Talk:Oja's rule

Rewrite
This article seems mostly copied from and offers little insight into the workings of the rule. I am rewriting it. SamuelRiv 00:19, 15 November 2007 (UTC)

Ambiguous
In the article it's not clear to me if A(expression) means function A applied on expression or function A multiplied by expression. This prevents me from understanding the formulas in the articles. — Preceding unsigned comment added by 46.126.114.10 (talk) 14:33, 17 February 2013 (UTC)

Little Technical
I am pretty technically proficient person, albeit in a completely different field, but so far this article is difficult to follow, in terms of high level "What does this Rule actually state?" If someone could write a better intro paragraph that gives the lay man an understanding of what this rule means that would be great. I have no knowledge of this field, and like probably a half million other wikipedians, I clicked on it from today's main page. But this article doesn't really tell me what Oja's Rule means. SamuelRiv, above, may already be working on this, but I just wanted to request that. Thank you. Dachande (talk) 17:22, 26 November 2007 (UTC)

Encyclopedic documentation and didactic
I have been reviewing the wikipedia on some hard math areas and I can see that despite knowing the material, even knowing the material so well to the technical profficiency level, sometimes I can barely follow the expositions and have to read several times and navigate! Is this now a lost comment not even I will remember and nobody will read? I see no problem, besides financing, for the keepers of this site to consider rewriting the math areas so that they become didactic, though other reorientations for this material are also possible. Particularly, some middle concepts should be explained in the same page without links or at least notation described thoroughly or exemplified despite redundancies. For example in this page, several formulae arrive at terms such as y(n) where actual samples for y are never exposed, though I know from other references that sigmoid functions ARE an important in these neural network contexts, though not the only possibilities available. Danilo. 67.100.226.74 (talk) 23:59, 8 December 2010 (UTC)

Derivation Error
Is there an error in the following derivation:
 * $$\,w_i (n+1) ~ = ~ \frac{w_i}{\left( \sum_j w_j^p \right)^{1/p}} ~ + ~ \eta \left( \frac{y x_i}{\left(\sum_j w_j^p \right)^{1/p}} - \frac{w_i \sum_j y x_j w_j}{\left(\sum_j w_j^p \right)^{(1 + 1/p)}} \right) ~ + ~ O(\eta^2)$$

I have an extra p-1 component on the $$w_j$$ term in the sum:
 * $$\,w_i (n+1) ~ = ~ \frac{w_i}{\left( \sum_j w_j^p \right)^{1/p}} ~ + ~ \eta \left( \frac{y x_i}{\left(\sum_j w_j^p \right)^{1/p}} - \frac{w_i \sum_j y x_j w_j^{p-1}}{\left(\sum_j w_j^p \right)^{(1 + 1/p)}} \right) ~ + ~ O(\eta^2)$$

Can someone check me on that?