Monthly Archives: April 2018

Machine Learning Speculation – Network Compression Ratio?? Water parameters?

From GAN network models I wonder. Since the generator depends on noise. How little noise is necessary to make it work.

Could we get it to work with a little as a single random data value, like a bit. Varying with time randomly.

So from this I speculate that there is an effective way to compress information. But requires perhaps a fast computing power.

So physically, water is important to life and water is a network. Since life is also complex. Water might be complex to. The weight parameters of water to sustain all life on the planet might be so large that it needs to be compressed somehow.

So what do I mean by water is needed for life. I guess that ?computing ready weight parameters in the water acts like the surrounding network between life networks.

?All contact surfaces to water needs water to sustain life.

Basically water is important.

Subtle Intelligence Idea – Perfect Representation vs Revealing Representation

Compare two images below and feel. What image gives you the most intelligent feeling?

For me its the flat shaded sphere image. Possibly because a perfect photo would be too amazing and a non perfect photo feels a little bit wrong. The amazing feeling does not give much information to the imagination. Which is The tool for intelligence.

Between these is the revealing image. The flat shaded one. The one that reveals how it works. A little any way. But still to the brain it makes all the difference.

Since Peace Is The Most Import Thing – What is A Peace Country Network?

Whenever you don’t know the answer, assume a network. So what does the peace country network look like?

What infrastructure do you need etc. Could creating a city result in a positive win win. Creating jobs. Also could reading peace as a school subject help. How do you change a troubled country to peace. What input to the network works?

Maybe peace is way more achievable than we think.

Gravity – Network Within A Network As Unsupervised Clustering?

Whenever you don’t know the answer, assume a network. Then for gravity you have at least three networks. The two masses and the in between space.

Then you can begin to speculate. In what sense does gravity make to all the networks.

Why and what reason could exist for masses to move?

In machine learning networks you have unsupervised clustering. Making groups of similar objects. So mass is a ?single class object.

The problem of gravity is then a physics network solution to single object clustering between mass and ?empty space. // Per Lindholm 2018-16-04

Battery Speculation – If The Electric Battery Is A Network. Is It Computing Unsupervised Classification?

A quick battery idea.

Since electric networks exist. Then surely the battery is a network. Then just because its possible for a network to calculate unsupervised clustering into groups.

Does it do this for some reason? Just because its beneficial somehow. To solve the problem.

Then how can the electric battery network be taken care of. Some further reason for new types of batteries perhaps.

Are there more properties of networks that can be translated to batteries?

Philosophy Idea – the (Problem, Question, Answer) triplet + Probability?

I guess there exist Problem, Question and Answer triplets that need to be adapted before solving. Problems are not always questions. Since you can ask different questions. They are  not always related to the problem.

Where in the triplet possibility does the relation lay. What I mean is.

To solve certain triplets the problem might have to downsized or up sized. I guess this could be said for the question and answer also.

Also. Can we be sure we got it right for every problem? There might be probabilities for the triplet. At least to the answer.

Mathematics Singularity Idea – Problem_Dependent_Division – All Possible Problems Are Not The Same

One interesting insight is that the set of all possible problems are not as compressible as mathematics assume. That is. Math and problems go hand in hand. Then since problems changes so could math or network math.

One such example is that of singularity. I speculate that math functions gets adapted to the problem. Not assuming the singularity problem are like every other problem. If you start at a singularity problem its perhaps a little bit harder. But otherwise I assume you approach the singularity from an easier problem.

So if you have a variable with a limited range from 0 to 9. Then any function that carries that variable outside its range would impose a problem. Likewise I think the singularity is related to infinity as its an mathematical idea number. So the solution to the singularity is also an idea.

So here is what I propose. Under smart network mathematics you could replace functions with network model functions or network ideas.

So for a problem of 1/0 you replace / with a different network_division(a,b) function. This particular division model function knows your entire problem and adapts its result.

