> A newborn foal can stand within 55 minutes of being born and can walk or run within 90 minutes. That is crazy fast training speed. Is it training initialization or is it a form of transfer learning?
Probably neither. How does the human baby "learn" to breath almost immediately after birth? I suspect the answer here is also the same.
> A newborn foal can stand within 55 minutes of being born and can walk or run within 90 minutes. That is crazy fast training speed. Is it training initialization or is it a form of transfer learning?
Or maybe having 4 legs (as a foal) gives more natural balance than a human or maybe even a cat or dog.
In the same way technically you only need one leg to walk (hop) around but that's much more difficult than having two.
> Are these innate patterns of activity priming mammalian brains in a similar way?
A newborn foal can stand within 55 minutes of being born and can walk or run within 90 minutes. That is crazy fast training speed. Is it training initialization or is it a form of transfer learning?
Most human babies take 9-18 months to walk, and this is commonly attributed to the fact that human babies have immature brains that cannot mature due to birth canal constraints, which is constrained by pelvis size which is constrained by the need to walk upright.
I have wondered about how true this is. Maybe our brains are just so much more powerful, or capable of much deeper understanding, that we require a much lower initial learning rate.
The lottery ticket could explain why some children learn to walk at 6 months, while others are closer to 24 months, with no different in intelligence or motor skills in later life.
Edit: As some people below have correctly pointed out, human babies could not walk within an hour for physiological reasons, but nor do they exhibit the basic motor skills, spatial reasoning or image processing required for walking. Many human babies even struggle to feed for the first 24-48 hours or longer.
Meanwhile baboon babies can hold onto their mother from birth while the mother climbs trees.
Not absolutely nothing, the neural net is initialized with some weights encoding basic things (breathing, sucking, crying, etc.). Newborn horse walks and follows mother after first 5-10 minutes.
A foal can stand shortly after birth. It doesn’t learn that from its environment. Say we train a quadruped AI, then what is our training simulating if not evolutionary development? It’s no different for an AGI and whatever analogue it ultimately has for our evolutionarily derived neocortex.
Is it? If we ignore compute requirements, how long would a good algorithm take to learn something like balancing a quadrupedal robot, based on real-time feedback to its outputs? It is a simpler problem, but drone flight software based on learning, can re-learn how to fly a quadcopter after something like losing a propeller and a sudden shift in weight distribution, in a few seconds.
A robot could be built that learns how to adapt to something like a limb being lost or added or half its weight shifted to the other end. It could probably learn an approximate optimal to move under those conditions in just seconds, as well. Though, I suspect that foal might actually be nearly as adaptable. Adult horses, or humans... not so much. We have many overlapping models of how to move, probably. I would take only seconds to adapt to half my weight being added on my shoulders, in the same ballpark as our best robotic systems. But if my left leg grew four inches it would take me a lot longer than the robot to learn and internalize the best way to move again. I could barely keep up with that when it was just a few inches a year when I was a kid.
The analogy given in the article is interesting. Some organisms perform certain actions even before they start to learn. I myself have seen some animals start running immediately after birth. Less number of parameters (shared parameters) could also be thought of as less complexity and hence less processing power requirements; which implies faster training. Phew! too much similarity.
Well, babies don't learn to stand and walk (and many other skills) from scratch by trial and error. The hard part of constructing a walking machine has already been solved by millions of years of evolution.
Indeed, a calf can walk within hours of being born. A human infant might need more time than a calf in part because walking on two legs is harder than walking on four, but much of the difference simply comes down to the fact that humans are born so early that a lot of predetermined brain development has not yet occurred (the muscles and bones are also too weak). The study [1] found that across varying mammalian species, walking is learned a predictable amount of time after conception (as opposed to time after birth).
There's a surely a continuum between brain functions that are completely hard wired and completely learned from scratch. I think it's accurate to say that for many functions, learning is used as a form of adaptive refinement to finalize specific predetermined neural programs. But the exact interaction between learning and pre-programming isn't well understood in most cases.
I guess it's firmware of some kind. Newborns already know how to breathe, how to control the heart rate and tons of other low level things. Legs movement is a small feature in comparison.
> Anecdotally, when I watch a baby learn to walk, it doesn't really seem like they are trying a zillion things and picking the one that works. It seems much more like their instincts are telling them to move a certain way, and once they grow strong enough, it succeeds and they are walking.
How much of that comes from seeing everyone around them walking?
It seems hard to argue that horses don't have a genetic component, but I don't think Occam's Razor applies as well for humans when it takes 1-2 years of observing the world in order to start walking unsteadily.
