U tweetove putem weba ili aplikacija drugih proizvođača možete dodati podatke o lokaciji, kao što su grad ili točna lokacija. Povijest lokacija tweetova uvijek možete izbrisati. Saznajte više
Did you know you can classify MNIST using gzip? 
You can get 45% accuracy on binarized MNIST using class-wise compression and counting bits 
No @PyTorch or @TensorFlow needed
BASH script and @scikit_learn classifier 
https://github.com/BlackHC/mnist_by_zip …
Sure but...45% accuracy is not exactly good. You can get close to 88% with a linear classifier. You can get 95% with nearest-neighbor/L2 distance. No deep learning necessary. But if you want more than 99% without losing your computational shirt, go with ConvNets.
Thanks! That's trueI would not recommend anyone use this classifier in seriousness
I was surprised it is working this well at all and better than nearest-neighbor on pixel sums.
At best, it's a simple proof-of-concept for information-theoretic approaches
After your comment about the accuracy, I dug in a bit. Turns out the zip compressor only uses the last 64 KB for its "statistics", which is about 82 samples.
When using all training data, zip compression obtains 74% accuracy on MNIST 
Not great, not terriblepic.twitter.com/tU9ZqDYCAs
Nice burn!pic.twitter.com/C26E5THsRQ
How does DJVU do on MNIST? (https://en.wikipedia.org/wiki/DjVu )
And you can get more than that with CapsNet
So we will get a MNIST-overfitted net which will work poorly with some rare real-world hand-writing. We need a more general net and dataset. It could be the grid cells based network. Maybe trained on videos of hand-writing
Sure ConvNets are fine if you have 3 minutes and 45 cents to burn. Is 99% worth that!?
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.
