[HN Gopher] To be energy-efficient, brains predict their percept...
___________________________________________________________________
To be energy-efficient, brains predict their perceptions
Author : sebg
Score : 235 points
Date : 2021-11-17 15:39 UTC (7 hours ago)
(HTM) web link (www.quantamagazine.org)
(TXT) w3m dump (www.quantamagazine.org)
| strict9 wrote:
| > _Through predictive processing, the brain uses its prior
| knowledge of the world to make inferences or generate hypotheses
| about the causes of incoming sensory information._
|
| I wonder if this mechanism is responsible for many shared false
| memories (or Mandela Effect). As an example, the spelling of the
| Berenstain Bears is misremembered by many, perhaps because we're
| so accustomed to seeing the suffix -stein in last names.
| EMM_386 wrote:
| > Berenstain Bears
|
| Now that's a weird one, never saw that example before. I'm
| familiar with the subject ... yet, if you asked me to spell it,
| it would have been -stein.
|
| Very interesting.
| mcny wrote:
| I still refuse to believe ET by Katy Perry always said
| "different DNA". I distinctly remember it used to say perfect
| DNA, not different DNA. Or I've lost my mind.
| rectang wrote:
| For a couple of years I made my living as an audio mastering
| engineer, which involves making tiny changes at or below the
| threshold of perception. One of the main ways I set myself up
| for success was to accept the human perceptual mechanism's
| propensity for illusion, and then to build systems which helped
| me avoid being deceived by it.
|
| * An instant level-matched AB switching mechanism for
| auditioning changes
|
| * Preferring tools which could be checked under optimal
| perceptual conditions, and choosing to avoid gear which did not
| lend itself easily to such conditions. This generally meant
| preferring software plugins and avoiding outboard gear unless
| the budget was very high (which it rarely was for my clients).
|
| * Isolating changes and amplifying the effect to well above
| perceptual threshold in order to get an impression of the
| change before dialing it back down. (This is a standard
| practice, especially for mix engineers.)
|
| Many audio engineers don't hold the same priorities, especially
| "audiophile" types those who advertise their "golden ears"[1].
| And many customers don't want someone who acknowledges the
| fallibility of their perceptions. But I was confident that I
| was absolutely doing my best work and offering the best
| possible value to my clients.
|
| [1] It's basically impossible to talk intelligently about audio
| in an open internet forum (including HN) because there will
| always be a swarm of participants who don't accept the limits
| of perception and their own propensity for illusion.
| skulk wrote:
| > gear which did not lend itself easily to such conditions
|
| could you expand on this a bit? I don't see how software vs
| hardware could impact this, doesn't it all just go through an
| amplifier to your headphones at the end?
| ubercore wrote:
| Only think I can think of is taking the effect of cables
| and preamps out of the loops when running through outboard
| gear?
| rectang wrote:
| I built my AB system in software, which allowed me to make
| a tiny crossfade on switching and avoid a big click. It's
| well understood that such clicks wreak havoc on your
| ability to perceive small changes. I didn't have access to
| an analog version of that at the time.
|
| It was also more difficult to set up and compare multiple
| "wet" settings for hardware devices, and it was less
| convenient to "save" settings, take a break and come back
| with fresh ears (because the studio might be used for
| something else in the interim.)
|
| It's possible to give yourself optimal conditions with
| hardware tools, it's just less convenient (and thus more
| expensive). I believed that giving myself optimal
| perceptual conditions and an ideal workflow was far more
| important than gear choice -- especially in terms of value,
| but in my view even in terms of doing my best work in an
| absolute sense.
| sneak wrote:
| It's always astounding to me the number of "engineers" who
| make fantastic claims and have never done an abx. It's nice
| to know I'm not alone.
|
| Do you give mentorship/lessons?
| rectang wrote:
| In my view, understanding the human perceptual mechanism
| was as essential as understanding how to set up good
| monitoring to doing my best audio engineering.
|
| I left the audio industry a long time ago. I love music,
| but I don't love audio engineering enough to sacrifice my
| life to it -- and because there's a tremendous oversupply
| of labor for art-adjacent professions, that's what it
| takes.
| Geee wrote:
| Not just the human brain, but anything that is _intelligent_.
| Life itself is a prediction machine. Evolution selects for
| prediction power in an energy-constrained environment.
| eevilspock wrote:
| > Life itself is a prediction machine.
