Machine Figures Out Rubik's Cube Without Human Assistance (technologyreview.com)

An anonymous reader quotes a report from MIT Technology Review: [Stephen McAleer and colleagues from the University of California, Irvine] have pioneered a new kind of deep-learning technique, called "autodidactic iteration," that can teach itself to solve a Rubik's Cube with no human assistance. The trick that McAleer and co have mastered is to find a way for the machine to create its own system of rewards. Here's how it works. Given an unsolved cube, the machine must decide whether a specific move is an improvement on the existing configuration. To do this, it must be able to evaluate the move. Autodidactic iteration does this by starting with the finished cube and working backwards to find a configuration that is similar to the proposed move. This process is not perfect, but deep learning helps the system figure out which moves are generally better than others. Having been trained, the network then uses a standard search tree to hunt for suggested moves for each configuration.

The result is an algorithm that performs remarkably well. "Our algorithm is able to solve 100% of randomly scrambled cubes while achieving a median solve length of 30 moves -- less than or equal to solvers that employ human domain knowledge," say McAleer and co. That's interesting because it has implications for a variety of other tasks that deep learning has struggled with, including puzzles like Sokoban, games like Montezuma's Revenge, and problems like prime number factorization. The paper on the algorithm -- called DeepCube -- is available on Arxiv.

Comments

Interesting name for a game

[Anonymous Coward]

:D :D :D :D

Traveler's diarrhea - Wikipedia
https://en.wikipedia.org/wiki/Traveler%27s_diarrhea
"Montezuma's revenge (var. Moctezuma's revenge) is a colloquial term for traveler's diarrhea contracted in Mexico."

Re: Interesting name for a game

[Anonymous Coward]

The Other Travelling Salesman Problem.

Wow amazing!

[110010001000]

This is great. Now that we have mastered Chess, Go, and Rubkis Cube all of these "researchers" will put them to work solving meaningful problems. Because AI. Right?

Re:Wow amazing!

[TFlan91]

Games are easy for "AI" because games have strict rules that a modeler can account for/predict.

Re:Wow amazing!

[religionofpeas]

Games are easy for "AI" because games have strict rules

Just because the rules are strict (or even simple) does not mean that the game is easy. You can achieve arbitrary complexity by iterating the rules a large number of times. For example, the rules of Go are strict, the question whether a given board position is winning for white is hard. The rules of a programming language are strict. Writing a Linux kernel is hard. The rules of math are strict. Providing a proof for Fermat's last theorem is hard. The rules of physics and soccer are strict. Making a robot that can beat a human at the game is hard.

Re:Wow amazing!

[gweihir]

Hehehehe, nice. No, not actually AI, just a planning algorithm as being used and researched for something like > 50 years now. This is another instance of machines getting faster, not of them getting any smarter. On the face of it, the Rubic's cube is a very simple problem with a very simple description and a low number of states. Sure, the number of states is actually pretty large when seen absolutely, but for a planning problem, it is not that large and, in particular, the score for a state is downright simplistic: The number of moves to it being solved.

Real-world planning problems are nowhere near that simple. Hence while we will continue to see these stunts, we will not see any real-world problems solved this way for a long, long time, if ever. Also take into account that single-core CPU speed scaling is dead and that most planning algorithms are, in the end, strongly constrained by single-core speeds.

Re:Wow amazing!

[gweihir]

Exactly.

Re:Wow amazing!

[Anonymous Coward]

How about a game
- where you have to figure out classification of a cucumber? https://cloud.google.com/blog/big-data/2016/08/how-a-japanese-cucumber-farmer-is-using-deep-learning-and-tensorflow
- where you have to detect greater than a mild level of diabetic retinopathy? https://www.technologyreview.com/the-download/610853/fda-approves-first-ai-powered-diagnostic-that-doesnt-need-a-doctors-help/

You could also say that games are easy for "humans" (I don't know why you use quotes, but I play along) because games have strict rules. If you put people into unexpected situation you get candid camera -> people really have no idea what is happening and what they should do.

