Machine Learning Physics Idea – Could light teleport information by creating a copy of the room to which it is going to teleport to. Then the universe updater would as it is optimized make the corresponding change to two rooms. As the rooms have the same classification label. Per
Machine Learning Math – Can a line approximation be an interesting machine learning problem. I mean with a model() function like the brain its not an approximation.
Machine Learning Idea – Future small prediction of the loss. Store loss values as a array then use np.polyfit(…) to get a linear prediction. Use that loss in a two transform prediction. // Per
Python code-blocks from if 1==1: statements. Just as a reminder to myself. import delayed_loss from delayed_loss import * importlib.reload(delayed_loss) // Per
Food Market Idea – Tent with solar panels for refrigeration of fruit and food. Would save the whole world. // Per
Food Market Idea – Expandable refrigerator. Extend the cooling volume. For flexible livelihood. // Per
Machine Learning – FAT matrix – Increase probability by removing the ability for the model to remove information. Like a universe. From model1() reduce(W2) = W1 and store W2 in larger model2. // Per
Machine Learning Physics – Assume there exist force mathematics. Then by this set of mathematical problems you can extract a force from acceleration, heat, magnetism, light or vibration… Simply by putting a more complex function in the output stream of the atom model() // Per
Machine Learning Physics – Could there exist variable time steps. Since to get to the next state the derivative must be constant under that small dt. So a time step is as long as the derivative is constant in some scenario.”Everything is relative to calculations at 100% accuracy”
I wonder. Would it be possible to create a virus detector from an old type capacitor. The idea is that the virus will jump from one side to the other and a current/voltage drop will be detected. // Per
Turn the electricity noise into .wav for listening and classifications of viruses.
Uses on y1,y2,y3 = model(X) loss = mse(mean(y1),1) loss += mse(mean(y2),2) loss +=mse(mean(y3),3) 3006^2 + 4008^2 = 5010^2 = 25100100
Super Intelligence Math – Dual math belong to inconsistant math. If a = b for t->inf but if t>inf a != b
Speed Of light – Speed is position iteration from start to target. With a loss function that goes to zero at the target. There could therefore exist superiteration or super speeds over lightspeed. The reason for limit of light speed for mass could be all the extra hidden variables all over ?must also iterate faster. // Per
Rain Innovation – To generate rain clouds. To make it rain at a location. Try spraying dissolved salt in ?hot water from a spray bottle in the air. This is not cloud seeding. Salt needs to have combined with hot water. I think it has to do with freshwater must not contain virus/bacteria so the water model will take advantage of salt being against viruses. // Per
Corona Virus – Develop a Double Effect Cure? – Steam Therapy from a boil of water+salt+citrus and some much simpler vaccine. Double effect cure. Steam for mucus vaccine for something else.
Universe – How it all works – guess – There need to be loss functions all over the place for functions to adapt its parameters to. Everywhere. So water and empty space could if the know enough simulate causality and loss functions for every other process. // Per
Corona Food Crisis – Could you keep food fresh with CPU processing calculations? Iterating the cool temperature and pressure with a machine learning model() should hold food fresh for much MUCH longer? Loss function is food gone bad so its simple. // Per
Super important idea. Keep food fresh much longer.
Machine Learning Uncertainty – delta learning rate * delta recall memory > zero . Recall many samples by running many or all samples in small transform form. Like crop of the samples. This is how you remember. First you train on full samples then on just the cheat sheets but a lot of them. Keeping batchsize data as before. // Per
Physics – Heisenberg uncertainty rewritten to kinetic and potential uncertainty delta_ke*delta_pe>value gives for light with high velocity high delta_ke and lowest delta_pe associated with zero mass. // Per
Machine Learning Battery idea – What if ion distance is a non linear function? Space expands around the ions non linearly depending on what the model or atoms it contains. So certain atom are in charge of space expansion? Check the outer border of space of such atoms // Per
Machine Learning Physics – Existance could be a training accuracy in time. Reverse training and the object ceases to exist. // Per
Machine Learning Math – Functions – Math can solve certain kinds of truths. But why not all truths? What if a mathematical function would output text and numbers. // Per
Machine Learning Physics guess – Predict the Heisenberg uncertainty and extract energy from the conservation of the uncertainty by calculating the ?migrating fly around bird formation algorithm.
Make Earth’s surface more reflective – No. If you heat space. The nearest star the sun will have a hard time with its processes. Residual space. It needs space to be free of extra disturbances. Creating a counter productive situation.
Energy generation guess – Predict the electron position and velocity for one instance in time and it moves. This ?is how you generate/get energy. Use computing power to predict a number of electrons. Get them to move in some magnetic field generating electrical power.Per Lindholm