[ml] Semi-supervised learning talk/workshop

[Joe Hale]

Hey,

You still need to provide the one class SVM with negative training
examples. I'll be having a go at that later on. I'll out the negative
examples in my dropbox and send you a link.

Joe

On 18 December 2010 14:50, Clay Woolam <clay at woolam.org> wrote:
> I have something else scheduled on the 12th, so I can't do that :/
>
> ps, I couldn't get the one-class svm to converge on the social network data
>
> On Thu, Dec 16, 2010 at 6:15 PM, Joe Hale <joe at jjhale.com> wrote:
>>
>> Hey,
>>
>> I'm going to be on the East cost then - any chance of pushing it
>> forward to 12 Jan?
>>
>> Joe
>>
>> On 16 December 2010 13:48, Clay Woolam <clay at woolam.org> wrote:
>> > I've been meaning to set a time for this, but I can give a
>> > semi-supervised
>> > learning talk sometime in the future.  I can cover the motivations and
>> > assumptions, touch on some approaches like clustering and transductive
>> > SVM,
>> > and mainly talk about graph based label propagation methods. I was
>> > thinking
>> > Wed, Jan 5th would be good. There might also be some tie-ins to the
>> > social
>> > network comp, too.
>> >
>> > _______________________________________________
>> > ml mailing list
>> > ml at lists.noisebridge.net
>> > https://www.noisebridge.net/mailman/listinfo/ml
>> >
>> >
>
>