The result is depending on your problem so we got a new idea. There is always a (problem, math model) pair that exists. So a math model could be a machine learning model that solves division near the problem specific singularity.


Peace Idea – You Get The ?Most Good Results For The Money From Empowering Poor People In Russia and America. Help Them Build Affordable Houses And Let Them Help Build Up Their Country Again

Why not let dignity and peace win. I believe a key to peace lays with the poor people. Those on the brink of becoming homeless. Empower them regardless of political relations. I believe this strategy is a smart way for peace and the people.

My Mathematical Idea – Complex Noise Number, a+i*model(noise), inital guess

Complex numbers solves a particular problem. My idea is that this the complex part can be expanded to solve other problems. However finding that particular problem is hard.

So assume the problem requires a network to be able to solve it. Inspired by this the complex part could be a machine learning model of noise, model(noise).

I assume a model with noise as input requires a discriminator model and some iteration to get a good result.

You should then be able to iterate the number to change the parameters according to the error discriminator system.

If successful the model will have produced a good solution.

Philosophical Possibility Idea – Line Shaped Noise – Classification Of Line Direction And Length

As a philosophical possibility its possible to have noise as classification data in a graph although not visible as in my below sketch.

So my idea is that you divide the data. The plot. Into a number of lines. Then calculate something about the would be noise that classified gave the lines its directions and length.

The idea could have predictive capability I think perhaps.

Line plot shaped by classification of noise – Idea by Per Lindholm

Machine Learning Math – What Is A Constant?

In machine learning ?every in-data are assumed to make sense to the target label. You update and learn the weights of the model. According to errors you get from the comparison of the target and the prediction.

From this I can guess that every decimal in a constant like pi should make sense to a network.

For me its close to saying that there are many significant decimals at least for pi.

So then should every decimal make sense to a network? Are there a constants which has an infinite number of significant decimals?

Math Innovation Guess – Set of Probably Inventions. The Faulty Calculator Number

A quick way to innovate is to imagine.

If a number is a rule element object. Then could certain numbers have other rules?

I’m thinking. What if this is a problem of certainty. Somewhere in the universe this could still happen. I mean. It could be an important problem that leads to new innovation.

So I imagine the calculator device was expensive item. Somewhere it broke down. Producing some faulty results for some number. This could happen. The people could not afford a repair.

This leads to a guess sentence. Problems that are very likely to happen should lead to useful things. Useful in that other similar problems can be solved.

So if the number 5 has some problem. How many workarounds does there exist?

Could the mechanical calculator provide other problems that the computer chip based calculator usually don’t show.

For instance. A mechanical calculator could perhaps experience fatigue. Then if this is beneficial in some way. Could this be simulated for a computer.

Are there virtual fatigue concept in mathematics. Where the system has not broken down but has ?other rules.

Machine Learning Idea – Other Ways To Solve A Problem. First Do A Cleaner Looking Compete Version?

The idea is to work the input towards an easier recognition. For a letter dataset this would mean that the first network makes a cleaner version of the hard to recognize letters. That is. A smart transformation that produces a simpler to recognize image.

I think this would mean that for already clean looking letters there would be no different image.

I wonder if you can take the images with the best confidence of score of recognition. Use them as examples for the rewrite network model. That is. Use the correct best selected images with the usual data,label network.

This so the network can ask if the generated image is clean or fake. It competes with the set of most recognizable images. That is. I think I should mix in all the recognizable images from the training set into one set of most recognizable images. To extract the clean property I mean.

Then I guess I run the procedure in reverse. First the clean competition then the recognition network.

One problem could be that if non letters get into the mix they would get produce the wrong decision.

So another problem is to recognize if its a letter at all. Again some network that tells if the image is a letter or fake. It almost feels like this is a kind of biological mimicry. Fooling the network that you belong to a another set of specie. But here we have another property, eatable.

For biological mimicry the fooling specie I think should be most successful if it can stay relatively close to the target specie but not too close. This so it will not try to reproduce with those it cant and are not eaten.

