Interesting.
i'd probably be interested to try it out.
On Thu, Apr 28, 2011 at 11:31 PM, Stanley Xu <wenhao.xu@gmail.com> wrote:
> Thanks Ted and Lance. And sorry for the jargon.
>
> For the delay Ted mentioned, we have already considered that, still thanks a
> lot for all the detail ideas, they were pretty helpful.
> For the parallelized SGD, just found a new paper about using DSGD in matrix
> factorization, it's different from logistic regression, but might helpful as
> well. Put the title here "LargeScale Matrix Factorization with Distributed
> Stochastic Gradient Descent" if anyone is interested.
>
> Best wishes,
> Stanley Xu
> On Fri, Apr 29, 2011 at 2:08 PM, Ted Dunning <ted.dunning@gmail.com> wrote:
>
>> Yes.
>>
>> Apologies for jargon and TLA<
>> http://en.wikipedia.org/wiki/Threeletter_acronym>
>> 's
>>
>> On Thu, Apr 28, 2011 at 7:04 PM, Lance Norskog <goksron@gmail.com> wrote:
>>
>> > CTR == Clickthrough Rate
>> >
>> > On Thu, Apr 28, 2011 at 12:06 PM, Ted Dunning <ted.dunning@gmail.com>
>> > wrote:
>> > > On Tue, Apr 26, 2011 at 8:00 PM, Stanley Xu <wenhao.xu@gmail.com>
>> wrote:
>> > >
>> > >> ... I understood as the algorithm, the time in training only relies
on
>> > the
>> > >> nonzero records, but per our test, there would be some overhead we
>> > could
>> > >> not ignore for thoso nonzero records, though the cost is sublinear
>> or
>> > >> logit to the length of the hashed vector.
>> > >>
>> > >
>> > > This is pretty close if we say "nonzero values". A record usually
>> > refers
>> > > to an entire training
>> > > example.
>> > >
>> > > The extra work refers mostly to deferred regularization that eventually
>> > has
>> > > to be
>> > > applied. My guess is that it is even less than log in the feature
>> vector
>> > > size.
>> > >
>> > >
>> > >> And in CTR prediction, I am not pretty sure it will converge very
>> > quickly.
>> > >>
>> > >
>> > > I was saying this purely based on the number of features.
>> > >
>> > >
>> > >> Because we will very possibly see some records has the almost same
>> > feature
>> > >> but different result in display ads.
>> > >
>> > >
>> > > The algorithm can still converge to an estimate of the probability
>> here.
>> > >
>> > >
>> > >> But we will see the result in the
>> > >> future.
>> > >
>> > >
>> > > You have to be *very* careful about this to avoid prejudicing the model
>> > > against
>> > > recent impressions. If you have a fast feedback to the ad targeting
>> > system,
>> > > you
>> > > can have severely instability.
>> > >
>> > > The key thing that you have to do to avoid these biases is to define a
>> > > maximum
>> > > delay before click for the purposes of modeling. You need to ignore
>> all
>> > > impressions
>> > > younger than this delay (because they may still get a click) and you
>> need
>> > to
>> > > ignore
>> > > all clicks after this delay (to avoid bias in favor of old
>> impressions).
>> > > For online ads
>> > > you can probably use a maximum delay of a few minutes because most
>> clicks
>> > > will
>> > > happen by then.
>> > >
>> > > To find a good value for maximum delay, you should plot the CTR for a
>> > bunch
>> > > of
>> > > ads versus delay. This will increase rapidly shortly after zero delay,
>> > but
>> > > then will
>> > > level off. The ordering of ads by CTR is what you care about so you
>> can
>> > > follow the
>> > > curves back and find the shortest delay where the ordering is clearly
>> > > preserved. Use
>> > > that as your maximum delay. Typically this is roughly where your CTR
>> is
>> > at
>> > > about
>> > > 8090% of the final value.
>> > >
>> > >
>> > >
>> > >
>> > >> (We were still working on creating a framework to digg all the
>> > >> features we need from the log, I would like to share our experience
by
>> > >> using
>> > >> Mahout SGD once we got our CTR prediction model release.)