The base model is likely to have lots of genetic determined stuff that is there before any training: some animals like horses can walk and perceive/navigate very soon after being born. There's also an innate fear of snake shapes in most mammals.
No, we don't "learn" how to balance, how to identify objects and navigate 3d environments. All of that is highly codified in our DNA from hundreds of millions of years of evolution. Human babies, similar to kangaroos, are just born too early for those systems to have developed.
Animals that are ready to go shortly after birth are called precocial. This includes chickens and most of the equines.
If you've ever seen a newborn foal stand up for the first time, it's clear that's a built-in behavior. Standing up on those long spindly legs usually works the first time. So does walking, and within hours, running.
Lying down, however, is not built-in. I've seen a foal try to get back down on the ground, which is clearly trial and error, often ending in a fall. There's no evolutionary pressure to have that work right the first time.
Of topic a bit but I recently watched some youtube videos about people hearing for the first time, I was surprised how instant things appear to be. Thinking about it you have baby animals that can just walk after a few minutes at tbirth so it makes sense that most of the basic mechanism are pre-programmed before birth.
We are often deceived by the fact that Human infants are optimised for plasticity (I know this is arguable - but it's a reasonable theory) and for their brain to get through a bipeds birth canal (and subsequently grow). Look at lambs in contrast (I've been on a sheep farm in Scotland for a couple of weeks so I've had the opportunity!) Lambs stand up about 3 to 10 minutes after birth (or there is a problem). They walk virtually immediately after that, they find the sheep's udder and take autonomous action to suckle within an hour (normally) and follow their mothers across a field, stream, up a hill over bridges as soon as they can walk. Within a week they are building social relations with other sheep and lambs and within three weeks they are charging round fields playing games that appear pretty complex in terms of different defined places to run up to and back and so on.
This kind of rapid cognitive development argues strongly (IMO) against the kind of experimental/experiential training that a tabula-rasa nn approach would indicate.
Human plasticity and logical reasoning are the apex of other processes and approaches, I think that because we have so much access (personally through introspection and socially via children) to models of theses processes, and the results are so spectacular and intrinsically impressive.
I used to go to the SAB conferences in the 90's, they're still going, but somewhat diminished I think. This was where the "Sussex School" of AI had it's largest expression - Phil Husbands, Maggie Boden and John Maynard Smith all spoke about the bridges between animal cognition and self organising systems. I am pretty sure that they were all barking up the wrong tree (he he he) but there was and is a lot of mileage in the approach.
If you're going to obsess over the nature/nuture thing, you have to consider the counterexample - animals that are born ready to go. These are called precocial species. Most of the large grazing mammals, including all the equines, are precocial.
You can see this by watching the first day of life of a horse.
Within the first hour, the foal stands up by itself. This is a complex coordinated operation for an animal with such long legs. The sequence for doing this is dynamic (not statically stable and can't be done slowly), and clearly built-in, but requires tuning. The foal may fail the first few times, but eventually gets up on the long spindly legs and wobbles. The first few steps are tiny and cautious, but the nervous system calibrates rapidly. Stable walking is achieved quickly.
Trotting appears after a few more hours. Within a day or two, a newborn foal can run with the herd. At that point, all the locomotion functions are working - balance, coordination, visual foot placement, obstacle detection, and collision avoidance. That's a lot of capability. Having worked on both automatic driving and legged robot balance, I know how hard that is.
This demonstrates that mammal brains don't come up blank and learn. There's a lot of hard-wired capability.
(I sometimes comment on mobile robotics that the main thing is getting through the next 15 seconds of life without screwing up. If you can do that, you can then add task-oriented back-seat driving to get something done, and that's the easy part.)
It's like a calf being born. It gets up and starts walking. Pretty amazing. Mesmerises everyone. The model contains everything it needs to know, to walk. But it's not going to dance nor is it capable of working out how to dance.
Babies do something completely different. They can't walk when born. Their model is to blunder about and work things out, building the model up thro a can I do this - can I do that - why not etc. Its only through this doing learning happens.
We have calf ai right now..you ask the calf what do you want to learn next or what are you curious about and you get to see how dumb it is.
Human babies are born knowing how to swim, I suppose they learn that in 3 seconds as well? Both of these are far more likely to just be genetic memory. If you can encode how to fold proteins to make wings, then packaging a control algorithm along is completely trivial in comparison.
Probably neither. How does the human baby "learn" to breath almost immediately after birth? I suspect the answer here is also the same.
reply