Posts by BenWilson
Last ←Newer Page 1 2 3 4 5 Older→ First
-
But thank you for spreading that image to the rest of us!
I was hoping to dilute it by spreading it. It's not working, perhaps the only cure will be to drink another thousand lime milkshakes (and not before boarding a train).
-
Oh, and God bless Paul Graham. I'm impressed that using his ideas my system had no problems at all with image spam. I still have no idea why, can only conclude that it picked up on textual tags that were statistically linked to those spams. And like him, I don't care to find out - let maths just do it's magic (until it breaks, of course).
-
I'm sure you didn't, Stephen. Just putting in my 2c for the edumication of the interwebz.
-
bro. just throw away the cup, and get a new one.
LOL. Being a tidy kiwi I was tempted to dispose of it conscientiously, but the idea of carrying a full (to the brim) cup of piss in my hand all the way home to a toilet got less appealing as the minutes passed, and the train entered more populous areas. So it went onto the tracks. But it still plagues my dreams.
-
That giraffe cup is basically one of my favourite things about New Zealand (yes, I am a weirdo).
It used to be for me until an unfortunate experience in Melbourne, boarding the last train from Warragul to the City after getting a bit sloshed. What was a mild desire to urinate quickly became an absolute imperative, but alas no toilets on that train. If I disembarked, it would have been a damned expensive taxi fare home. Fortunately in an empty carriage I discovered a used "Longest Drink in Town" cup. It was just enough.
Now I can't shake the memory. I'm grateful for their awesomely large cup, but I don't want to drink from it any more.
-
btw, I wrote and support a naive Bayesian spam filter. The 'naive' part pissed me off a bit so I looked into some far more complex methods. My conclusion was that naive was just as good on large data samples. And it's really a lot simpler. It's not always a dirty word.
-
Or they say nothing, using a lot of words to do it.
Probably it would be better to train a classifier to work directly on statements of fact. "Taxes will be lowered within 1 year", and then link those to political utterances.
My guess is that there will almost no connection between what they say and what actually happens. As in, it's random rather than lies.
-
Marcus, NZ was purportedly the 5th wealthiest country shortly after WW2 (I can only assume per capita). Unfortunately there has not been another world war so it's been hard to hold on to this position ahead of more industrially advanced and populous nations.
-
Wouldn't some Bayesian voodoo using newsfeeds and stock prices be more profitable? I think you might have a hard time finding statements from politicians that have factual veracity to train off.
-
let the students have their wanderjahr before they decide on what they should do.
I'm still in my Wanderjahren. Die Lust ist nie verloren.
Last ←Newer Page 1 … 971 972 973 974 975 … 1066 Older→ First