Re:Wow amazing!

[sfcat]

Hehehehe, nice. No, not actually AI, just a planning algorithm as being used and researched for something like > 50 years now. This is another instance of machines getting faster, not of them getting any smarter. On the face of it, the Rubic's cube is a very simple problem with a very simple description and a low number of states. Sure, the number of states is actually pretty large when seen absolutely, but for a planning problem, it is not that large and, in particular, the score for a state is downright simplistic: The number of moves to it being solved.

Real-world planning problems are nowhere near that simple. Hence while we will continue to see these stunts, we will not see any real-world problems solved this way for a long, long time, if ever. Also take into account that single-core CPU speed scaling is dead and that most planning algorithms are, in the end, strongly constrained by single-core speeds.

That's not all that off base but its not quite correct either. You are just missing some context. So this appears to be using a part of AI called Reinforcement Learning (I did my thesis on RL at CMU 20 years ago). So in RL, there are a bunch of techniques for taking a pre-defined domain with a reward function that shapes the behavior learned by the system. There are plenty of algorithms (Value Iteration, Polity Iteration, Q-Learning, etc) for solving a problem with an existing reward function and a discrete set of states (and actions).

But now we want to learn the reward function somehow. This is where the current research is focused. To that end, there are techniques to try to learn this discrete reward function that best supports the learning process (Double Q Learning, Actor-Critic, and others). This research in the article seems to be another technique to learn these reward functions but in this case, there was still an existing model of the game (the states of the cube). But in this case, it seems they just reinvented Value Iteration (which is about 30 years old), slapped a new name on it, but didn't seem to link up their work with where other RL researchers. So just a big yawn...

Re:Wow amazing!

[gl4ss]

it's still far from AI though.

making a robot that runs around after a ball and would fullfill the rules of soccer is indeed hard.

also in this case, the "AI" was shown the wanted end result. it's all very neat except the headline says it solved it by it's own, which would have been a feat if it figured out semi randomly on it's own that this is how it's supposed to be presented for someone to be impressed. after that it's just trial and error loop so remind me again how is this ai?

if you had _Actual_ frigging ai teaching it to play soccer would be fairly easy, but first you would need to ask it to make a robotic vessel for it to play with anyways.

AND SERIOUSLY SOCCER IS THE MOST LEAST STRICT FUCKING PIECE OF GARBAGE ON HIGH LEVEL YOU CAN FUCKING IMAGINE. IT DOESN'T EVEN MATTER IF PEOPLE CAN SEE YOU CHEATING. so what would probably happen is that it would just straight up kill the opposing teams players when the ref isn't looking, because it's fair game if the ref doesn't catch it then and there.

Re:Wow amazing!

[religionofpeas]

after that it's just trial and error loop so remind me again how is this ai?

It's not just trial and error. The Rubik's cube has 10 to the power of 19 combinations, and most of them look like fully scrambled cubes. You cannot randomly try things until you stumble on the solution. The AI part is where it learns the patterns that tell it that it's making progress. Most humans who have come up with a solution to the Rubik's cube start by first solving one side, and then the second layer, and then the top. This AI system has done something similar, except it doesn't work in layers, but came up with its own intermediate stages.

Re:Wow amazing!

[Anonymous Coward]

Providing a proof for Fermat's last theorem is hard.

Non-sense, you only say that because no one ever has!

Re: Yawn

[Anonymous Coward]

Computer are good for sequencing, imagine that. No wait, imagine real news about Stefan Halper instead.

With one exception; the goal state

[Anonymous Coward]

Someone had to tell it what is a solution. If you give it a solved cube, that's assistance. Is it really that hard not to inflate headlines?

Re:With one exception; the goal state

[PPH]

If you give it a solved cube

And you give it a scrambled cube. The AI shouts "Hey look! Haley's comet!" And while you are looking up, it switches them.

