Supervised Virus treatment – From the set of first possibilities. You have to eat even if your sick. So the idea is to boil salt and water together with toilet paper with some of your nose snot in. The idea is that water will learn what virus it is and instruct the salt residuals in the vapor to kill it. // Per

Mega Battery Idea – Transparent surface battery. Photochemistry activation and adaptation with laser like a CDRW recorder. Exact control enables lots of energy storage? // Per

Mega Battery idea – Fully connected battery. Extra connections between three batteries. Inspired by a simple multilayer perceptron machine learning model. To be able to better calculate its ion causality the battery needs more connections. // Per

Machine Learning Aided Coding – Can programming code generate feelings. If so then it could be used to mitigate bugs? Learn supervised bugs then express the label with a feeling face?

Machine Learning Physics Guess – Could magnetic fields be filled with quantum particles. A way to create a fully connected network. Directed north south in a loop iteration. Calculating everything the forces should be able to do? // Per

Series Generator Invention? Guess – Connect electric generators in series with loosely connected magnetic gears? // Per

Vertical take off Airplane Invention – Connect electric motors in series with loosely connected magnetic gears. The idea is to use magnetic forces to bypass one revolution if not in perfect sync? More adaptability means you can connect in series. Much more effective for the power

Machine Learning Math – Given pressure all problems can be solved. Now how to implement that guess in ML. // Per

Water Science guess – Water molecules are anything but simple. Its needs to know and predict all logic so it does not disturb anything in the brain.

Forest fire – Fire Fighting Competitions. Shortest amount of time to put out a forest fire. Its an engineering problem that needs new ideas.

Forest Fire Prevention – Large Self going Water spreader could perhaps aid forest/bush fires. How many water spreaders would you need for a km wide target?

Idea – Stand by me movie with Syrian refugee kids

Grid idea -Could ?unaccounted for grid energy losses be mitigated with irreversible filters. An electric generator connected an electric motor. The idea is that everything with electricity must be used to make the grid better. Battery, transformer, flywheel, motor, generators.

It could be that model2 are life organism with freshwater and saltwater looks at the complexity required for perfect freshwater. Not all water is the same is my guess.

Water Desalination Idea – If freshwater is the target. Then it ?must have an information signature that separates it from saltwater. For water itself to solve the problem of producing freshwater from the sea. «Freshwater information» // Per

Changed my Rpi4 today. Maybe it has a power issue, some undervoltage. With a official power adapter though. Disconnected the usb ssd to make sure and it still runs. Please hardware update the adapter. 300MB/s from SSD usb3 is pretty good to leave out. Manjaro rocks barryboot.

Looking at possibilities. Seriel dependent information could be energy and mass is then parallel energy. If you got information that has to be unraveled in a certain order then it can perform work. Thats is my guess. // Per

Philippines Idea – I liked the collect trash for rice idea so I wonder if a collect trash for a raspberry pi 2, 3 or 4 could work. The idea is to teach office. Here Libreoffice. Write text documents and export to pdf cv, presentations, excel sheets. Basic computer knowledge. /Per

Invention – DLSR Camera addon for the Raspberry Pi? Innovate the camera space with a computer and linux. For proffesional photographers. // Per

Super Wikipedia Idea – Use machine learning to pair a problem with wikipedia links to articles relevant to that image or text. Something very useful. Wow // Per

Like super conducting cables. Maybe they dont exist for constant or sinwave voltages. It has to be the right voltage at the right time.

Mega Machine Learning Physics – Circle Stochastic Voltages. Inspired by ML iteration. See what inventions can be made from voltages taken from a random set. Ex. np.random.choice([0,1,2,3,…,12]) and repeat that for the duration. // Per

Machine learning physics speculation – Can work be split into motion work and thought work. Guess based on that we need to think while we train. So efficiency or heat reduction is possible if material thinking got easier. Light, pressure, heat, vibration, forces. Easy for a brain

Machine Learning progress – Superstable error correction. x = F.relu(self.l0(x)) x = F.relu(self.l1(x)) x = (self.l2(x)) c0 = F.relu(self.l3(x)) #corrections? By Per Lindholm c1 = F.relu(self.l4(x)) c2 = (self.l5(x)) return x,c0,c1,c2 F.mse(c0-yt,F.squared_error(c0,yt))

Machine Learning progress – Last layer is the prediction of the loss. As predictions are easy when loss is constant zero. It helps as an extra loss function. Last layer is the predicted loss. // Per