Peroglyfer – Machine Learning changes math. Math changes everything.

Search

Skip to content

Home
List It Like 1813+ Ideas
My best ideas
About
Unsorted coolab notebooks
Changed Ideas

Search for:

Unsorted coolab notebooks

Google coolab note books. Machine Learning tests. Unsorted.

http://peroglyfer.se/colab-notebooks-20190503t054545z-001/

Ideas Are Free To Use As Long As You Give Me Credit

By Far The Smartest Idea - Hot Milk With Cooked Rice In A Cup

If free will can be recognized in the machine learning sense then it could probably also be optimized with max as the target. // Per Lindholm

Is a proof some knowledge of the probability of the network.

The Universe Is A Compression Decompression Model() // Per Lindholm 2018

Peace Idea - Any Type Of Probability Can Move Us Forward. Peace Will Do It 100% // Per Lindholm 2018

:wow

Peroglyfer Feed RSS
www.facebook.com/peroglyfer

Search for:

Recent Posts

Cancer Innovation Guess – Cancer is a ?classification problem. I made an earlier unrelated discovery. What can you do with loss = MSE( b + c , b). If b is blod and c is cancer then c should be zero. But if everything must be classfied. // Per
Machine Learning Test – My trick worked better than expected converged L.Convolution2D(10,10,3) x 4 to 99.17% MNIST // Per
Machine Learning – Interesting result. Some sort of classification separator. Adapt x,y so that they can be classified and the same time be part of a circle /// Per Lindholm
Machine Learning Newtons Law Is an ?Existens Law. Keep( 1 / inf ) existing and Keep( 1 / 0 ) existing. For 1/0 -> inf. Here you need to shed a lot of extra material. For this mass generates forces to move both away from each other. So the mass can keep existing in a number sense. // Per Lindholm 2019-jun-12
achine Learning Adapt Math – How to classify if value is an x-cord or y-cord. Add a model_x_zero() resp. model_y_zero() to x and y. They will shape so they can solve the problem and at the same time be close to zero. // Per
Machine Learning Innovation Guess – Extra Functional Classes. Recalculate with some random extra input. Break calculation if at some point. // Per
MAchine Learning Math adapts – Solvning Singular? How? I guess by adapting the solution with singular_relu(x). It outputs 0 when undef else x. // Per
Peace Idea – The simplest thing you can do is to work for peace. This way you don’t have to think so much.
Regreening The Desert With Bitcoin Money For People, Seedlings and Water? What if buying a coffey is as easy as regreening the desert. Simple choice in my opinion why we need crypto currency. Instant climate change mitigation // Per
Machine Learning Speculation – Why wave-forms? Since the problem of fitting a matrix is like a play-springboard. One things goes right and another goes wrong. It could be why wave forms exist. Adjust the springboard for zero total error // Per Lindholm
Machine Learning Thought – Is square matrices inefficient? Inspired by DNA a 2xN matrix is the best? How do you adapt machine learning calculations to this? Going backwards and forwards transformations f(g(x))
Machine Learning – How do you prevent dice throwing in Machine Learning. Data sees no differences I think between random dice and updates that seem looking random. Adjust the update vector so it looks less random. // Per stuck at 97%
Machine Learning – Magnetic field inspired model(). The idea to let mass model layers transform input X to output y under the common loss with the outer field lines. Through another model with a differential equation or mask as extra loss. // Per
Machine Learning Physics Guess – Should not the magnetic field be turned off some time before fusion starts? Otherwise you missconfigure the mass? // Per
Amazing Machine Learning Math – Where there is smoke there is fire. Meaning that normal probability function should not be so clean in the long tail. It should smoke out. // Per Lindholm

Archives

site owner: Per Lindholm (peroglyfer@gmail.com)

Play Soundbite freewill

Proudly powered by WordPress

We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.Ok