Here’s what I made this week:
It's been 35 years since Seymour Krelborn was on the cover of Life Magazine. In a matter of weeks, the world was infested with Audrey Twos, and with it, massive deaths. It's been only seven years since humanity collectively started avoiding the plants rather than trying to fight them. In that time, the amount of living pods has plummeted, with some pretending to be dead and eating the less cautious. Today it was announced that the population of Audrey Twos has reached "extinction" level, with only a handful living in controlled environments.
For decades, "Krelborn" has carried with it the stigma of the near destruction of our species. Cities decimated, millions dead. In that time, however, the amount of carbon absorbed by the leafy monsters stopped the precipitous pace of global warning. Urban reductions caused massive amounts of CO2 levels to fall without the massive amounts of gas-powered cars on the road. Coastal cities no longer fear flooding or being eaten alive by a plant.
In a way, Krelborn might have saved the planet.
Didn’t really do much during the week due to the last few Mardi Gras parades. I worked on the neural network lottery project a little more. I’m leveraging a Java project called Neuroph that can make & train networks out of the box. I started fleshing out the training of the network with the data set I got from last week.
Follow my progress here.
Agh, more Carnival. I started a github project for building a neural network that can try to predict Powerball numbers. Mainly because 1) I want to learn some neural network frameworks (Java is easier for me), and 2) I want to test how a neural network will perform when trying to predict what should be random numbers. It’s no better than the random numbers that quick picks make.
I scraped the list of winning numbers here, but only from October 7th, 2015 (last time they changed the number ranges). From there, I made a file that has one week’s numbers along with the next week’s numbers for training the network, then testing to verify its accuracy (or lack thereof).
Anyway, click here to see the project in github. I’ll make a post once it’s actually working and I can feed it a set of numbers, and it’ll try to predict the next batch.
It’s Carnival time, I made a beat
* Finished the nap tracker that I started last week. Check it out here: https://output.jsbin.com/gasefiw
* Started planning out & researching a project idea to make an artificial “spidey sense”. Tired of stubbing my toes in the dark! It’s a pretty big project (should take about 16-20 weeks), so if it’s going to take that long, I might as well make it a video series!
The plan that I have so far is have a bunch of distance sensors around either a hat or a belt, run those into a Raspberry Pi, then output to some sensors on a shirt or vest to give me haptic feedback. If it works, it’ll be awesome! If it doesn’t work, then what a waste of time!
So sleep training is over (yay!) but it didn’t come without difficulty. One of the problems we encountered is trying to figure out when our kid’s naps will be. So I decided to throw together a simple JS/HTML page that can handle this. Made the layout here.
Some minor display bugs, but it works from the start to spit out ideal times for whatever you want for your baby. Once it’s completed, I’ll host it on here.