Oh! I have a somewhat inconvenient question. I am ok, if you don't answer it. But... Why not work for Elon Musk or the USA government? There are rumors that Uber are owned by Russians and report directly to Putin, jk ; )
Isn't it interesting what is more efficient: neural nets or a learning Mealy machine? Anyway, an optimization of an exhaustive search is a slow but assured way of solving a car driver problem. You don't need the most accurate simulation for it as Elon says here:
A "brute-force" algorithm (an exhaustive search, in other words) is the easiest way to find an answer to almost any engineering problem. But it often must be optimized before being computed. The optimization may be done by an AI agent based on Neural Nets, or on a Learning Mealy Machine.
A Learning Mealy Machine is an finite automaton in which training data stream is remembered by constructing disjunctive normal forms of the output function of the automaton and the transition function between its states. Then those functions are optimized (compressed with losses by logic transformations like De Morgan's Laws, arithmetic rules, loop unrolling/rolling, etc.) into some generalized forms. That introduces random hypotheses into the automaton's functions, so it can be used in inference. The optimizer for automaton's functions may be another AI agent, or any heuristic algorithm, which you like...
Some interesting engineering (and scientific) problems are:
- finding a machine code for a controller of a car, which makes it able to drive autonomously;
- finding a machine code for a controller of a bipedal robot, which makes it able to work in warehouses and factories;
- finding a CAD file, which describes the design of a spheromak working with a guiding center drift generator (hypothetical device, idk!);
- finding a CAD file, which describes some kind of working Smoluchowski’s trapdoor (in some specific conditions, of course);
- finding a file, which describes an automaton working in accordance to the data of a scientific experiment;
- finding a file, which describes manufacturing steps to produce the first molecular nanofactory in the world.
Related work by Embecosm is here: superoptimization.org Though it seems people have superoptimized only tiny programs so far as you can see from the ICRL 2017 paper (App. D): arxiv.org/abs/1611.01787 And loops can also be rolled, not just unrolled. That kind of loop optimization seems to be absent here: en.wikipedia.org/wiki/Loop_optimization
That sounds very different from what Uber is doing here, which is basically accelerating training with synthetic data to accelerate otherwise standard neural architecture search tools. The focus is on the data synthesis network.
Also, the system you describe sounds impractical for any of the complex learning tasks you suggest, especially if it hasn't even done much simpler things yet. Why would machine code be the right level of abstraction for a vision or robotics problem?
> Why would machine code be the right level of abstraction for a vision or robotics problem?
That code would be used to calculate the output function and the transition function of the automaton. At first, as the automaton tries some action and receives a reaction, those functions are constructed accordingly in plain movs and cmps with jmps (suppose x86 ISA here). Then a whole machine code of all actions-reactions is optimized by arithmetic rules, loop rolling and unrolling, etc, so its size is reduced. That optimization may include some hypotheses about Don't Care values of the functions too, which will be corrected in future passes, if they turn out to be wrong... Imagine that code running on something like Thomas Sohmers' Neo processor or Sunway SW26010.
Yeah, it is completely different to Neural Nets. I posted it here because I feel the urge to popularize the idea : ) I am a dilettante in machine learning actually.
Indeed. Unlike many developed countries which have those big box retail stores available in almost every neighborhood, local commerce activities in developing countries are dominated by smaller businesses, and their stores are often distributed randomly across a city.
In the U.S, you kind of know ahead what brands of retail stores you will be able to find a product, as opposed to some other countries where it's a lot more difficult to locate them due to the lack of specialized retail stores.
Isn't it funny that people still don't use a messenger with decent filters, which assess personal ratings of posters to show you more relevant results first? Even Discourse doesn't have personal blacklists. Though its search is more advanced than Reddit's. Isn't that strange in 2019?
Also even popular people like Guido van Rossum can't get enough funds to improve compilers of such an important language as Python [0]. And current patent system doesn't support developing of software tools because everything is already owned by some patent trolls. For instance, why do they still not use AI agents [1] in machine code optimization? I think, it is not a coincidence. Just read recent news about Chinese hacking the USA hardware. Along with Russians they just dumb you down.
