davmre

26 ms·

181.
by davmre 12y ago
> Automatic machine translation, and query expansion was supposed to make these inter-lingual barriers go away. But it feels like in many ways, the effort has died. Machine translation is still very much an active area of research. It su
182.
by davmre 12y ago
The great thing about math research, or really research in any field, is that it's highly parallelizeable. There are a few geniuses like Terry Tao exploring specific directions of thought, but there are so many things to think about
183.
by davmre 12y ago
> Not so much what n and k to set, but how to get such an entity to do anything at all. I don't understand what you think the problem here is. Computers are machines that run programs; my laptop doesn't require an evolved self
184.
by davmre 12y ago
There are pretty good arguments that the goals of superintelligent AI will not, by default, be aligned with human values unless we specifically work to make them so. The Russell piece I linked makes (a very short version of) this argument,
185.
by davmre 12y ago
Expert opinion is not unified on this. E.g., Stuart Russell: "A highly capable decision maker – especially one connected through the Internet to all the world's information and billions of screens and most of our infrastructure –
186.
by davmre 12y ago
For those claiming that no reputable AI researchers believe there's a serious risk, I'd like to point to this short reflection by Berkeley professor Stuart Russell (disclaimer: my PhD advisor): http://edge.org/conv
187.
by davmre 12y ago
It's kind of unfortunate that the spectre of "AI doom" has blown up in the popular press the way it has, in that it's forced a lot of top people to push back against caricatured, sensationalist arguments in a way that di
188.
by davmre 12y ago
That doesn't change the calculus because the offset applies equally to both sides. Suppose you had paid the workers the same amount to build schools instead. They would still be able to buy houses with their salaries, but now there w
189.
by davmre 12y ago
The comparison is to humans that see the same data as the algorithm. The "true" labels can be assigned using additional data. For example, if I make a dataset for the task of distinguishing cocker spaniels from springer spaniels,
190.
by davmre 12y ago
Not sure if this would solve your issues, but next time you try Firefox you should check out Instantfox ( https://addons.mozilla.org/en-us/firefox/addon/instantfox/ ). It implements a Chrome-like unified U
191.
by davmre 12y ago
Facebook gets all of the value produced by their own developers, but the benefits of GPG are shared. If every big tech company matched Facebook's $50k, the total sponsorship would be equivalent to several engineer salaries.
192.
by davmre 12y ago
Aside from making conceptual sense, using the original value has the nice property that there's no dependence between the calculation of separate pixels, so the filtered image can be computed fully in parallel. GPUs are very good at th
193.
by davmre 12y ago
The point is that they worked to establish the norm that pedestrians must defer to cars on city streets, instead of cars deferring to pedestrians. This makes it viable to actually get somewhere by driving a car in the city, but at the cost
194.
by davmre 12y ago
You can donate to keep software bug-free by funding research into safer programming languages (e.g., Rust), tools for formal verification and static analysis, testing frameworks, software engineering methodology, etc. Sure, someone could al
195.
Obama calls for municipal broadband(vox.com)
520 points by davmre 12y ago|275 comments
196.
by davmre 12y ago
Just to avoid confusion: Norvig is the co-author of AIMA, the AI textbook. The research priorities document that is the subject of this thread was drafted by Stuart Russell, Daniel Dewey, and Max Tegmark, though Norvig (along with many othe
197.
MOOCs are closed platforms and probably doomed(lemire.me)
56 points by davmre 12y ago|27 comments
198.
by davmre 12y ago
NIPS gets a ton of submissions, so the law of large numbers governs pretty strongly. Imagine that each paper submitted is independently either good or bad, with 22.5% probability of being good. With 1660 submissions, the total number of goo
199.
by davmre 12y ago
To be fair, calling it "Neural Information Processing Systems" isn't significantly more informative. The name is just a quirk of history; NIPS in its modern form includes research in all areas of machine learning, not just ne
200.
by davmre 12y ago
There are different ways to be "good" at math. I have a few friends who are very good at contest type math (e.g. well-ranked on the Putnam) and I also have friends doing math PhDs at top schools. They are mostly not the same peopl
201.
by davmre 12y ago
I interpreted the article as making exactly the opposite of your point: because Khan Academy is limited to automated grading, they can only ask standardized-test style questions: solve an equation, select a multiple-choice answer, etc.
202.
Warm West Coast Reception for China’s Web Czar(bits.blogs.nytimes.com)
1 points by davmre 12y ago|0 comments
203.
Wikipedia’s ‘complicated’ relationship with net neutrality(washingtonpost.com)
108 points by davmre 12y ago|85 comments
204.
Rationality and Intelligence: A Brief Update [pdf](cs.berkeley.edu)
47 points by davmre 12y ago|1 comments
205.
by davmre 12y ago
Geoff Hinton gave a talk last week at Berkeley on exactly this problem - in pixel space, object identities are all tangled up with location/pose information in a very nonlinear way; it would be nice to find a representation that actual
206.
by davmre 12y ago
For those unfamiliar with the reference (from http://www.ams.org/notices/200410/fea-grothendieck-part2.pdf ): One striking characteristic of Grothendieck’s mode of thinking is that it seemed to rely so littl
207.
Gift from Ballmer Will Expand Computer Science Faculty at Harvard(bits.blogs.nytimes.com)
2 points by davmre 12y ago|0 comments
208.
Big Data, Hype, the Media and Other Provocative Words to Put in a Title(amplab.cs.berkeley.edu)
15 points by davmre 12y ago|1 comments
209.
HBO will let you watch their shows online – without having to buy cable(vox.com)
1 points by davmre 12y ago|0 comments
210.
by davmre 12y ago
> Speaking like genderless objects addressing other genderless objects with genderless pronouns and genderless generalizations doesn't really help anyone. If you're speaking in the first person, about your own experiences, it&#

More ›