Turing test: Passed.

Re:With one exception; the goal state

[Hognoxious]

I did wonder how it decided for itself that the completed cube was somehow "better".

Re:With one exception; the goal state

[Anonymous Coward]

I would argue that is a command. They are telling the computer "Do this task" with no guidance on how to perform the action just specifying the desired result. That is important if the methods are extendable to more complex commands.

Re:With one exception; the goal state

[Anonymous Coward]

Well I did RTFA and the solution presented is clever but not worthy of the headline.

They give the AI a solved cube and this is the goal state. The AI scrambles (randomly I presume) scoring each state as worse and worse. After enough trials, you would be able to score moves in the opposite direction to reach goal state from a random scrambled state.

But you still need to provide some initial state information.

Backwards search basically

[Anonymous Coward]

Yep one way humans solve tasks is to work backwards and forwards. Makes sense to employ an algorithm that works similarly.

Re:Backwards search basically

[PPH]

This is what I was thinking of as well (putting my 37 year old copy of The Handbook of Artificial Intelligence back on the shelf).

Re:Backwards search basically

[sfcat]

This is what I was thinking of as well (putting my 37 year old copy of The Handbook of Artificial Intelligence back on the shelf).

I think you mean Norvig Russell...here is a free PDF Artificial Intelligence: A Modern Approach

Re:Backwards search basically

[PPH]

You couldn't find one older?

Odd definition of "without human help"

[Entrope]

This algorithm was able to figure out how to solve Rubik's Cube with no help from humans other than humans providing the (simulated) cubes, describing what the solution looks like, and designing an algorithm specific to solving Rubik's Cube?

Color me less than impressed.

Re:Odd definition of "without human help"

[Anonymous Coward]

And flipping this around, if you were to provide me with a large database of current configuration and suggested move I could also do it in 30 moves after doing a bunch of database searches.

This research seems especially not exciting for AI because there is a definite right answer and they are just replacing traditional expert systems programming (including designing the algorithm to solve the problem) with a generic algorithm that needs to be trained instead of programmed.

Problems where there isn't a right answer from which to work backwards but you need to get to the best answer amidst a lot of uncertainty are much harder, and deserve more research dollars

Re:Odd definition of "without human help"

[DontBeAMoran]

Color me less than impressed.

What do you mean? White, red, blue, yellow, orange or green?

Re:Odd definition of "without human help"

[Anonymous Coward]

This algorithm was able to figure out how to solve Rubik's Cube with no help from humans other than humans providing the (simulated) cubes, describing what the solution looks like, and designing an algorithm specific to solving Rubik's Cube?

Color me less than impressed.

18374, 183743, 3973, 39483927, x

What is x?

Re:Odd definition of "without human help"

[Anonymous Coward]

Agreed. You can hand a scrambled Rubix Cube to any child and that kid will start attempting to rotate the pieces to make solid surfaces without being told to or shown how to rotate them. That is the I lacking in AI.

Re:Odd definition of "without human help"

[gweihir]

Oh, it still is a nice result. But you are describing exactly the core problem with it: Everything was clear and described in simple, clear statements from the start. That is not how a real-world problem presents itself.

Re: Odd definition of "without human help"

[Anonymous Coward]

X = Y - 1

Re:Odd definition of "without human help"

[Anonymous Coward]

It sounded like the human had to tell the machine what the solved cube looked like. That isn't the problem we are trying to solve. The problem we are trying to solve is for the machine to tell ME that the solution is to make all the sides the same color. And I expect it to tell me it's not possible with the cube I gave it, because I swapped a couple stickers. When I did this in middle school (solving the cube was a raging ego contest in my childhood) the human would detect the fraud (although not usually immediately, which is why it was funny.)

A real AI would take the puzzle given, learn from it, and solve it. Without having to be primed on what the damned puzzle is, or what the solution looks like.

Not impressive.

Re:Odd definition of "without human help"

[religionofpeas]

Everything was clear and described in simple, clear statements from the start. That is not how a real-world problem presents itself.