|
| I beg to differ. That assumes evolution is teleological, that
| it isn't _reactive_. It 's like saying positive and negative
| feedback loops are predictive. You're diluting the meaning of
| "predictive" to say something that sounds deep.
| hypertexthero wrote:
| "The observer is the observed."
|
| --Jiddu Krishnamurti
| podgaj wrote:
| I have Aspergers and suffer from something called closed eye
| visions. When I close my eyes, mostly at night, I get flashes of
| images, like faces and people and places. The idea is that the
| brain is trying to make sense of the random light patterns I am
| seeing even when my eyes are close. And so it's hyper predictive.
| Like others are saying, this is nothing new to me.
| somebodythere wrote:
| I get these too, although I wouldn't describe it as suffering.
| Sometimes the images are detailed and tangible enough that it
| feels like looking at a printed photo through my eyelids. When
| the images are particularly interesting, I like to draw them.
| podgaj wrote:
| Sometimes mine are horrific though. Like a face in bandages
| with an eye hanging out. But sometimes they are interesting
| and nice, like tree tops of a forest.
| guerrilla wrote:
| Same here. I used to narrate them and make up stories about
| them to my girlfriend :)
|
| There's also a different TED talk than the one I already linked
| about your theory which gives a lot of interesting examples[1].
|
| 1. https://youtu.be/SgOTaXhbqPQ
| vga805 wrote:
| Andy Clark's Surfing Uncertainty[1] is a nice monograph of the
| topic
|
| 1:
| https://oxford.universitypressscholarship.com/view/10.1093/a...
| ithkuil wrote:
| Funny coincidence, I'm in the middle of reading of "a thousand
| brains" by Jeff Hawkins
| eevilspock wrote:
| Do you notice that your visual center of focus shifts to
| different parts of the duck vs rabbit image?
|
| While I strongly agree with the ideas of the article, the fact
| that you have to focus on different parts of the image to shift
| perception, altering the "bottom up" sensory data your brain
| receives, undermines at least a little the use of this example.
| paradaux wrote:
| This would certainly explain why my brain periodically suffers
| from meltdown.
| nefitty wrote:
| One of my "executive function" problems is that I get
| overwhelmed when I think about things I need to do. I noticed
| that a sort of movie starts playing in my head of how to
| accomplish the task, i.e. the steps entailed, resources needed,
| etc.
|
| What happens frequently is that my brain takes multiple
| discrete tasks and attempts to simulate each one's steps to
| completion simultaneously. I suddenly find myself at the front
| door paralyzed for five minutes, "Should I grab the trash since
| I'm going to go check the mail? I have to walk the dog, I can't
| carry all this trash while holding the dog. Where will I put
| the dog when I get the mail. Ok, put the leash on the dog, grab
| the trash... Crap, there's so much trash! Ok, just take the
| office trash out..."
|
| ADHD medicine stopped working after a month. What has helped a
| lot is N-acetylcysteine. In fact, it's been several months and
| I'm comfortable saying that it has changed my life. The
| trainwreck thought loops give way to singular chains of focus.
| phkahler wrote:
| Tell me more about acetylcisteine. How much per day? Is there
| a theory behind that use? Etc...
| nefitty wrote:
| It has to do with the glutamine-glutamate system,
| glutathione, GABA and dopamine. There are various possible
| etiologies for ADHD, but I'm lucky this is one path I
| discovered that actually helped me. It also explains why
| Adderall and Wellbutrin didn't help me, as they affect
| dopamine and norepinephrine respectively.
|
| I take at least 2g per day. Some NAC supplements have some
| selenium and molybdenum included, so in those cases it's
| important to be mindful of not taking too much of those
| trace minerals.
|
| Check out the introduction of this paper: https://www.scien
| cedirect.com/science/article/pii/S014976341...
| digilypse wrote:
| How long did it take for you to observe a difference
| after you started taking NAC?
| nefitty wrote:
| I noticed a difference on day one. I'll note that I also
| take l-glutamine to further the effect on the system I'm
| targeting. You'd also want a molybdenum and selenium
| source because NAC uses them up.
