Re: size of variables in hyperspace?

chris' solution, where you gradually get rid of the old .chs-files,
sounds more implementable than bill's version with doubling weights.

i'm now running a new .chs-file for the sports articles, and plan on
ditching the old one when the classifier gets confident on the new
one.

i have not trained enough data to notice ger's comment on big O for
hyperspace, but that could become a problem as the amount of data
increases. that's partly why i would like to do it chris' way and get
rid of old .chs-files every now and then.


On Thu, Jun 11, 2009 at 2:06 PM, Thomas Michael
Hagen<[email protected]> wrote:
> this sounds like a very interesting landscape to explore. my summer
> holiday just started, however, so that'll only happen some time in
> august.
>
> the amount, and quality, of help i've received from this list is very
> impressive. i'm grateful, and i wish you a great summer!
>
>
>
> On Wed, Jun 10, 2009 at 8:53 PM, Bill Yerazunis<[email protected]> wrote:
>>
>>   From: Chris Babcock <[email protected]>
>>
>>   > and 'sport' fill up first, with 'foreign', 'economy' and 'consumer'
>>   > not far behind. this is, of course, a rather stupid short term
>>   > solution, and i'll need some way to refine the learning without
>>   > filling up the .chs-files.
>>
>>   This isn't as stupid as it may appear. You're emulating a natural
>>   strategy for information overload. Admittedly the barrier is an
>>   artificial one, but the strategy is acceptable... providing, of course,
>>   that it works acceptably well.
>>
>>   The next step would be to set up sub-categories when categories fill.
>>   A script that could monitor its own storage, recognize when to stop
>>   training a category, re-analyze its corpus in that category and define
>>   relevant sub-categories would be very powerful. There's no "out of the
>>   box" way to do that with CRM, but it might be worthwhile to do manually
>>   until you can develop a sufficient handle on something like the Python
>>   Natural Language Toolkit (NLTK) or duplicate the relevant functions in
>>   CRM.
>>
>>   Also, if a category is one that evolves then you might employ a
>>   'mentoring' strategy. When a feature file matures, use feedback from a
>>   classify based on that file to train a replacement. Once the
>>   replacement has been trained to an acceptable level, it can be used
>>   instead of the older file.
>>
>> There is actually a name for something similar to this called "Boosting".
>>
>> One implementation of it works sort-of like this:
>>
>> 1) Start learning (say, with SSTTT).
>>
>> 2) At some point, STOP... freeze your learners, and start
>>   a new batch of learners, trained only on the _errors_ of the
>>   previous learners.
>>
>> 3) Again, at some point, STOP... again, freeze the learners, and
>>   start to train a new batch of learners, trained only on the
>>   errors of the previous two sets of learners.
>>
>> ... and so repeat.  The challenge then comes in how you combine
>> the results of these successive generations of learners.  I seem
>> to recall that successive doubling often worked, i.e.
>>
>>   pR_classifier_gen_1 + 2*pR_classifier_gen_2 + 4*classifier_gen_3...
>>
>> and so on, doubling the weight at each generation was effective
>> (but that, I believe, was in a face-recognition task, and not
>> with CRM114.  But it's not that hard an experiment to try.)
>>
>> Another method is to change the weights of individual examples.
>> Unfortunately, none of the CRM114 classifiers right now support
>> other than integer weights (i.e. repeated training with the
>> same example doubles, triples, etc. the weight of that example).
>>
>>     - Bill Yerazunis
>>
>>
>> ------------------------------------------------------------------------------
>> Crystal Reports - New Free Runtime and 30 Day Trial
>> Check out the new simplified licensing option that enables unlimited
>> royalty-free distribution of the report engine for externally facing
>> server and web deployment.
>> http://p.sf.net/sfu/businessobjects
>> _______________________________________________
>> Crm114-general mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/crm114-general
>>
>

------------------------------------------------------------------------------
Enter the BlackBerry Developer Challenge  
This is your chance to win up to $100,000 in prizes! For a limited time, 
vendors submitting new applications to BlackBerry App World(TM) will have
the opportunity to enter the BlackBerry Developer Challenge. See full prize  
details at: http://p.sf.net/sfu/Challenge