That's exactly how the problem is described on the box when you buy a Rubik's cube.

Re:Odd definition of "without human help"

[gweihir]

So?

Re:Odd definition of "without human help"

[Anonymous Coward]

I think the point is that the algorithm is agnostic to the relations of the states between the moves. It can be used to calculate an algorithm for solving rubik's cube but it could have been trying to estimate an optimal policy for solving any state space transition given any state space relations (as long as the state space is known)

Re:Odd definition of "without human help"

[parkinglot777]

If you are talking about an algorithm to get from the start to end, it could be interesting. However, this is what I don't see useful is that solution of Rubik is simple a pattern from one to the other. Thus, you need to run the algorithm only once to create links to each of pattern to the solution state. Even though there are 6 faces, it is still possible to store them all. Why? Because that's how those Rubik geniuses do -- memorize certain patterns from one to the other. A computer should be able to do it better than human.

Re:Odd definition of "without human help"

[Anonymous Coward]

Slightly less than that, but mostly yes. Most Deep/Reinforcement learning can be cut into 4 parts : How to see the world (the state), how to interact with the world (the action), how to say if something is good or now (rewards for a given action/state), and some data to train the Neural Network to recognize which action to do in which state.

The first 3 things are usually programmed/quantified by a human : You need to tell him how it can see the world, how it can interact with it, and what is a good action.

Data though can come from anywhere : existing data from human (ads, labeled pictures, public chess games, etc...), or it can be generated (It can play against itself, like in AlphaZero/AlphaGo, or it can be generated data). In this instance, the data was generated by a couple of iterative loops from the 'solved' state of the cube.

After a while, the AI essentially learend what was a 'good' move in many situation, and could generalize it in new situation that it never saw before, and thus, solve cube from an unknown state.

starting with the finished cube

[Anonymous Coward]

wait, if this algorithm was "starting with the finished cube" how did it LEARN anything?

Re:starting with the finished cube

[DontBeAMoran]

The same way rich people "learn" how to become rich.

Re:starting with the finished cube

[Anonymous Coward]

I learned from high school and college to avoid debt and invest. It's not hard to be rich, but you have to work at it for a long time. Most people I know are not able to stay focused like that. They want to spend, not save. This is the irony of being rich. People who are not rich want to be rich so they can spend all that money. But spending money is the behavior that keeps them poor. And the guy who is actually rich is not spending that money until he has so much of it that the only reason to spend it is to keep the pile from growing too big.

GOAP?

[Anonymous Coward]

Sounds like GOAP - goal orientated action planning. You start at the end state then perform actions (in reverse) until you get to the current state. I read the article which doesn't tell you much more about how they did it. It sounds like they brute forced a bunch of moves to build up a tree then used A* on the tree. Then they trained a neural net on the brute-forced solutions. They talk about evaluating how close a cube state is to the goal state, but they don't explain how the AI determined that. It sounds like they hard coded what closeness means, so they're lying when they say the AI worked without human assistance. Without human assistance means they would have needed to use a GA, self-playing, or some other learning method to determine what closeness means. The article's "without human assistance" refers to them not hard-coding any move sequences. I consider that a very far stretch of the word. I'd call hard-coding moves as cheating. If you give it everything it needs, it turns an AI into an algorithm in my mind.

Here's a link to the paper from the article. I don't have the time to read it right now: https://arxiv.org/pdf/1805.07470.pdf

Re:GOAP?

[sfcat]

Sounds like GOAP - goal orientated action planning. You start at the end state then perform actions (in reverse) until you get to the current state. I read the article which doesn't tell you much more about how they did it. It sounds like they brute forced a bunch of moves to build up a tree then used A* on the tree. Then they trained a neural net on the brute-forced solutions. They talk about evaluating how close a cube state is to the goal state, but they don't explain how the AI determined that. It sounds like they hard coded what closeness means, so they're lying when they say the AI worked without human assistance. Without human assistance means they would have needed to use a GA, self-playing, or some other learning method to determine what closeness means. The article's "without human assistance" refers to them not hard-coding any move sequences. I consider that a very far stretch of the word. I'd call hard-coding moves as cheating. If you give it everything it needs, it turns an AI into an algorithm in my mind.

