change: move the period, so it's inside the quotation mark.

Placing commas and periods inside dialogue quotation marks is proper syntax/grammar in normal writing. .... It's improper syntax only in technical/computer writing. However, it's possible that -- stylistically -- this may have changed in recent years, as technical (and technology) writers have influenced mainstream writing somewhat.

change "Said" to lower-case "s"

My understanding of strict grammar rules, with regard to writing numbers, is as follows:

• When writing a number less than ten, one spells the number (and refrains from writing the digits).
• When writing a number greater than 10, one writes the digits (e.g., 24).

(Side note: I don't recall what the rule is for the number ten itself!)

Anyway, that being said: My sense is that in journalistic writing (and in fiction, and in novels), one can take stylistic licenses; it's entirely up to you, based on what style, context and goals you have in mind.

If it were me, I'd spell it out (but it's your call): "but three is still pretty good."

Flip the X and X' in X -> X'

Shouln't there be a space between "2,000", and "calorie"?

"share ~the~ a fundamental"

cannot

...In Antonio Machicao y Priemer, Andreas Nolda & Athina Sioupi (eds.), Zwischen Kern und Peripherie ;)

There are many Index-entries which appears twice or in this case 4 times. It is probably a problem of the scope of \index \index{\textbf{x}} =/= \index{x}

an

an N

that?

*gets

This paragraph does not build a sense-unit with the example before, and should be therefore indented, similar to the next paragraph beginning with Interjections.

without "."

subdivided

an

an

trick

a sentence

"changed in"

These vertical bars do not seem to work for me, in either the regular input mode or ":paste" mode.

lower case

to beat random search in a carefully-chosen intervals.

remove the "a"?

That is, we are generating a random random with a uniform distribution, but then raising it to the power of 10.

The word random is repeated?

True to the denominator term in the RMSprop update

remove the word dropout?

theoretical convergence guarantees

update that has recently

appears as a slightly more interpretable (remove first more)

can have sizes in the millions parameters can have millions of parameters

Maybe it should be: hack the code to remove the regularization loss contribution.

Typo

\(x_i\) is scaled

The figure above has a value of 1.04 for the softmax case. I think that should be \(0.452\).

Typo. Grammar.

The synapses are not just a single weight ,but a complex non-linear dynamical system

The usual spelling is "no one" and to a lesser extent "no-one". Just nit-picking. B-)

typo: drop-out

Typo/grammar: The motivation for this is based on a compromise and an equivalent analysis

Editorial nitpicking: spaced en dash needed here :)

Double-space here

typo? or just confusion grammar?

Another issue that should be addressed

cease

cease