|
| The effect is not dramatic like a stimulant. It's more
| like it makes my attention stickier, and allows me to
| hook into long problems more easily. It doesn't "feel"
| like anything, just causes a noticeable change in what I
| find engaging. One weird effect it has had is I
| completely stopped craving alcohol, news and social
| media. I usually spend hours on Reddit when I don't have
| anything on my plate. I'm talking 40k+ karma, multiple
| posts/comments per day type of addiction. I see multi-
| day, multi-week gaps between comments on my account now,
| which is unheard of for me. I know it's a silly
| heuristic, but it's an example of the behavior change.
|
| I've seen much smarter people than me on Reddit say that
| effects may take 2-3 days. The supplement itself is
| pretty cheap, and negative side effects are minimal to
| non-existent. I am seriously astonished that such a
| common and accessible supplement has had the impact it
| has had on me. I personally think the cost/benefit ratio
| is so good that warranted skepticism can be overcome with
| a self-experiment.
|
| If it doesn't have an effect, then at least a person can
| check off the "glutamate-glutamine system" checkbox in
| their quest to address executive function problems.
|
| https://en.wikipedia.org/wiki/Glutamate%E2%80%93glutamine
| _cy...
| hawski wrote:
| I'm sorry if that's a dumb question, but is monosodium
| glutamate consumption related to ADHD?
| nefitty wrote:
| No, that's a smart question because MSG is so prevalent
| in our food. You'd have to ingest a lot of MSG to make an
| impact on glutamate levels, which would then affect GABA,
| which then affects ADHD symptoms.
|
| https://www.karger.com/Article/Fulltext/494782
| afarviral wrote:
| Your comment on N-acetylcysteine is really left-field. Its
| not even listed as a use of that medication, whos primary
| purpose is to treat panadol overdose. Can you elaborate?
| nefitty wrote:
| I hesitated to mention the specific supplement for that
| reason, but maybe it will help someone.
|
| Apparently, the glutamate-glutamine system is implicated in
| psychiatric problems. Through my research I found that GABA
| deficiencies can cause ADHD symptoms. GABA is produced in
| the gut, I have stomach problems, so this seemed feasible.
| I learned that NAC increases glutathione and helps regulate
| glutamate. The effects people report point to an effect on
| this system.
|
| The Wikipedia page for NAC has many references, but this is
| one review of its possible uses in psychiatric problems: ht
| tps://www.sciencedirect.com/science/article/pii/S014976341.
| ..
|
| https://en.wikipedia.org/wiki/Acetylcysteine
| astrange wrote:
| N-acetylcysteine is very effective (for some people)
| against anxiety, addictions, amphetamine tolerance, acne,
| and also some things that don't begin with "a".
|
| Amazon also stopped selling it this year because people
| thought it was a COVID cure.
| nefitty wrote:
| Yeah. I've had several convos with my health store owner
| about this. She told me she's found herself stocking up a
| lot lately because of the increased popularity and COVID
| misinformation-motivated stocking changes on Amazon.
| jackallis wrote:
| the saying "My brain runs faster than my mouth" lends to this
| idea.
| swayvil wrote:
| A kind of reverse-science if you will. Don't think about what you
| see, see about what you think.
|
| And what we think derives from so many different sources.
|
| It explains a lot.
| guerrilla wrote:
| This TED talk has a great example where the same sound sounds
| different to you based on what you expect to hear, if you need to
| prove this to yourself [1]. His thesis is also that the brain is
| a prediction machine.
|
| 1. https://youtu.be/lyu7v7nWzfo?t=365
| tasty_freeze wrote:
| Michael Shermer played Stairway to Heaven backwards and it
| mostly sounds like gibberish. But when he displays the
| 'translation' as text and plays it again, the words seem pretty
| clear.
|
| https://www.youtube.com/watch?v=wYG2oZXSvdE
| jodrellblank wrote:
| In 1985 two young men attempted suicide, and the family sued
| the British rock band Judas Priest, accusing them of hiding
| subliminal messages in the music that said "Do It!" when
| listened to backwards.
|
| From memory, part of the trial was the defence taking a song,
| suggesting what words the court would hear in it played
| backwards, and then playing it backwards. e.g. this clip
| https://youtu.be/uyIsu93zAoM?t=1123 and showing that you hear
| the suggested words out of the noise.
|
| The trial was thrown out, after deciding there were words and
| they were subliminal, but were not deliberate. The whole
| situation included: "In a pre-trial motion, the judge ruled
| that subliminal messages were incapable of being protected
| speech under the First Amendment to the United States
| Constitution, since they were by definition not noticeable
| and thus could not form part of a dialogue."