Here's a link to the paper from the article. I don't have the time to read it right now: https://arxiv.org/pdf/1805.074...

Search is the basic AI problem and thus part of AI. And no, they are not using GOAP here (which is just a type of search). It looks a lot more like Value Iteration which is a type of RL (Reinforcement Learning). The point of this research is that it learns to not blindly try random things but learns to search in a more intelligent fashion. But unfortunately for them, their technique was invented decades ago and so just slapping a new name on it doesn't really impress anyone.

Re:Yawn

[infolation]

From TFA:

it has implications for a variety of other tasks that deep learning has struggled with, including... problems like prime number factorization

If it could help with finding the prime factorization of large semi-prime numbers – ie two or more prime numbers that multiplied together result in a target original number - then that would be quite useful.

*cough* cryptography

Bring on the robot overlords

[Anonymous Coward]

This is great. Now that we have mastered Chess, Go, and Rubkis Cube all of these "researchers" will put them to work solving meaningful problems. Because AI. Right?

That would be nice, actually. See, much of our problems in this World is because of humanity's stupid ape brain. It's easily manipulated and we all strive for ape goals.

Dominance. Control. Accumulation of wealth. Fighting over resources. etc ....

Not the things that really make us happy and healthy.

Humanity would be best served by AI overlords.

Re: Bring on the robot overlords

[Anonymous Coward]

All those things helped us get where we are... the dominant species on this planet. We are so dominant that we can easily destroy overselves many times over on this puny little planet... yet we don't.

Really stupid, smart, or really lucky?

Re:Bring on the robot overlords

[PPH]

Dominance. Control. Accumulation of wealth. Fighting over resources. etc ....

Heuristics optimization. It has worked well for us so far.

Humanity would be best served by AI overlords.

And yet in practically every book or movie involving the 'benign' transfer of social control over to benevolent overlords, the humans end up unhappy. Even if not explicitly stated, such control is generally perceived as 'evil'.

Re:Salshdot figures out how to get rid of creimer

[Anonymous Coward]

https://www.youtube.com/watch?...
Cryptofeces Lepidoptera Creimerus infestation is a serious problem. Not only are they capable of reproducing asexually like amoebas, they can also lay eggs hermaphroditically in unexpected places. They can disguise eggs as something useful to fool the unaware, sometimes pretending to be a haiku author, blogger, vlogger, or IT closet cleaner.

Very dangerous. They can seemingly reproduce out of the cosmic background radiation, even if you step on twelve of them, there's always one you miss.

Don't be fooled by the C. Lepidoptera Creimerus's innocuous, rolly-polly, and almost friendly appearance; despite its great size, stupid demeanor, and bedraggled toothless appearance, they have the hardiness of a tardigrade.

Only a concerted, targeted downmodding campaign has been shown effective in controlling this dangerous pest.

Experience shows that stopping such a campaign leads to C. Lepidoptera Creimerus returning within days.

Don't let it happen again!

Re:Yawn

[AHuxley]

Assembling any type of IKEA furniture from the box?

It just created a look-up table.

[Fly Swatter]

Find state of cube and look up the next move, the lookup table could also include every move from that point to solve with very little future work. Granted, creating that table efficiently is where the magic happens.

Re:It just created a look-up table.

[religionofpeas]

The number of possible cube configurations is too large to make a lookup table practical.

Not impressed

[Anonymous Coward]

How close is A.I. in the field of cat-girls?

Long lost

[AndyKron]

Searching for long forgotten Sokoban.exe game now... Found it.

