bold: search problems and learning problems. It will help these important terms to stand out from the rest of your article and mass of text. Define: supervised vs. unsupervised learning. In the section on geometric transportations, please explain this idea more. It's a little too simplistic.

Honours and Awards

I would recommend putting this information in a table. It would be much easier to read than having it in a block of text.

See Also

This was a nice touch. Well done !

References

Be careful because some of your references repeat themselves.

All in all, congratulations on a job well done and good luck in the final stages of the assignment!!

Angelamarti (talk) 19:10, 3 April 2013 (UTC)

Hi Rpau23,

   I appreciate your comments and I agreed with most of them.  I realized after you mentioned it that I needed to change the wording of one of the sentences under the Career heading.  I think it works a bit better now.  Creating hyperlinks for some key words was beneficial too.  However, in regards to creating a hyperlink for Alex Krizhevsky and Ilya Sutskever - they are graduate students working with Geoffrey Hinton and do not have Wikipedia pages written on them, so I can't make any links for them.  As for the picture of the Hinton, Wikipedia seemed to give me a hard time for trying to upload a photo unless I had explicit consent from the owner of the photo.  I had contacted Hinton previously to get some information from him personally, but I got the sense that I was annoying him slightly as he said he was really busy and was terse in his responses, so I decided not to pursue him for the picture.  Thanks for your comments and criticisms though - they were helpful.

Angelamarti,

   Thanks for your input, it was quite useful.  I decreased the font size of Geoffrey Hinton at the top - I agree it looks much better at the size it now is.  I thought that I would keep the additional "although he disliked it" bit.  I emailed him and he told me this personally, so I know it to be a fact, and I thought it added a personal touch to the article.  I understand that an encyclopedia article is supposed to be purely factual, but it is still a fact that he disliked school, and in my opinion it humanizes an article that contains mostly talk of computation and mathematical models etc.  You might be right that it is better gone, but I still think it's nice to keep it.  I added a hyperlink to the word "psychology" and I made it more clear what I meant by relaxation methods.  I had actually defined relaxation methods in the same sentence that I mentioned it, but it sounded like I was describing a certain type of relaxation method or a particular instance of relaxation method rather than defining the thing itself.  I think adding the dash in the middle of the sentence makes this a little clearer.  I also added a few words to clear up what was meant by supervised and unsupervised.  As for the honours and awards, with all due respect I don't think a table would improve the article.  They are more of a "by the way" thing and I think adding a table would draw the attention away from the real content of the article.  Honours and awards probably mean relatively little to the reader in comparison to Hinton's life and work.  Had the article been substantially longer and included more diagrams, pictures, etc., then yes I would probably agree that a table would suit this article well.  Thanks again!