Slashdot Mirror


Machine Learning Spots Treasure Trove of Elusive Viruses (nature.com)

Artificial intelligence could speed up metagenomic studies that look for species unknown to science. From a report: Researchers have used artificial intelligence (AI) to discover nearly 6,000 previously unknown species of virus. The work illustrates an emerging tool for exploring the enormous, largely unknown diversity of viruses on Earth. Although viruses influence everything from human health to the degradation of trash, they are hard to study. Scientists cannot grow most viruses in the lab, and attempts to identify their genetic sequences are often thwarted because their genomes are tiny and evolve fast.

For the latest study, Simon Roux, a computational biologist at the DOE Joint Genome Institute (JGI) in Walnut Creek, California, trained computers to identify the genetic sequences of viruses from one unusual family, Inoviridae. These viruses live in bacteria and alter their host's behaviour: for instance, they make the bacteria that cause cholera, Vibrio cholerae, more toxic. But Roux, who presented his work at the meeting in San Francisco, California, organized by the JGI, estimates that fewer than 100 species had been identified before his research began. Roux presented a machine-learning algorithm with two sets of data -- one containing 805 genomic sequences from known Inoviridae, and another with about 2,000 sequences from bacteria and other types of virus -- so that the algorithm could find ways of distinguishing between them.

28 comments

  1. Where's the real intelligence? by Mister+Liberty · · Score: 1

    Between Humans, AI, and Viruses?

    1. Re:Where's the real intelligence? by Anonymous Coward · · Score: 0

      We need to stop referring to these things are "artificial intelligence"; they're "pseudo-intelligence", at best.

    2. Re: Where's the real intelligence? by Anonymous Coward · · Score: 0

      Pseudo intelligent. Like many humans.

    3. Re: Where's the real intelligence? by Anonymous Coward · · Score: 0

      What are you saying? That you actually believe there is a mind inside that box? That these self-driving cars are capable of carrying on a conversation about, I dunno, the news of the day with you, while it drives you to work? Or are you scoffing at people who do believe these machines they keep trotting out are actually 'artificial intelligences'? If you're scoffing at me then you're barking up the wrong tree. An amoeba is more intelligent in ways that matter than any of these so-called 'AIs' we keep hearing about. It's all hype.

    4. Re: Where's the real intelligence? by rogoshen1 · · Score: 1

      I'd say an amoeba is on par with these machines.. it's basically just a stimuli -> output type situation, which these machines can approximate pretty well.

      Had you said 'rodent', or 'dog', i'd totally agree with you. But makes one wonder, how far down the evolutionary tree do you have to go before you get anything approximating a 'mind' ? And when / what caused that change to occur.

    5. Re: Where's the real intelligence? by Anonymous Coward · · Score: 0

      What are you saying? That you actually believe there is a mind inside that trump? That these self-driving trumps are capable of carrying on a conversation about, I dunno, the news of the day with you, while trump drives you to work? Or are you scoffing at people who do believe these trumps they keep trotting out are actually 'artificial intelligences'? If you're scoffing at me then you're barking up the wrong tree. An amoeba is more intelligent in ways that matter than any of these so-called 'trumps' we keep hearing about. It's all hype.

    6. Re: Where's the real intelligence? by Anonymous Coward · · Score: 0

      I was actually drawing the amoeba reference from an article I'd just read minutes before -- but yes, even a mouse has more of a mind, more 'intelligence' than these machines do.

    7. Re:Where's the real intelligence? by Dutch+Gun · · Score: 0

      We need to stop referring to these things are "artificial intelligence"; they're "pseudo-intelligence", at best.

      There's no intelligence at all. Not even pseudo. "Machine learning" seems like a better fit for these algorithms. They're really more about fuzzy pattern matching than any sort of logical deduction or reasoning, as expected of "intelligence."

      --
      Irony: Agile development has too much intertia to be abandoned now.
    8. Re: Where's the real intelligence? by orlanz · · Score: 1

      Humm, this article says ameobas are smarter. Maybe that gap has gotten smaller in the last 2 years.

      https://www.npr.org/sections/1...

    9. Re: Where's the real intelligence? by Anonymous Coward · · Score: 0

      You need professional mental help

    10. Re:Where's the real intelligence? by HeckRuler · · Score: 1

      The search-space presented by all these repetitive arguments for why AI doesn't exist would bore an AI to SHED_TEARS.sh

    11. Re:Where's the real intelligence? by mikael · · Score: 1

      "feature recognition" would be another term. These system look at petabytes of data (individual genomes, gene interactions) and try and identify which mutations or interactions correlate to a particular genetic problem. No human could print out all this data, sit down at a desk and compare then line for line.

      Imagine taking millions of lithographic electron microscope photographs of CPU/GPU cores, then trying to corrrelate those to broken logic circuits. Eventually, you would be able to associate particular regions with particular faults.

      --
      Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
    12. Re:Where's the real intelligence? by religionofpeas · · Score: 2

      Dictionary says: "The ability to acquire and apply knowledge and skills.". The machines are doing that, so it's not unreasonable to call them AI.

    13. Re: Where's the real intelligence? by shaitand · · Score: 1

      Alphazero can read the rules for a game and then beat pretty much anyone at it. That isn't Pseudo intelligence, that is intelligence.