Actually that's a great idea: knapsack problem

[goombah99]

While Ikea furniture is designed with assembly in mind other things are not. Say for example, an airplane. So the assembly process might not be optimal. Letting the computer look for a more optimal process might be useful.

Or more practically, packing items into a shipping box. the famous knapsack problem.

  I hate these slashdot summaries of algorithms. you end up thinking gosh that's stupid. When it's not. just the description is stupid. like a car analogy

Re:Actually that's a great idea: knapsack problem

[religionofpeas]

While Ikea furniture is designed with assembly in mind other things are not. Say for example, an airplane.

I would assume that an airplane designer is considering assembly (and maintenance, which is even harder) in every part of the design.

Re: Actually that's a great idea: knapsack problem

[Anonymous Coward]

I agree. That's the fault of mischaracterizing these algorithms as some kind of sentient consciousness - which the most certainly are not - and tech itself is largely responsible for that. These algorithms are just like other algorithms. It's difficult to have useful conversations amidst all the science fiction nonsense.

Re: Actually that's a great idea: knapsack problem

[Anonymous Coward]

Weight matters a lot for planes. If you can make the plane 5% lighter with twice the assembly time, they go for it.

Re: Actually that's a great idea: knapsack problem

[religionofpeas]

If you can make the plane 5% lighter with twice the assembly time, they go for it.

In order to make the optimal choice, you have to know in advance how much assembly time is required for each of the various options.

Re: Actually that's a great idea: knapsack problem

[zwarte piet]

My guesses are usually in the 1700% ballpark

Brute force method...

[cre1mer]

The fastest way to reset a Rubik's Cube is to pull it apart and reassemble it. That works well on the original Rubik's Cube (3 x 3 x 3). Other variations (4 x 4 x 4, 5 x 5 x5, or 17 x 17 x 17) are increasingly difficult as the parts get smaller.

Re:Brute force method...

[Anonymous Coward]

Failed yer AI test, dummy. Get bigger cubes.

Re:Brute force method...

[gweihir]

Overall, when starting from nothing, this _is_ the fastest way. I figured this out as a teenager. But is also is a way these "AI" things cannot come up with, because they do not actually have any general intelligence. They cannot think "outside of the box" at all. Still lots of applications for this, but replacing a smart person is not among them.

Re:Brute force method...

[Anonymous Coward]

I call BS on this. The fastest official 3x3x3 solve is 4.22 seconds. There's no way you can pull apart and reassemble a cube that fast. Post a video of it if you can. I think what you meant to say is that it's faster to cheat when you are just not very good at solving it conventionally.

Re:Brute force method...

[Anonymous Coward]

We wish we could pull your brain apart and reassemble it into a normal brain.

"increasingly difficult as the parts get smaller."

I can imagine, with those Abominable Snowman fingers of yours.

Embedded FOR loops

[nadass]

So these 'Researchers' are calling an 'autodidactic iteration' what is otherwise known as embedding FOR loops with every iteration added to one of two collections: success or failure. Where is the innovation, exactly? Its increased in performance is attained solely by using newer generations of hardware -- give it older hardware and its performance would comparatively deteriorate.

Games??!!

[PPH]

If ever you've traveled, you know that Montezuma's Revenge is no game!

Something's not adding up

[Dynedain]

Either the article writer didn't understand the whitepaper, or the researchers haven't actually done anything novel.

Having been trained, the network then uses a standard search tree to hunt for suggested moves for each configuration.

This works because the beginning state and end state of a Rubik's Cube are effectively identical. It's the same number of tiles, in a specific arrangement. As humans, we've defined the "solved" state to be all the tiles color-matched to a side. But the "solved" state could just as arbitrarily be any pattern or arrangement of colors across the cube.

Reversing the simulation to work backwards from the "solved" to some specific state of scrambled is exactly the same problem as starting from some specific state of scrambled and trying to get to the solved.

Re:Something's not adding up

[religionofpeas]

The summary and article are quite confusing. They started with solved cubes, and scrambled them in different amounts to generate training data so they could train a neural net to give the most likely candidate turns that would solve the cube.

