Cybiote

29 ms·

181.
by Cybiote 10y ago
My favorite part of the second article was how the octopus was able to attend to 3 or 4 people at once and still manage to steal the bucket of fish. It's part hivemind, part single entity and I agree, very alien. What's also inter
182.
by Cybiote 10y ago
This is why I think the ability to ask good questions is a better indication of understanding and intelligence than the ability to generate answers. As a gauge for how far we are from AI you can consider what sort of modeling capacity is re
183.
by Cybiote 10y ago
My point was that you've underestimated the problem difficulty, when the signal is not of a smooth manifold and where the dynamics are far from stationary, it is incorrect to infer that techniques which have worked well in stable and s
184.
by Cybiote 10y ago
There are varying levels of sophistication from which data can be learned from. Pigeons can learn non-trivial word concepts and statistics but they do so at a rate that is much slower than a human infant. Thus far, machines have not done we
185.
by Cybiote 10y ago
I did not actually. I pointed out the approx energy required to get to 1 dan professional then pointed out that a system trained with orders of magnitude more energy was still far less capable. To get to Lee Sedol is still < 3000 MJ (30
186.
by Cybiote 10y ago
There is also a difference between learning and playing. During play, the human operated at ~20 watts on computation while the machine ran at a rate of anywhere from 26,000 watts to 260,000 watts, depending on how efficient the TPUs are (an
187.
by Cybiote 10y ago
I'm skeptical he made up any definitions since it hews quite close to existing ideas. With developments in AI, neuroscience and even physics, this stuff is starting to enter into science proper. The errors of the piece are no worse tha
188.
by Cybiote 10y ago
I see that there is much disagreement with this article but I could not find anything with which I seriously disagreed. I suspect most readers are pattern matching in that the arguments sound very dualist but they need not be; most of the p
189.
by Cybiote 10y ago
Yes, I certainly do think it possible. But it will be more difficult and slower going. Read widely, as the biggest hurdle for a self-learner will be in identifying and filling their gaps of knowledge. Read papers so you learn how to use lan
190.
by Cybiote 10y ago
First of all, let me say that what you all have put together is truly excellent. It covers the sort of things toy treatments and papers leave out or gloss over if you actually want to get something working in the real world. That said, I st
191.
by Cybiote 10y ago
If you want a solid understanding of DL⊂ML, you'll also need a good understanding of probability theory and information theory. While the mechanics require just matrix math and some vector calc, you'll also want to know linear alg
192.
by Cybiote 10y ago
That's WorldBankData. I'm fairly certain that Current US$ is unadjusted. If you use their constant 2010 dollars you get: 96 billion on research in 1970 and 126 billion in 2014
193.
by Cybiote 10y ago
This is astounding! If requests are being taken, it would be useful to be able to search through the listed environments. And a poker environ for the internet section would be a good balance of fun, widely appreciable and a straight forward
194.
by Cybiote 10y ago
As rm999 points out, this book is vastly more useful than limiting distinctions such as supervised/unsupervised learning (what happens in brains is learning while predicting, which is not completely well captured by that delineation, n
195.
by Cybiote 10y ago
The comment which originated this subthread was about telling a bot to get a sandwich and clean the toilet. This implies a single bot and the verbiage implies an agent with strong autonomy. The disadvantage of the specialized machines you s
196.
by Cybiote 10y ago
Making a sandwich or cleaning a toilet is certainly not an easy thing to do. Getting a robot that can understand a scene, any scene and plan motions within it has been very difficult to do. Far more difficult than playing Go at a superhuman
197.
by Cybiote 10y ago
While I do not accept ragebol's notion of real creativity, I very much doubt your definition is correct. The use of ' just ' is too strong, excluding the possibility of influence from other inputs. When humans learn, they do
198.
by Cybiote 10y ago
I'm talking about learning modularly. Children can categorize things they've never seen before by inventing labels on the spot, they're not limited to selecting from a preexisting set (e.g. a lion is a "big cat"). T
199.
by Cybiote 10y ago
Very much so and that's why we don't want our cars to perfectly replicate humans. More sensors plus a limited and narrow AI is better than just vision and a smarter AI.
200.
by Cybiote 10y ago
> People drive reasonably well using vision primarily This is not accurate however. Other important senses in use include proprioceptive, hearing and tactile feedback from wheels. In addition to vision and the improved dynamic range of e
201.
by Cybiote 10y ago
This incarnation of neural networks will never be artists true, but some future iteration of combinations of algorithms (some of which will be informed by current neural networks) likely will produce many great artists per turn. That said,
202.
by Cybiote 10y ago
I think that's about as fair as saying "No, those three lines of code is actually millions of lines of code". After all, it's arbitrary wherever you set the cut off (e.g. it's worth noting that human level intellige
203.
by Cybiote 10y ago
This can also be seen as tight constraints leading to convergent evolution. There's nothing shameful in recognizing when something else's invention is relevant to your needs.
204.
by Cybiote 10y ago
This is an excellent post smallnamespace. The area of research you're looking for is known as evolvability , reading these wiki/wikibook [↓] links is a good start for the topic. Evolution seems to have figured out a way to build
205.
by Cybiote 10y ago
I don't have any disagreement with you--except with your fermi estimate--I've made some of those points before too. In retrospect (and I'd realized it before you posted) I shouldn't have emphasized neuron counts but watt
206.
by Cybiote 10y ago
You see the problem yourself by pointing out the required increase in computational capacity. Quantum mechanics can be described and predicted using fairly straight forward mathematics. Its descriptive complexity is lower than for a cance
207.
by Cybiote 10y ago
I strongly disagree that you can only compare difficulty after the fact. The full class of cancers is an object of lots higher complexity than quantum mechanics. For one, you need a theory of quantum mechanics to fully predict from the deta
208.
by Cybiote 10y ago
You're right! and I've known this but that distinction didn't come to mind as I was writing the post. It's not really relevant though, as my point was about how adaptable and surprisingly intelligent behaviours even thes
209.
by Cybiote 10y ago
> Sort of. Hyperparameter tuning is indeed difficult, but in principle it can be automated and there has been great work on this in the past few years. There's far more to modification than just hyperparameter tuning; this is why pe
210.
by Cybiote 10y ago
I took care to point out deployed models. I did not say the models did no learning. In fact, one can draw an analogy from evolution and instinct to offline training/hyperparameter search/architecture tweaks and deployed model. A

More ›