|
| Documentary - https://www.imdb.com/title/tt0104140/
| eevilspock wrote:
| did Grover drop an f-bomb?
| https://www.youtube.com/watch?v=1mU7t9GZ9Os
| _Microft wrote:
| Here is a sound that sounds like either _brainstorm_ or _green
| needle_ , depending on which words you focus. Try own word
| combinations. I have no trouble hearing _brain needle_ or
| _green storm_ , for example.
|
| https://www.youtube.com/watch?v=1okD66RmktA
| shard wrote:
| Hmm no matter how hard I try, I can't get _storm_. Closest I
| can get is _brain needle_. It 's just too clear to me that
| there are three syllables, and I can't map _storm_ to the
| second half.
| AnIdiotOnTheNet wrote:
| I had trouble hearing it too, so I closed my eyes to remove
| the flashing light as input and, I shit you not, just
| straight up told myself "you will hear _brain storm_ ". It
| was interesting not only in that that worked, but that my
| perception of the word was so completely different that it
| sounded nothing like _green needle_.
| kibwen wrote:
| And no matter how hard I try, I can't get "needle". There's
| an "s" sound in there, how does anyone get "needle" out of
| that? :P
| defaultname wrote:
| Related but also not: One of the most interesting aspects about a
| "shroom" trip is the incredible ability to visualize. Visualizing
| is something that we all think we can do normally, but if you
| really try to picture something in your head -- specific faces,
| the design of a bicycle, the layout of a room -- it's quickly
| evident that we are actually _terrible_ at it (it 's hard to
| reverse information from that neural network). Even if you're
| drawing it's often an iterative process where each line drives
| the next.
|
| Maybe other people are better at it, but when self-interrogating
| and inspecting one's own vision, it just completely falls apart.
| https://www.booooooom.com/2016/05/09/bicycles-built-based-on...
|
| Normally.
|
| I can only speak to personal experience, but under the influence
| of psilocybin I find that closing my eyes and visualizing complex
| environments and machines -- of literally inspecting and walking
| through gearing systems, for instance, and rationalizing their
| operation -- with complex lighting, etc, leaves me just in awe at
| the mind's capacity. It literally feels like looking into another
| universe, the construction of the reality simply too complex to
| be imagined and visualized.
|
| I've actually felt my head to ensure I'm not overheating,
| contemplating the process like the mind was a GPU.
| dgan wrote:
| This happens to me (almost) every night, before i fall asleep,
| and for some short time right after I woke up
|
| Amazing state, i love it!
| chwzr wrote:
| I like the way you are thinking here. Speaking of the brain as
| a neural network, what might be the counterpart of psilocybin
| in our cs neural networks?
| rolisz wrote:
| Randomized weight updates?
| jareklupinski wrote:
| along with amplifying feedback bleeding into unrelated
| parameters
| frisco wrote:
| > Visualizing is something that we all think we can do
| normally... we are actually terrible at it
|
| > leaves me just in awe at the mind's capacity
|
| Want a really trippy realization? All you _ever_ see is brain
| activity! Sounds obvious, but most people haven 't really
| internalized it. That's all regular perception, which feels
| totally real and solid, is. The psychedelic just gave you a
| greater ability to volitionally influence the percepts.
| penjelly wrote:
| yeah, those color illusions really demonstrate this. like
| when you stare at a point and the picture changes, the
| picture looks fully colored but in reality its turned black
| and white and your brain just hasnt realized it.
| zero_iq wrote:
| The ideas in this article are strongly reminiscent of those in
| Jeff Hawkins' "On Intelligence" (2004).
|
| The idea of the brain operating (at least in part) as a
| "prediction machine" is certainly not a new one. I'm actually
| surprised it's taken this long for this sort of experimental
| confirmation and theory to become more mainstream.
| cr4zy wrote:
| Prediction machines, yet critically, taking actions that pursue
| states which we are not quite able to predict.
| https://reader.elsevier.com/reader/sd/pii/S0896627315007679?...
| SubiculumCode wrote:
| On any serious discussion of neuroscience, I'd advise to keep
| Jeff Hawkins out of it. Not only are the ideas he publishes
| often gross simplifications with little data to back it up (but
| does wonders as marketing), they were ideas pushed and
| developed by real working neuroscientists. Just my opinion.