After the network is trained, you can give it a random scrambled cube, and it will do a Monte-Carlo tree search based on the guidance of the neural net.

Re:Yawn

[Anonymous Coward]

All they did is have it start with a solved cube, build up a database of cube combinations and moves it took to get there, and then have it solve an cube by looking up the shortest move set to get from a solved cube to that combination...

You could do that for any simple problem but it's going to require building a lookup table which is all they really did. This is very old tech with new buzzwords slapped on.

Also, a *machine* is already "human" by definition

[Anonymous Coward]

People always say things like "the computer/machine did X". Like it isn't just a distilled frozen version of the behavior of its designers/programmers.

The entire machine was already built by humans. There is no such thing on this planet, as a machine that does anything without having been assisted to get there by a human.

Novel?

[Anonymous Coward]

This uses to be called the Brute Force method - find all possible solutions - then pick the solutions that fit the test case.
But this time it is novel "In an AI Computer" ?

gweihir, a question... apk

[Anonymous Coward]

See subject & answer this: Did YOU write this? https://tech.slashdot.org/comments.pl?sid=12238544&cid=56791244/ or was some unidentifiable anonymous fool impersonating you as they often do me in some puny attempt to "stir me up" vs. you?

* I've never personally seen you do THAT kind of post (submittal by AC & it was off your usual time-pattern) & as far as trolls doing it, wouldn't surprise me in the least - they constantly use "Bitch Tactics" & never fight a battle themselves, trying to get others to fight one another etc. so they "win" etc. - et al.

APK

P.S.=> That all said & aside, just asking for an honest answer from you... apk

Re:gweihir, a question... apk

[gweihir]

I am going to assume you really are APK.

I very, very rarely post as AC and I never sign it when I do. I also never do it to rile anybody up. I most definitely did not write that posing.

There are some deficients here that cannot stand that I can understand things they are incapable of understanding and that I dare to tell them they are stupid when they have written (again) something extremely stupid. Cowardly, dishonorable and utterly pathetic trolling.

But we already know a shortcut to solve it

[corezz]

There are countless videos online that show the same simple technique that can solve any rubix cube. You repeat the same movements over and over -- even blindfolded -- and its solved. Depending on your hand speed you can solve any combination in under a few minutes. I had my mom, who has never seen a rubix cube, repeat the same steps and she was able to solve it with ease. Then she asked what a Rubix Cube was for/about. 'nuf said.

How did it know what it should look like?

[Anonymous Coward]

Its hard for an AI to figure out what is the correct answer on its own...

It's me & thanks for honesty... apk

[Anonymous Coward]

See subject & I mirror your sentiments on what you said in regard to what you call (lol) "deficients" (I call them the "ne'er-do-well" not-men myself).

* I am assuming you told the truth & take you @ your word. Don't even assume it's me, it IS me & the dorks you describe have been "at me" for years now (since 2012 mostly).

APK

P.S.=> They are EXACTLY what you state & obviously, whoever does it to ME also has some "bone to pick" w/ you & for the reasons you state & THEY WERE TRYING TO "Goad me" into giving you guff!

(I do as you say you do to when others attack me - no "PC" from me then, they do NOT merit that much courtesy & I speak to 'em in a language they understand (wish it could be face to face sometimes using the "ultimate diplomat" which backs all law unfortunately) - crudity RIGHT BACK @ 'EM except loaded w/ truth about them being what we BOTH call them in our OWN ways (losers that can't stand winners & productive hard working guys since THEY just CANNOT compete))... apk

Re:It's me & thanks for honesty... apk

[gweihir]

I have had them from time to time, but apparently I am not interesting enough to rate permanent stalkers. Now I will make damned sure to not ever do any AC postings that resemble this crap, you have my word. Not that I ever intended to do anything like this, it is just completely dishonorable and to me, that counts for something. People sniping from the dark are destroyers of communities and have not place in civilized society.