>> > >>
>> > >> And for parallelize SGD, what do you mean for help with sparse inputs
>> > that
>> > >> exhibit longtail frequency distribution? Would you like to share some
>> > of
>> > >> your ideas, Ted?
>> > >>
>> > >> Currently, what I could think about is split the data to multiple
>> mapper
>> > >> randomly and let every mapper to learn from the local data and get
an
>> > >> average on the whole model, or let multiple model to vote for every
>> > >> feature's weight. A little like the idea of AdaBoost or RandomForest.
>> > But I
>> > >> am not a scientist or mathematician, so no idea if it is correct or
>> not.
>> > >>
>> > >>
>> > >> Thanks so much.
>> > >> Stanley Xu
>> > >>
>> > >>
>> > >>
>> > >> On Tue, Apr 26, 2011 at 11:16 PM, Ted Dunning <ted.dunning@gmail.com>
>> > >> wrote:
>> > >>
>> > >> > On Mon, Apr 25, 2011 at 11:46 PM, Stanley Xu <wenhao.xu@gmail.com>
>> > >> wrote:
>> > >> >
>> > >> > > 1 hour is acceptable, but I guess you misunderstand the data
scale
>> I
>> > >> mean
>> > >> > > here. The 900M records didn't mean 900M Bytes, but 900M lines
of
>> > >> training
>> > >> > > set(900M training example.). If every training data has 1000
>> > dimension,
>> > >> > it
>> > >> > > means 900 million X 1000 X 16 B = 14TB. If we reduce the
logs
>> > collected
>> > >> > to
>> > >> > > 14 days, it would be still 23TB data.
>> > >> > >
>> > >> >
>> > >> > Oops. Forgot that last multiplier.
>> > >> >
>> > >> >
>> > >> > > Per our simple test, for 1000 dimension, 10M lines of record,
it
>> > will
>> > >> > take
>> > >> > > about 12 hours to do the training, so 90M lines of data
will cost
>> > at
>> > >> > least
>> > >> > > 90 hours, is that correct?
>> > >> > >
>> > >> >
>> > >> > 10M x 1000 x 8 = 80 GB.
>> > >> >
>> > >> > 12 hours = (approx) 5000 seconds. So this is
>> > >> >
>> > >> > 80 GB / 5000 s = 80/5 MB /s = 16MB / s
>> > >> >
>> > >> > Yes. This is reasonable speed. I think you can get a small
factor
>> > >> faster
>> > >> > than this with SGD. I have seen 100 million records with more
>> > nonzero
>> > >> > values than you describe with a training time of 3 hours. I
would
>> not
>> > >> > expect even as much as a factor of 10 speedup here.
>> > >> >
>> > >> >
>> > >> > >
>> > >> > > And from the PPT you provided
>> > >> > > http://www.slideshare.net/tdunning/sdforum11042010
>> > >> > > You said it would take less than an hour for 20M data records
for
>> > >> > > numeric/category mixed dimensions. I am wondering, how many
>> > dimensions
>> > >> > per
>> > >> > > record?
>> > >> > >
>> > >> >
>> > >> > These are sparse records records with about a thousand nonzero
>> > elements
>> > >> > per
>> > >> > record.
>> > >> >
>> > >> >
>> > >> > But let's step back to your data for a moment. Where do these
>> > thousand
>> > >> > dimensions come from? Do you really have a thousand handbuilt
>> > features?
>> > >> > Do you not have any sparse, textlike features?
>> > >> >
>> > >> > If you really only have a thousand dimensional problem, then I
think
>> > your
>> > >> > model might exhibit early convergence.
>> > >> >
>> > >> > If not, it is quite possible to parallelize SGD, but this is only
>> > likely
>> > >> to
>> > >> > help with sparse inputs that exhibit longtail frequency
>> distribution.
>> > >> >
>> > >>
>> > >
>> >
>> >
>> >
>> > 
>> > Lance Norskog
>> > goksron@gmail.com
>> >
>>
>