| LesZedCB wrote:
| isn't that pretty face value what Numenta's goal is though?
| They read research papers on neuroscience and try to distill
| it down to applicable engineering problems for their HTM and
| see what works.
|
| i can't speak much as to their original research efforts, but
| i personally appreciate the engineering-research side of what
| they are doing.
| Invictus0 wrote:
| Thought of this as well. Have you read his newer book? I
| haven't decided if I should pick it up or not.
| chalcolithic wrote:
| Oh, he has a newer book! Thank you!
| KingFelix wrote:
| It's alright, I didn't end up finishing it. I really like On
| Intelligence, wasn't pulled in to finish. Might have been
| state of mind at the time etc
| metanonsense wrote:
| I have read the book and I really liked the first half, which
| explains the "thousand brains theory of intelligence". Very
| inspiring and thought-provoking (at least to me as an
| interested amateur in this field). The second half, however,
| would better have been a book of its own. It's about Hawkins'
| ideas on AGI implications and whatnot, which is quite
| entertaining but devalues the first half, in a way.
| ethbr0 wrote:
| It's been known in many communities, I think.
|
| There's been a number of articles on HN over the years on
| saccades / visual interpolation.
|
| See: https://www.portsmouthctc.org.uk/a-fighter-pilots-guide-
| to-s...
| sea_things wrote:
| This article is so fascinating... and it reminds me of one of my
| favorite topics! I'm excited to share this link:
|
| https://en.wikipedia.org/wiki/Ideasthesia#Neurophysiology_of...
|
| > Ideasthesia is congruent with the theory of brain functioning
| known as practopoiesis. According to that theory, concepts are
| not an emergent property of highly developed, specialized
| neuronal networks in the brain, as is usually assumed; rather,
| concepts are proposed to be fundamental to the very adaptive
| principles by which living systems and the brain operate.
| exporectomy wrote:
| A lot of comments here seem to be overlooking the point of the
| article. That brains predict their sensory inputs isn't new. What
| seems to be is that a NN where correct predictions are done with
| small values of the weights and error correction has large
| weights uses lower weight values overall. That being energy
| efficient seems to be more a coincidence of how biological brains
| happen to work - larger weights use more energy. At least that's
| how I understand it. Doesn't seem quite as incredible as
| predicting sensory inputs.
|
| Obviously being able to predict the future has value beyond
| conserving energy.
| Barrin92 wrote:
| I think this is a very intuitive theory. I feel you can actually
| notice this consciously when you do some mundane repetitive task
| you've done a lot and something is off. For example if you lift a
| cup you thought was full and it isn't and for a moment it
| _really_ throws you off. It 's not just perceived as a difference
| but as something 'out of whack'.
| euroderf wrote:
| Or when you lift an egg carton with one hand but it's only
| partially full and all the eggs are at one end and you lose
| your grip oops. I always rearrange our eggs to have left/right
| symmetry in the carton and my wife must think I'm nuts.
| birriel wrote:
| Friston on the Free Energy Principle and Markov Blankets:
|
| https://youtu.be/NIu_dJGyIQI
| supperburg wrote:
| The brain is a prediction machine largely. How is it possible
| that you can experience a lucid dream where every sensation and
| interaction is as real as real life? It's because your
| consciousness lives inside a simulation. When you're awake, your
| brain uses sensory input to populate the simulation. But when
| you're asleep it populates it itself. When you're awake, your
| brain doesn't scan every square inch of the real world looking
| for things to populate the simulation; it scans sparsely looking
| for key indicators and then guesses the rest. It's an
| optimization for time and energy. Most of the things you
| experience are guesses. And when you're asleep this guessing is
| all that you're seeing. Dreams are just the guessing machine.
|
| There are scattered reports of lucid dreamers who take
| benzodiazepines. They can't escape their dreams. They simply wake
| up into the dream again if they die. Everyone reported it as
| being terrifying. Some reported that their reality checks stopped
| working; double taking at a watch didn't change the position of
| the hands. The world became frighteningly real. It's likely that
| the guessing machine you experience during a normal dream is not
| operating at full capacity.
|
| It explains everything. How people swear on their life that they
| saw something, ordinary or extraordinary, that couldn't be
| explained by a trick of light on the eyes or anything else. It
| explains dreams of course and other realistic hallucinations had
| by schizophrenics or drug users. The simulation doesn't just deal
| with the physical world, it deals with abstract things like the
| "presence" of a person or entity. It's all haywire simulation.
