The other day there was a wasp going in and out of my room on the second floor by the window. I closed the window because it was annoying me.
The wasp found an entry through another window, in another room open in the second floor. It was from the room next door. So, I just closed this other window.
Again, it found an entry from another window. But this time, it started to impress me: it was from a room across the corridor! The wasp flew to the back of the house, found an open window, entered it and mapped to path through the corridor to my room.
I then closed all the windows in the second floor of the house and bet it was over. I was wrong: the wasp came in from the front door, flow upstairs and found the room. It was even capable of doing the path in reverse so it could come in and out of my room.
I closed all the doors and windows on the first floor, but forget to close the small bathroom window. This was enough for the small wasp map an entire new path to my room again almost immediately!
Once it entered my room again, I opened the window again. It was smart enough to use it like it was simply choosing the shortest path.
So, even the small brain of a wasp is able to run "computer vision" algorithms and are fast running "machine learning" algorithms. They are even able to run A* perfectly and merge input from various sensors to perform superb navigation using very little energy, producing almost not heat in a very lightweight package in real time with very low latency!
I'm a beekeeper, and one of the most profounding things to show people when they (anctously) look over my shoulder is this "bee waggle".
It is clear and easy to see, (easier than that d* b* of a queen when you need her, most often), and it tells a lot about the cognitive abilities of a social group, the hive. How they can tell eachother complex directions by twerking.
Kids love it too. And will often play around, wiggling their behinds (a cousin called it bee-twerking) and giggling how they are showing eachother where the sweets are by dancing.
And as per usual, there’s a futurama episode covering the bee dance: where bender, with his fantastic dancing skills goes out to communicate with space bees.
So, even the small brain of a wasp is able to run "computer vision" algorithms and are fast running "machine learning" algorithms.
Now i'm actually curious about its efficiency of the computation. If we could somehow map neurons to transistors (e.g. the logical calculations/algorithms, not the structure and not simulate neurons with transistors), which would be more efficient, the brain or the hardware?
Comments
The other day there was a wasp going in and out of my room on the second floor by the window. I closed the window because it was annoying me.
The wasp found an entry through another window, in another room open in the second floor. It was from the room next door. So, I just closed this other window.
Again, it found an entry from another window. But this time, it started to impress me: it was from a room across the corridor! The wasp flew to the back of the house, found an open window, entered it and mapped to path through the corridor to my room.
I then closed all the windows in the second floor of the house and bet it was over. I was wrong: the wasp came in from the front door, flow upstairs and found the room. It was even capable of doing the path in reverse so it could come in and out of my room.
I closed all the doors and windows on the first floor, but forget to close the small bathroom window. This was enough for the small wasp map an entire new path to my room again almost immediately!
Once it entered my room again, I opened the window again. It was smart enough to use it like it was simply choosing the shortest path.
So, even the small brain of a wasp is able to run "computer vision" algorithms and are fast running "machine learning" algorithms. They are even able to run A* perfectly and merge input from various sensors to perform superb navigation using very little energy, producing almost not heat in a very lightweight package in real time with very low latency!
I'm glad this critter is small. Nature is scary.
The wasp might be using smell to figure out the way into your house; any odor would come out of any open window.
It might be using A* then :P
Exactly! The intensity of the smell is a perfectly fine consistent heuristic.
Was it a paper wasp? They are so cute and friendly.
This little one didn't mind a bit when I put my macro lens an inch or two away, and even did some tricks to show off!
https://geary.smugmug.com/Macro/i-VsCPgMS/A
https://geary.smugmug.com/Macro/i-3xN7pMk/A
Can't tell, but looks similar.
What if I told you it wasn't the same wasp?
One wasp trasmitted a map of your house to another and another one!
Your comment may sound like a joke but bees are known to transmit direction and distance to one another and are also able to solve complex problems:
I'm a beekeeper, and one of the most profounding things to show people when they (anctously) look over my shoulder is this "bee waggle".
It is clear and easy to see, (easier than that d* b* of a queen when you need her, most often), and it tells a lot about the cognitive abilities of a social group, the hive. How they can tell eachother complex directions by twerking.
Kids love it too. And will often play around, wiggling their behinds (a cousin called it bee-twerking) and giggling how they are showing eachother where the sweets are by dancing.
And as per usual, there’s a futurama episode covering the bee dance: where bender, with his fantastic dancing skills goes out to communicate with space bees.
Yep, it was intended as a joke, halfway between a citation of The Matrix and the known joke about the Scottish black sheep: https://www-users.cs.york.ac.uk/susan/joke/3.htm
Of course bees can communicate between them, and probably also wasps.
Maybe it spoke with the same accent
It had a red mohawk.
genuinely thought the last line in this was going to be and then I found a wasps nest in the room next door!
Now i'm actually curious about its efficiency of the computation. If we could somehow map neurons to transistors (e.g. the logical calculations/algorithms, not the structure and not simulate neurons with transistors), which would be more efficient, the brain or the hardware?
You may have already, but you'd enjoy reading 'Conscious Mind Resonant Brain' by Stephen Grossberg.
Thanks for the hint/tip/advice. Never heard about it. Hope you found my tone similar or with some close ideas.
You might want to consider the “screen” algorithm :-)
Could there be a wasp nest outside your window or worse inside your room?