So if it is AC and claims to be from me, it is not.

Re:Yawn

[Anonymous Coward]

So reverse engineering is AI now..

HS Trig proofs

[guy_scree]

Some trig tests consisted of proposed equalities. You had to determine which ones were valid. My technique was to assume it was true, and work backwards. The teacher objected to my starting off assuming it was true. So I wrote the steps from the bottom of the answer box to the top, announcing I had derived it the proposition from a known equality. Teacher couldn't say a damn thing. Just goes to show the quality of math teaching in the 1950s.

The shits

[Anonymous Coward]

other tasks that deep learning has struggled with, including puzzles like Sokoban, games like Montezuma's Revenge,

It could have the solution to traveller's diarrhea?

It's called a "heuristic"

[Anonymous Coward]

Fancy name -- "autodidactic iteration." It's just another term for a heuristic, a way of characterizing how close a search through a problem space is to the desired solution. They've been used in machine learning and AI for decades. https://en.wikipedia.org/wiki/Heuristic

Old news.

[Anonymous Coward]

Someone already got a Guinness record for making a Rubik's Cube solving machine back in 2016. How is this anything new?

Useless, non-repeatable

[Anonymous Coward]

All this excitement is not called for as long as there is no real chance of replicating this.This is not science.

How to deal with CR!MER on slashdot

[Anonymous Coward]

__,.--=#[How to deal with CRIMER on Slashdot]#=--.,__1)Why-are-people-upset-with-him? 2)What-can-I-do 3)What-are-his-names 4)Who-is-FatCashewsLovesMe 5)How-to-defeat-his-hustles 6)Why-are-there-dashes 7)Pastebin-Copy

1)Why-are-people-upset-with-himHe makes frequent low quality posts for two reasons:
Money) BASICALLY: He made thousands of shitty posts & bragged about how much money it made him.
DETAILS: He wants u to folow his referer links & pick up his cookie. Even if u dont buy what he linked but do buy somthing else from that site later on he often makes money;He ALSO tries to drive TRAFFIC to his BLOGS & vlogs.
Karma)Hi krma helps him disply his ads. He believes karma acumulates infinitely So he makes lots of pointles posts that r not bad enuf to mod down;hoping they wil get moded up;He was a raging ahole when he thought he had a karma surplus

2)What-can-I-do DOWNMOD u wil usually get more mod points. If he is postng from a new sock acount w/ krma, get his oldst posts first. DOWNMOD him and AC in fresh thrads early on;Metmods wil reward u. METAMOD his posts. REPLY ONLY ANONYMOUSLY to the most deeply nested coments in his threds it helps hide his posts. Dwnvote his SUBMISSIONS, he uses to get krma. REPORT HIM to slshdot & the afiliate progrms he is usng. DONT MENTION his 'brand names' c**mer.

3)What-are-his-namesMost famous:Cre1mer Cdre|mer ILoveFatCashews, Anonymous Cashews, The Fat Bastard aka TCDR
4)Who-is-FatCashewsLoveMe AKA Tardu Lardo,FCLM Funny & anoying; Not me or crimer;He keeps lookout for infestation

5)How-can-I-avoid-his-hustles - --===DONT FOLLOW HIS LINKS!!!===-- -
IF YOU MUST:Use a privte tab & nevr buy anything on the same sesion. If he fools u, close tab, cler the cookies for that site. There r sites other than yutube that wil let u watch his videos. I dont know if people view his contnt but I can pictre his jowls jigling at the thot of people subvrting his 'business model'
6)Why-are-there-dashes & weird stuffI know most only skim thse posts. I want the most imprtnt infrmton to pop out at a glnce & to keep it shrt. I dont use TCDRs name becase he may think tht he benfits from geting it indxed by serch engnes. Id lik 2 thnk TCDR & FCLM for editrl advce 7)TardCopy: http://archive.is/HYiM2