Hill Climbing Algorithm & Artificial Intelligence – Computerphile

Hill Climbing Algorithm & Artificial Intelligence – Computerphile


generally when you think about artificial intelligence you are talking about whats the difference between einstein and normal people or between a normal person and a stupid person actually it’s like difference between normal person and a mouse??! or what’s the difference between mouse and a rock?? like that’s the…that’s the difference in ..umm ..between just levels of intelligence within humans which actually …actually doesnt vary that much in the grand scheme of things vs. whats intelligence is in general one way of thinking about what intelligence is ..is as an optimization process .most basic of all optimization process is evolution which …umm..is looking for good replicators in the search space of animals ..right?? ..or creatures..or umm..organisms..yeah..Organisms is the word!


100 thoughts on “Hill Climbing Algorithm & Artificial Intelligence – Computerphile

  1. Your assumption that progress in evolution is dependant only on random chance is flawed. Consider the self organization properties of chaos and periodicity. Learning from experience colours any living thing and helps it to make choices. The recording of that learning in memory is key to evolution and the length of time that one remembers is critical to success. Also, forgetting is even more important; we need to forget maladaptive 'learned' experiences to not keep repeating them.

    Evolution is a lot more than you seem to perceive it is. Dive deep into other sciences and then tie it back to mathematics.

  2. if we define intelligence as the ability to hit a smaller target in a search space in a lesser amount of time, then i'm not sure that humans are necessarily more intelligent.  I think we are predisposed to think in certain (sometimes presumptuous) ways when attempting to solve some problem, be it optimization or otherwise.  but in the case of optimization, i think evolution's inability to premeditate its actions is an advantage, in that it taps into solutions that would be entirely unintuitive to a human. this is demonstrated in biology (like that optimizing slime mold), but also very convincingly in genetic algorithms, which are evolutionary in essence.  the fruits of genetic algorithm-based optimization are often nearly optimal solutions (like those antennae that nasa uses) that a humans probably would not be able to come up with

  3.  I object to this characterisation of evolution. Evolution is not an intelligence, nor is it aiming towards a specific point. There are numerous evolutionary traits that have been invented and reinvented (sight for instance), and characteristics that have gone "backwards" as well as forwards – the evolution of swimming for instance. This talk fails totally at every premise.

  4. Very interesting, though i do not agree fully. Evolution made us that smart. People forget that although we call our products artifical we are natural animals which are (thankfully to evolution) always undergoing evolution. Therefore AI is indirectly a product of evolution. Sry for my english but i hope you get my point. I wish humans started to remember we and therefore all our actions are part oft nature.

  5. The car is a poor example for evolutionary intelligence. The people involved in the industry still haven't designed that brick they are trying to push trough the air on a road from an aero dynamical perspective. But I guess it's logical because car designers aren't rocket scientists…

  6. "Evolution as an algorithm"; never thought along that line, very interesting and illuminating information.

  7. Intelligence in a life-form begins at the molecular level; protein folding, dna replication. Sub-cellular scale. By the time a basic nervous system can develop, a highly intelligent system already exists.

  8. Just to make it clear. Genetic algorithms (evolution) and hill climbing algorithm are different optimization methods. The former is a global method, which in fact can find "the huge hill off in the distance", it doesn't always do, but it can. Others like hill climbing or gradient algorithms are (mostly) local methods. So all due respect, but the guy in the video mixed these things up. Also, for the more complex engineering problems (where simulation is unavoidable) we use genetic algorithms, because the evaluation of the fitness function would be too computationally expensive with other methods. Nature wins so far…

  9. People have developed 'AI' for video games, namely Sethbling, for Super Mario World and Mario Kart. It works in the same way as evolution, with each generation having a fitness level as it plays the game, trying random inputs. The fitness increases depending on how far through the course the AI gets, and decreases based on the time taken to do something. Then, two 'species' from a generation which each have the same fitness level are 'bred' together, resulting in the next generation. This keeps on happening until the AI is an expert at the game.

  10. I don't think evolution gets necessarily trapped in local maxima. Think legs or wings: How did an animal go: I'm gonna sacrifice my front 2 (of 4) limbs to have wings thousands of years in the future? Because it was a little bit better for immediate offspring? Hardly… How did fish come up with legs?* (or worse: lungs**) – It took incredibly long before the mutations were useful… – I think there's actually something "smart" in evolution – not that I understand evolution, just a "feeling".

    (* they might have lived in some shallow puddles, so that's possible, but how would one need to fly? …) (** I know about mudskippers, so ok.)

  11. To avoid being stuck in a local maxima, could a program run a cheap "probe" generation. One that is really stupid and just goes in a single direction for a while to test whether or not it's on a local maxima or a global one? For instance, the new generation could be programmed to have 1 stupid for every 4 intelligent. The intelligent ones would use the info already gained, but the stupid one would be like a radar ping, going off in a spiral to discover the landscape. Then take the data from that probe generation and enhance the current algorithm. Thereby giving the AI a form of insight.

