How to rewrite my sentence?

This is a discussion thread · 8 replies
Anonymous:
Hello,
I would like to rewrite the following sentence, but could not think of any better way.
"A classifier predicts the class of an unknown example by rules (boundaries) that are induced from the training examples to discriminate the classes in the training examples."

Is the following sentence clear and sufficent?
"A classifier predicts the class of an unknown example by rules (boundaries) that are induced from training examples to discriminate their classes."
Wen-Feng
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Puzzle Monster:
How about this:
"A classifier predicts the class of a new X using rules induced from training examples."
Replace X with whatever you're trying to classify, i.e. "soil sample," "plant," "comic book hero."
"Unknown example" doesn't really make sense. It's not unknown, it just hasn't been classified yet. And it's not an example - the examples are what you used for training.
As for "to discriminate the classes in the training examples" vs. "to discriminate
their classes" - both are awkward, and as far as I can tell don't actually add any information. I'd take that part out entirely.

-PM
http://www.puzzlemonster.com
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Anonymous:
Thanks! Should we add 'the' before 'rules' here?
"A classifier predicts the class of a new X using the rules induced from
training examples."
How about if I want to describe the rules as being used to discriminate the class boundaries?
"A classifier predicts the class of a new X using the rules, being used to discriminate the classes, induced from training examples."

Sounds awkward?
Puzzle Monster =E5=AF=AB=E9=81=93=EF=BC=9A
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Anonymous:
[nq:1]Hello, I would like to rewrite the following sentence, but could not think of any better way. "A classifier predicts the class of an unknown example by rules (boundaries) that are induced[/nq]
I do not like "induced". The 'correct' word is "Inferred" meaning assumed on the basis of evidence although not expressly stated. "Induced" is just wrong. It means cause to happen, as in "Her labor was induced by the doctor." as in having a baby.
[nq:1]from the training examples to discriminate[/nq]
I do not like "discriminate" either. I prefer "differentiate".
[nq:1]the classes in the training examples."[/nq]
Thus:
"A classifier predicts the class of an unknown example by rules (boundaries) that are inferred from the training examples which differentiate the classes in the training examples."

GFH
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Richard Yates:
[nq:2]"A classifier predicts the class of an unknown example by rules (boundaries) that are induced[/nq]
[nq:1]I do not like "induced". ..."Induced" is just wrong. It means cause tohappen, as in "Her labor was induced by the doctor." as in having a baby.[/nq]
"Induce" also means "to infer by means of inductive reasoning", that is, by discerning a rule from many examples.
Richard Yates
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Anonymous:
Thanks, but I meant to express that the rules are also used to differentiate the classes in the training examples. Every training example has its corresponding class (positive or negative). The rules are used for discriminate positive examples from negative examples. Is it better if I write it as:

"A classifier predicts the class of an unknown example by rules (boundaries) that are induced from the training examples. The rules can be used to differentiate the classes in the training examples."?

How can we combine these two ideas into one sentence?

Wen-Feng
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Anonymous:
[nq:1]Thanks, but I meant to express that the rules are also used to differentiate the classes in the training examples. ... that are induced from the training examples. The rules can be used to differentiate the classes in the training examples."?[/nq]
I would use "deduced". It seems that you are now saying that there are actaul rules which have been deduced from examples.
[nq:1]How can we combine these two ideas into one sentence?[/nq]
Why do you want to do that? The two sentences make the concept clearer.
GFH
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Clark S. Cox III:
[nq:1]Thanks, but I meant to express that the rules are also used to differentiate the classes in the training examples. ... be used to differentiate the classes in the training examples."? How can we combine these two ideas into one sentence?[/nq]
To me, that sounds like deduction, not induction. You're inferring a conclusion about particular instances from general rules. Induction would be inferring a conclusion about general rules from particular instances.

Clark S. Cox III
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Blue Hornet:
[nq:1]Hello, I would like to rewrite the following sentence, but could not think of any better way. "A classifier predicts ... the class of an unknown example by rules (boundaries) that are induced from training examples to discriminate their classes." Wen-Feng[/nq]
There seem to be several things wrong with the sentence, to my ear. First off, the thing being described, the classifier, sounds more like an algorithm or some other software, rather than a machine or a mechanism. So I would state it that way: "the classifier program" or "the classifier algorithm" (works in such and such a way).

Next, I think that I would describe the rules, rather than the outcome: "The classifier algorithm sets parameters based on the machine's reading of the training examples, in order to ..." I had trouble with "discriminate their classes".
Finally, what "predictions" can possibly be made about an "unknown example"? Does the classifier actually make predictions about things it hasn't experienced?
As much as I try, I can't even imagine what is being discussed here.

To my way of thinking (and my experience with "classifiers" as machines and machine parts may be blocking my understanding of what you're trying to say):
A classifier sorts incoming materials into the different classes of material that it is set to recognize by physical characteristics such as particle size, density, magnetism and weight, or moisture content or other characteristic, and shunts or exports the material into the correct processing path, or into a waste bin for 'unrecognized' or 'nonconforming'.
I've just pulled off the top of my head; I don't play with classifiers. How does your classifier differ from this example? Are you handling data or physical samples? How are the 'rules' established by the classifier?
This thread originates from within 'usenet', and as such the content and users are not guaranteed to have been moderated by our community.
Live chat
Registered users can join here