jf-

32 ms·

271.
by jf- 8y ago
I think it’s important to bear in mind that most machine learning models still aren’t very good, and the main problem you’ll encounter with them isn’t bias, it’s whether or not they’re producing gibberish. In this thread we seem to have ass
272.
by jf- 8y ago
Why don’t you ask them?
273.
by jf- 8y ago
It’s a trope that’s supposed to show that google searches are biased in favour of black people. If you search for American inventor you get lots of image hits for black inventors. The real reason for this is that American inventor is a subs
274.
by jf- 8y ago
The “American Inventor” result is easily explained by being a substring of “African American Inventor”. No conspiracy necessary, it’s just a substring hit.
275.
by jf- 8y ago
Indeed, I know what a bias is and what the common cognitive biases are, though they always bear repeating. I’m referring to a specific statement at the end of the video lecture, where the viewer is encouraged to validate ML results against
276.
by jf- 8y ago
The video lecture calls for validation of ML results with regard to “appropriate social behaviour”. Is this actually a call to hide results that are politically inconvenient? A concrete example would help here.
277.
by jf- 8y ago
Depends on how needlessly argumentative your team is. Generally you can reach a consensus through discussion, example and experimentation. We do have intuition, generally we know what simple looks like when we see it, likewise we recognise
278.
by jf- 8y ago
Well I wasn’t quoting Einstein, but sure. I would err on the side of less rather than more.
279.
by jf- 8y ago
Software developers are systematisers by default. We tend to value complexity for it’s own sake, hence the over-engineering common to software projects. The methodologies we use fall victim to the same tendency. We build complex, rigid rule
280.
by jf- 8y ago
This may be an unpopular opinion, but I’d prefer if this culture war crap were kept off HN. There are plenty of other places to discuss this, just because Twitter is involved doesn’t mean it’s related to tech.
281.
by jf- 8y ago
Maybe they think it’s funny?
282.
by jf- 8y ago
This is the most important bagel-related news I’ve read all day.
283.
by jf- 8y ago
Thanks for that. I really did want readers of my comment to look him up. There’s a reason Chomsky describes his work as being without value, but now we’ve just got two appeals to authority going. I’d rather people look up some Zizek on YouT
284.
by jf- 8y ago
I’ve said what I’ve said, and I’m no fan of Zizek, to whatever extent I’ve been exposed to him. Take what you want from it, I’m not arguing further.
285.
by jf- 8y ago
Zizek is a purveyor of nonsense. I think it’s obvious to most readers that I have a belief system, and that I am advocating for that belief system, and that belief system is roughly described as centrist. You’ve written rather a lot of word
286.
by jf- 8y ago
It’s already peaked and has started to decline. There’s a definite sense that questioning the radical left is less transgressive than it used to be. We’re currently coming out of a period which will be looked back upon as being very left, a
287.
by jf- 8y ago
Thanks!
288.
by jf- 8y ago
No worries! Nothing is ever perfect on launch day. It’s a nice app you have here. My gut instinct is that you will need to allow under the hood customisation if you want it to really catch on. Your potential customers will be afraid of gett
289.
by jf- 8y ago
Also can’t find a valid username/password for Hours. There is what looks like a username (hello world) and password in the source but they don’t let me log in.
290.
by jf- 8y ago
Tutorial has some issues. “from_scratch.migration” should be “from_zero.migration”, and visiting server_ip:7584 gives an invalid address error. This is after downloading the virtual box image, installing Alan connect, doing the build steps,
291.
by jf- 8y ago
No. You’ve had my opinion, I’m not arguing with you, I’m telling you.
292.
by jf- 8y ago
No, the left has recently adopted anti-centrism, which is just a form of the ultimatum “you’re either with us or against us”. Attacking the center to try to force moderates to lean further left. I don’t agree with everything the Republicans
293.
by jf- 8y ago
People today seem to think this level of incivility is not only acceptable, but a virtue.
294.
by jf- 8y ago
It is both sides. How grounded in reality you think either are depends on your own biases. From my point of view both the left and the right have become more extreme and more removed from the truth. The left thinks that the right is distort
295.
by jf- 8y ago
Google are an incredibly sinister privately run surveillance system, and are pushing AMP to EEE the web, but you know, MS are still the bad guys, because we’re stuck in 2003.
296.
by jf- 8y ago
The same way you know that you walked out of your front door this morning without it having to be published in a scientific journal.
297.
by jf- 8y ago
Plenty of people try to replicate, in trying to build on work from other labs. What they don’t do is publish when replication fails, but the failure to replicate usually becomes known in the community.
298.
by jf- 8y ago
Engineering and science have an advantage, however, in that they are empirical and readily tested. Things evaluate to true or false, even if doing the evaluation is difficult. In the humanities, things often come down to value judgements an
299.
by jf- 8y ago
Withdrawing the nomination just sends the message that due process doesn’t exist. I don’t want him confirmed, but I’m not sure that public rejection of the presumption of innocence by the US government is a price worth paying. Rock and a ha
300.
by jf- 8y ago
Seems like a no-win scenario, whether he’s confirmed or not, everyone loses.

More ›