  12. I don't understand why evolution should have a maximum, imagine that the faster an animal is, the longer it lives, the animal will keep getting faster, right ?

  13. Robert Miles sounds really interesting but i can't find a thing about him in google. I would really like to find his youtube channel or find his blog or something (assuming it exist).
    Does any1 know where to find him?

  14. You cannot claim that you are able to find the global optimum of truly complex systems analytically. The nonlinear interactions and their outcome on the fitness landscape are unfathomable. The only way to achieve global optimality in real world, complex systems is through genetic process.
    If there is something better don't worry. The GA will find it.

  15. I just want to say that this video expandend my understanding of… basically "the world" by so much, it's hard to even describe. Thanks for making it. Even at the risk of sounding extremely corny, I'll just go ahead and say that it changed my life.

  16. This is a really good way to articulate why evolution doesn't always result in the best possible outcome and why all animals aren't evolving into perfection all the time, and why we're the only intelligent species. I've tried to explain it in the past, and he totally nailed it with this analogy. It's because intelligence is this huge peak, but it's surrounded by valleys. It took a huge fluke for us to acquire it.

  17. so when you make your step and ask yourself: "am I higher than I was?" how would you know? if you are indeed higher would you even remember your previous step? …. in short : kids…don't do drugs.

  18. What a fantastic way of explaining evolution's "myopia", that 3d graph. It clearly represents both the long-term brilliance and short term limitations of darwinian evolution. Great video.

  19. Computer Science and AI are wrong!
    Quran and Bible are Right!
    Evolution is just a theory!
    Computer Science is nothing but a hypothesis. No evidence that Computer Science is right!

  20. But the entirety of life exists in a space where the possible outcomes are often taken into account. If there exists an open space for supporting life where there is no contestors or where the contestors leave a space open, a species will often adapt to this area. So the entirety of the natural search space is the natural universe. This search space increases with intelligence as they will create search spaces that are not natural (i.e computers). Not written by julian

  21. saying "improvements" can be subjective. A car designer makes changes based on his perception for that moment. While evolution occurs overtime moving back and forward.

    Besides isn't the process of evolution based more on adaptability? I see evolution as a self learning algorithm

  22. The main problem is that Pythagora's theorem on right angled triangles provides negative values of length …

  23. i think thats because evolution doesnt want global maxima ie one particular species to survive. it gives room for local minumum to try out various experiments within and this diversity in species is important as almost all species are interdependent for their survival.
    evolution is a big game which is not limited to only hill climbing algorithm.

  24. I very much enjoy Robert's interesting, insightful & thought provoking ideas, I thought that I would interject however on what evolution is optimizing for (@ ~1:33). Contrary to the statement that it's optimizing for the number of offspring, in fact evolution sometimes performs the opposite function if it is needed to further the lineage. In other words, heredity taken as a whole (DNA,RNA, Ribosome,Epigenetics etc) is self perpetuating and as such, in some cases it may be advantageous to decrease the number of offspring. This is thought to be one factor in the success of mammalian evolution. Some have also argued that in some cases individuals that display self-destructive characteristics are performing some function that ultimately contributes to the success of the whole or a "generalized heredity class" if you will. While many of these statements are debatable, they are submitted simply to further the conversation and promote additional levels of thought on the matter.

  25. What program do you use for your illustrations? The animations and illustrations are quite well done and I would love to be able to experiment and try it out for myself! 🙂

  26. "NATURAL SELECTION IS A POWER INCESSANTLY READY FOR ACTION, AND IS AS IMMEASURABLY SUPERIOR TO MAN'S FEEBLE EFFORTS, AS THE WORKS OF NATURE ARE TO THOSE OF ART."

  27. Mind : BLOWN

    speaking of evolution, after the big bang matter was at its simplest form and then becoming more complex overtime and then become molecule continue to evolve and become even more complex until it became an organism (I skip and simplified thing a lot but you got the point) and organism evolve until becomes us human who then could create something that could evolve faster than evolution could ever be and if so, is then is the meaning of life the universe and everything is to create the most complex being/thing that could possibly exist?

    I'm not convinced that this world is not a simulation

  28. but surrondings can change , basicaly the ''evolutionary topography can change over the time, because well one spiecies usualy doesnt exist in vacuum

  29. A car designer makes a better car than evolution and faster.
    Evolution makes a WAY better car than the car designer, it just takes a couple million years.

  30. That's why I have such a hard time losing weight, my fitness is stuck on the local maximum and I can't seem to find/ get to the global maximum. ;P

  31. Eventually we will end up destroying ourselves in war and the AI will survive and continue to evolve over time and become sentient and a whole world of machines will inhabit the earth. Soon they would start to question their existence and some will claim that they were created and the creator had a strict plan and rules for them to follow. They then start to create artificial intelligence out of carbon material in order to improve their lives.