Dropbox abandoned their incompatible python 2 jit compiler. Pypy remains available and there was really no point in having another around.
And Guido had nothing to do with that decision afaict.
Finally, the reason why they dropped it is absolutely correct. Python isn't meant to be used for performance critical work. Python is however fast enough for most usecases - especially with pypy, but definitely not all.
Why not have single language for all kinds of work? And why not spend some tiny part of those money dedicated to improving hardware by 2x-3x, when there are people [0] claiming Python run-time has potential to 1000x speed-up? I know, I know, it will make manual optimization job nonsense, and everyone needs a job nowadays it seems...
Yes, you can have a substantial performance boost for python by rewriting it fully in c... But that's exactly like saying that c performance is substantially better than Python's.
...with R&D price tags attached. Then everyone who uses something a court considered similar must pay his contribution to the price of the oldest patent involved.
Exactly. Poor Western Europeans are robbed by their corrupted governments, IMHO. In Ukraine programmers pay only 5% tax, when everyone else - around 40%. So, come to us! : )
Isn't it kind of exagerating to call western EU governments corrupt and invite investors into Ukraine, a country that struggles to get out of Russia's grip and had several swift regime changes, a poisoned president and an ongoing war in the last 20 years? Also, tax extemptions for a job sector that isn't threatened in any way are discriminating.
Well, yeah, unfortunately majority of Ukrainians are not smart enough to overcome those problems (mainly with Russia). But even without IQ qualification for voting, I hope, they will be able to rule over the state right. I don't participate in voting as you can guess from my profile...
Regarding the taxes. They are not that discriminating! Because in absolute values other workers may pay to the state even less than programmers. A good programmer may earn up to $3000 every month, so $3000·5% = $150 is what he gives to the state in taxes. An ordinary worker earns (official stat) $350 a month, so he gives only $350·40% = $140. I am not sure about 40% though...
We also have very low prices here too. For instance, potatoes cost only ¢15/kg, so if you are European or American you are welcome to spend your money here : )
So in absolute values poor people are ripped off (live off $210) while IT workers are well off (live off 2850$). How is that not discrimination? A flat rate of 19% for everyone, like Poland has is quite fair in comparison. Other countries in Central and Eastern Europe have cheap food producs as well:
That is because they are doing useless jobs or work which is not yet in demand here. If they invested in education or some equipment which were raising in demand, they would be fine. Many of ordinary workers in Ukraine emigrate to the EU actually - even more would do that if the EU did not establish those discriminating boundaries with visas : )
But the main reason of their poor life is corrupted government which have been here for many years and which didn't give them opportunitytoinvest properly (not everyone can be a programmer, I admit). We had several hyper-inflation periods with banks going bankrupt, with too complicated legislation no one seriously followed, unfair privatization of former state property, policemen killing and robbing people, etc.
...€2.32 for 1 kg of potatoes in Norway?! I got a millionaire idea.
Take now... some hard-headed business man, who has no theories, but knows how to make money. Say to him:
"Here is a little village; in ten years it will be a great city; in ten years the railroad will have taken
the place of the stage coach, the electric light of the candle; it will abound with all the machinery and
improvements that so enormously multiply the effective power of labor. Will in ten years, interest be any
higher?" He will tell you, "No!" "Will the wages of the common labor be any higher...?" He will tell you,
"No the wages of common labor will not be any higher..." "What, then, will be higher?" "Rent, the
value of land. Go, get yourself a piece of ground, and hold possession." And if, under such circumstances, you
take his advice, you need do nothing more. You may sit down and smoke your pipe; you may lie around like
the lazzaroni of Naples or the leperos of Mexico; you may go up in a balloon or down a hole in the ground;
and without doing one stroke of work, without adding one iota of wealth to the community, in ten years you
will be rich! In the new city you may have a luxurious mansion, but among its public buildings will be
an almshouse.
I would add: "Or get some intellectual property everyone must buy." /s
Oh, I am neither against IP, nor in favor of buy-out of all natural resources. I just think it would be good if each IP had reasonable price [0] and everyone had a land for subsistence.
