Signal Processing
FABLES FOR OUR GRANDCHILDREN
READING TIME: 3 minute
Do you know that Artificial Intelligence (AI) has to practice thinking before it can analyze stuff. As you probably know, AI takes large sets of data and processes them through their AI neural networks to find patterns in data. To practice, the AI is given “similar” data to what they will analyze so it can teach itself how to recognize the patterns in the data. Once it has created the sophisticated algorithms it needs, then it is fed the real world data.
But guess what: AI makes too many mistakes!
So scientists found they have to contaminate the practice data to prevent mistakes. They add some chaos and some signal noise to screw with the AI processing. And it works! AI is way smarter and does a much better job finding patterns when it’s messed up.
All this junk in the data is to prevent what they call “overfitting.” What this means is that AI has a model of the data in its virtual head and if there aren’t crazy signals now and then, it over fits the data you feed it to its model. That’s bad because it misses patterns. Chaos makes the AI more vigilant, flexible, and open to new data.
It’s kind of like driving your car for a long time. Pretty soon you go into a zone in your head and you ignore processing some of the data. Have you ever arrived somewhere and don’t really recall the details of the drive? That’s because our brains are like AI. The familiarity of the same things will cause us to “overfit” data to our mental model of the world. Overfitting for humans means we miss opportunities and threats. We get too comfortable.
Scientists are now proposing a new theory they will be testing in the lab as to how the human brain protects us from overfitting data to models. They theorize that we use a little thing called dreaming. The bizarre wackiness of dreams keeps your brain from getting too locked into your models. Here are two of my dreams from this week.
1. I had a dream where I was upset with Keith for calling Mollie at 4am to buy a large box of marijuana from her. The box was about the size of a microwave. The part I was upset about was that he called her so early in the morning and I knew that would upset her. It was unlike him to be insensitive to her sleep and I wondered if this escapade (buying drugs without discussing with me and waking Moliie) was a sign of something. Then I woke up. My first question: Why Mollie? She is so unlikely to have access to this quantity of pot.
2. I also had a dream that Maggie walked in my back door and started roaming around my house as if in a trance. She was holding some sort of handlebar thing in front of her with blinking lights, wiring, glowing coils, puffs of smoke. I kept trying to talk to her, but she just kept walking around all the rooms leaving this faint mist as if she was degaussing our house. Never made eye contact. Never said a word. Then she disappeared and I awoke startled and confused.
My first question: Does this have anything to do with crypto?
Scientists think our brain feeds us contaminated data on purpose in our dreams to ward off over-fitting. Keeps us on our toes so we expect the unexpected. So thank your head and realize what a miracle you are.
And, while you’re at it, also rethink the value of video games. I surmise they also ward off overfitting and are producing a generation of people who will be neurologically superior to past generations in their attentiveness and signal processing.
I hate you,
Granny
P.S. .
Nah! Of course I love you. I’m just messing with your input data so you are smarter. (Hee Hee)