Amiga Soundenhancer Idea – Enhance Youtube Audio Realtime Using Linux calf-plugins

The little cheap Sound Enhancer hardware was a great addon to the Amiga Computer. I always wanted that sound ever since. Now hmm a little late I tried the calf audio plugins with Linux. So I thought I run some Amiga Demos from youtube into the calf audio plugins via qjackctl.

You need QjackCtl, pulseaudio-module-jack and calf-plugins. Which you apt install from the terminal.

Then you run ”pactl load-module module-jack-sink” in the terminal. Then you need to select Jack sink under the sound settings as output.

Be sure to check qjackctl so that you dont have multiple outputs to the system out channels.

In the video you can see how it went.

Towards Train Station Innovation – How Do You Create A Great Train Station Experience. Museum Experience Or Maybe Swedish Nature Films By Regular People

I was wondering. How do you improve the train station experience?

If you start simple. To see if things work. Why not add an entertainment page at the website. The idea is to show something like Swedish nature films on the Just youtube links. Then let travelers vote.

So you can create competitions on the best nature films. By the public and professionals. I got this idea from a TV program where they showed nature photos from places around the country from viewers.

High Resolution cameras are cheaper today. So why not take advantage of them.

Then maybe in the future if successful we could have best films air at the train stations somehow.

Machine Learning Idea – Using A GAN For SuperResolution

A quick idea this morning.

What if there is fast way to make a photo super high resolution?

Splitting the photo into a larger grid with photo pixels and additional transparent pixels should do the trick. Just use a GAN network to imagine the extra pixels to real looking color pixels. An extra pixel could be stored in a separated pixel matrix.

Then this could work for other GAN problems. Just use the model on a small downscaled photo and then apply pixel separation and GAN imagination to increase the resolution. Should be much faster than to do it all at once for a large image.

Machine Learning Physics – Balls Of Steel As A Machine Learning Network?

I was wondering. Philosophically you can place balls of steel in a 2D pyramid and apply a little voltage. So you created something that looked like a machine learning network.

So I was wondering. Would it be possible to learn a little more about the electric network. For use as inspiration in batteries and motors.

From my machine learning speculation it could be that the layers of the network oscillates a little in finding its target weights. Knowledge about this could perhaps make things more effective?

Are there electric machine learning elements objects one can use to enhance motors and batteries for instance.

Thinking outside the box creating a parallel electric network. Instead of the crude fully connected network. Convolution networks are important in machine learning. What counterpart is it possible to create physically.

Maybe we get a CPU Network electrical Wire in the future.

Internet Innovation – Make Use Of Subdomains. Attract More Companies To Your City.

Why not create a positive spiral of Internet visible companies in a city or municipality. Add an optional city subdomain to the Internet address like

So to for example. We have Gekås in Ullared then you just register .ullared. in the address field.

We have many companies in different cities. So why not add the .cityname. as an optional subdomain. It will create new websites and advertising for the city.

Draw in more companies because of the probability of success estimation people do when they associate many companies with a place.

Optional here means the old address will point to the same webpage.

Physics Speculation, Time – Everything Is Just Classifier And A Generator Of Noise

Using a GAN network I think I have the answer to the many difficult questions. Noise. Noise put trough a generator and iterated with the help of a discriminator can create just about ?anything. From mass, energy, time and so on.

So if you want to travel in time you need to figure out a way to generate mass and time from noise and iterate it with a discriminator associated with that mass and target time.

So to understand this idea use the following guess sentences.

Because it makes sense to a network.

Since it makes sense to nothingness.

So we have one new idea to the why question.

So in my mind what came before nothingness. Maybe there is some math here. If time is a generator(noise) and noise is related to frequency then there should be the possibility the noise somehow restarts itself. So we got the start of a new time ?counter. To make use of a fully expanded universe looking like empty ?noise another generator is needed perhaps with new time to use.