  32. This is a 2D input – 1D output maximization problem. It's easier to draw the 1D input – 1D output maximization that is found in conventional math and then move to higher dimensions.

  33. I really enjoy your content about AI, do you have any Book puplished or would you recoment one ? I would be cool when its avilable in geman but not nessary.

  34. But isn’t a automobile engineers mind created by evolution?

    An automobile engineers mind has the ability to think and plan. It seems that the human mind is infinitely better, but it is itself a product of evolution.

    Maybe evolution does have the ability to plan and reason

  35. I appreciate these discussions. I will say i'm having trouble understanding how the car maker becomes smarter than process that created them. But I suppose it brings attention to the nature of the 'randomness' of evolution or adaption / intent towards a particular relevant outcome by the individual processes.

    I feel it's bizzare that we can imagine processes in ways that don't/can't happen,but some times it's because of or despite this ability to abstract something entirely 'incorrectly' that we use that to achieve the 'correct' desired outcome. Curious to assume the process that allows us to behave/perform this way doesn't appear to operate in this way itself.

    Basicslly, Are the processes that influence the progression of evolution not at work when a carmaker imagines how not to make something before they make it?

  36. the epistemologist Gregory Bateson (one of the fathers of cybernetics) would say that a human is capable of learning type 0-3 but that evolution is learning 4 because it is a system capable of making systems that do learning 3. in that way evolution is far more intelligent than any particular organism.

  37. Am I missing the point if I assert that the car designer is a part of evolution and therefore it's output is really evolution's output?

  38. IN FACT the larger amount of dimension in a sense makes hill climbing more feasible (in machine learning for example) if your dimensions behave at least somewhat independently. This is because a local minimum or maximum is a place where all derivatives are zero. If you have
    #1:A lot of dimensions, and
    #2:They are not hugely interdependent,
    then almost always there is at least some dimension where the derivative is not zero.

    I've dubbed this "the blessing of dimensionality" as opposed to the curse of dimensionality that is a plague in other areas where dealing with high dimensional data.

  39. If evolution was as fast as ai-development could be, it would destroy itself.
    There is a reason for mistakes.
    If every organism developed to it's highest level, but the environment/rules changed, all organisms would die, because they would be all the same.
    So the way to go might be to acccept some amount of failure as a better solution for different circumstances.

  40. At first glance in the thumbnail I thought it's Guy Martin teaching about computers in his spare time while he's not riding. Haha
    Nice video though

  41. In the words of Ray Kurzweil the definition of intelligence is; measured by only two factors #1 accuracy and #2 speed of resolve …

    the United States education system is built for "neither"… Common Core curriculum in the US … was selected for its enslavement mentality…

    It reduces self-image and self-esteem and accuracy is not a factor in common core…

    So the globalist are educating American children to be non-competitive and enslavement to artificial intelligence is inevitable …

    unless of course the white hats of the US military who put our Great President Trump in successfully take down those that would Enslave Humanity

  42. I disagree that the hill climbing analogy is a good one to illustrate evolution. Evolution is too complex to compare the two. You would have to make modifications to the hills problem.
    For example you would need to have the hills slowly moving up and down.

  43. And because of this, you would want DNA to have a very fractal behavior, allowing for small changes in the root to give huge changes for the organism, allowing you to possibly break out from a local maximum. This is done by inheriting earlier properties is the DNA sequence, for example, all skins humidity is based on a common first introduction of the humidity in very early in the DNA.

  44. I AM SOOOOOOOOOOOOOOOOOOO HAPPY TO HEAR THAT YOU UNDERSTAND HOW EVOLUTION ACTUALLY WORKS>>> OMFG its a full time job just helping ppl understand it let alone trying to correct someone that has it wrong

  45. I suppose it's a bit odd to treat evolution like a hill climbing function, since in reality evolution is an emergent property of biological replication. Over time, biological replication is optimised, but evolution isn't the agent causing this. Evolution is just the name given to the property of gradual change of an organism over time.

  46. This is further confusing the point that most people misunderstand when they are under the impression that evolution somehow an agent making choices. When it’s just what we call the random process that occurs when an organism happens to mutate in such a way that is more successful at mating.

  47. The descriptions of evolution here are fairly inaccurate. Non optimized organisms still reproduce and can step to other worse optimized points on the path to the best optimization. It's just a lot harder.

    Organisms can also jump all over the graph you made, if the graph represents the intensity of a trait. A random mutation means it gets anywhere from almost no to way stronger in one step.

    It does take a very long time though

  48. I now can see what's happening with humans. Today, children are collectively less fit than their parents in almost all domains. I was confused and had started to think that evolution has become disfunctional. Now I undertand that it's because we have hit a 'local maximum' and therefore we need to go downward before we can reach the neighborhood of the 'global maximum' mountain and start climbing again. Genius!

Leave a Reply

Your email address will not be published. Required fields are marked *