|
| This idea that synapses are a part of an energy optimization is
| fascinating. It is widely speculated that inflammation has to do
| with schizophrenia, and inflammation is also involved in
| metabolic interference. African sleeping sickness for example is
| a purely inflammatory disease that causes people to run out of
| energy and fall asleep. Perhaps schizophrenia is some kind of
| metabolic knob being turned down by an inflammation signal which
| then results in the progressive shutting down of synapses as a
| way to conserve energy...
| selimthegrim wrote:
| You wonder if that Batman episode about people not being able
| to read in their dreams holds up
| AnIdiotOnTheNet wrote:
| > There are scattered reports of lucid dreamers who take
| benzodiazepines. They can't escape their dreams. They simply
| wake up into the dream again if they die. Everyone reported it
| as being terrifying.
|
| I am not a lucid reamer nor to I take benzodiazepines, but I do
| occasionally have dreams that I suddenly realize are dreams and
| then wake up into another dream. Occasionally, this too results
| in my realizing I'm dreaming and waking into another dream,
| where the pattern repeats at faster and faster pace until I'm
| just examining the way my bedroom appears in each iteration
| looking for how it is wrong and trying desperately to wake up
| for real this time.
|
| I can confirm that it is pretty terrifying. Fortunately for me,
| I've developed an appreciation of nightmares and actually don't
| mind it all.
| amelius wrote:
| Makes sense. If someone throws a ball into the air and you want
| to catch it, you predict where it will go. Along the way, if
| there is an error between prediction and actual position, you
| adjust for it.
|
| Classic control theory :)
| FearlessNebula wrote:
| Speculative execution
| motohagiography wrote:
| Rich, rich analogy. SPECTRE and Meltdown attacks for the mind
| have to be a thing, possibly as interrogation and cross
| examination techniques?
| forgetfulness wrote:
| Well... from the looks of it, we've gotten diddly squat from
| trying to think of the brain like the thing I'm writing this
| comment on, so probably not.
| 988747 wrote:
| We use SPECTRE and Meltdown attacks against brain every day,
| we call them "jokes".
| motohagiography wrote:
| Woah. That could make me a _very_ dangerous person, because
| I am not funny at all.
| shostack wrote:
| Is this akin to asynchronous space warping or synchronous space
| warping used with VR headsets to improve perceived frame rates?
| CyberDildonics wrote:
| No, that has nothing to do with this at all. This is about
| making predictions and comparing them to new information.
| xyzzy21 wrote:
| And in many cases only the prediction is used - people see what
| they want to see. It's a natural extrapolation of this (which can
| be very wrong and even deadly).
| vez- wrote:
| Considering these three statements:
|
| - the brain predicts what its upcoming input will be,
|
| - quantum biologists ask if the human eye is sensitive to quantum
| effects, and
|
| - measuring quantum information under different bases result in a
| different quantum state of not only the measurer, but of the
| world being measured.
|
| I wonder if it is possible that the brain uses its predictive
| power to change the basis under which the eye measures photons,
| resulting in different perceptions as well as a different
| reality. A bit of a crazy idea but I don't see any reason why it
| shouldn't be the case other than if it is shown that biological
| sensory organs are simply not that precise.
|
| [0]https://www.templeton.org/grant/is-the-human-eye-able-to-
| see...
| leobg wrote:
| Classic in marriage counseling. Many fights occur not because of
| what the other person has said or done, but because what your own
| brain has added to their behavior. Often, you see what you are
| most afraid of. Or what you used to experience in childhood. To
| "see" that your partner here and now is actually completely
| different takes a lot of energy.
| rks404 wrote:
| holy crap this is a very useful insight for me. Thank you!
| SubiculumCode wrote:
| THE BIGGEST issue in marriages, no doubt.
| xeromal wrote:
| You have to clear the cache constantly.
| HenryKissinger wrote:
| Insufficient memory to perform operation. Your brain will go
| to sleep in 5...4...3...2...1...
| trhway wrote:
| It isn't that straightforward for GAN.
| alx__ wrote:
| You can actually just overwrite the neurons. But it requires
| multiple exposures to new perspectives.
___________________________________________________________________
(page generated 2021-11-17 23:00 UTC)