    14. Re: Where's the real intelligence? by eaglesrule · · Score: 1

      An example could be captive fish that learn to recognize the source of their food and become excited in anticipation just prior to feeding. Non vertebrates like the Portia spider display complex behavior too, like learning from trial and error, but I hesitate to call that any kind of 'mind'.

    15. Re: Where's the real intelligence? by Stickasylum · · Score: 1

      Alphazero might be more general than human-tuned algorithms for specific games, but it is still very much restricted to finite games (or games where the infinite component can be ignored), and relies heavily on the expert encoding of the rules to aid learning. The rules are not simply "read" by alphazero, the boardspace and legal moves are built into the network in a very specific way to facilitate learning. Then it requires truly massive computing power to train to a *specific* game, and additional massive power to play against outside opponents, during which it is no longer learning. Alphazero is an impressive feat, but more as a technological step showing that hefty computing power can allow more generalized neural networks can be used to produce good results in these sorts of finite games. It's definitely not crossing any AI thresholds.

  2. The singularity is coming. by BlueCoder · · Score: 1

    Looks like someone beat me to the punch at bring up the question of "AI" or pseudo intelligence.

    Many smart people are shouting the dangers of AI at the moment when we couldn't be further from it. First of all your have to define what intelligence is with whom you start having a discussion first. For instance an I.Q. test can easily be passed by a knowledge engine with a high score but it's hardly a "mind".

    To me a "mind" is one that is capable of doing all the things a human mind can do. A human mind is capable of emotion, confusion, and going crazy as well as logic and intuitive reasoning. There is a structure of neural networks that we are born with. We as of yet have little control over the "wetware".

    But the time is coming for when the brain and artificial neutral networks will combine.Think of it as a smartphone in your head at first. We will have a classic microchip in our head and we will be able to program it as such. Anytime you try to add it's automatically rerouted to the chip which can do it much faster. Similarly we will be able to create and manage massive databases within our minds. It will be much faster than wetware.Then we get into artificial neural chips which we will similarly be able to program. They may start out as very tiny chips but eventually they will become three dimensional. It could at some point reach the size of a golf ball and it will be inside your brain. After that all mental growth will be through external computers we network with.

    The singularity is the point when after you are implanted you don't need to be taught how to use it or how to do anything for that matter. There still will be natural aptitude and the physical knowledge like playing a piano or a guitar but one will be able to pick it up exponentially faster.

    There will be no war with "AI" per say. It will be enhanced "evolving" humans verses natural humans (luddies). There is no skynet.

    1. Re:The singularity is coming. by Anonymous Coward · · Score: 0

      Luddies?

    2. Re:The singularity is coming. by q_e_t · · Score: 1

      There is no skynet.

      Don't tell the British army.

    3. Re:The singularity is coming. by Anonymous Coward · · Score: 0

      "We will have a classic microchip in our head and we will be able to program it as such."

      Oh hell no, do not want! What company would you trust to make a chip for your brain that didn't have a back door, or at least have exploitable bugs? Intel? AMD?

    4. Re:The singularity is coming. by eaglesrule · · Score: 1

      If we're going to go that far to create wetware interfaces, then general AI would also be a natural byproduct of understanding the processes of how the mind and human biology works.

      We wouldn't even necessarily have to understand how general AI works exactly or why, only the processes used to create it. At first it won't be a threat, but eventually it would be weaponized or handled carelessly. The potential is just that great, and our capacity for hubris also is unlimited.

      I think you're correct that our competitive nature will lead to cyborgs in order to obtain an advantage over the competition. This may be exactly the weakness that an AI would need to exploit in order to achieve dominance.

    5. Re:The singularity is coming. by dinfinity · · Score: 1

      There will be no war with "AI" per say. It will be enhanced "evolving" humans verses natural humans (luddies). There is no skynet.

      You say "Many smart people are shouting the dangers of AI at the moment when we couldn't be further from it." yet imply that full fledged high bandwidth BCIs are somehow around the corner. They're not. Current BCIs are a joke compared to what current AI can do.

      I do think we agree where sentience will end up, but I also believe you overestimate the role organics in general and humans in particular will have in participating in the transition. Let me put it like this: tacking chips onto a human brain is like putting wheels on a horse. It could work, but not as well as just ditching the animal part altogether.

  3. Once more around the circle. by HeckRuler · · Score: 1

    Grab your partner dosie-do!

    What's intelligence? Here we GO!

  4. Grammar nazi here, sorry... by BankRobberMBA · · Score: 2

    It's 'per se', not 'per say'.

  5. Notice! by EETech1 · · Score: 1

    Take note that th they are not referred to as viri!

  6. I would do some serious spot checking by Anonymous Coward · · Score: 0

    because there might be a lot of false positive in that 6,000.

  7. They will never find that elusive virus C teapot by 140Mandak262Jamuna · · Score: 1
    The existence of such a virus was first hypothesized by the scientist Bertrand Russell.

    This is a very special virus that infects its host, H sapiens. It affects their brain and make them identify the virus strains that compete with it (genus Celestial species teapot) and go on an all out effort to get rid of the competitor. The C teapot virus will occupy the niche vacated by these aggressive treatment, but the brains of the H sapiens it affects will never see it or identify it.

    --
    sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
  8. Re:They will never find that elusive virus C teapo by Anonymous Coward · · Score: 0

    Sounds like an AI would be able to find it even if we couldn't.