Re #Angel_Investor: #Machine_Learning, #Genetic_Algorithms, cryptocurrency and #SPEL


On 16-05-11 08:46 PM, Angel Investor wrote:
> > Hi
> >
> > Can you describe what is ‘machine learning’? Actually, if we are
> > into it, what is ‘genetic programming’ too?

machine learning encompasses a wide range of different things,
but mostly it has to do with analyzing and making sense of large
amounts of data. For instance face recognition, and self driving cars
use machine learning. Machine learning has a variety of types, but the
most hyped up one right now is deep learning, which is inspired by
brain neurons.

Genetic programming is a form of AI based on genetic algorithms that
evolves programs. So the DNA is the program, there are many random
combinations made initially, and then they are tested for fitness,
those which are closest to the goal (most successful/fit) are mated,
and mutated to make children, and those are tested, and so on. It can
also include “geographies” where different processors work on their
own populations, with a certain chance of successful individuals from
one population going over to another.

> >
> > Would it be possible to earn or supplement living in these areas
> > or these are mostly ‘academic curiosity’ stuff? It’s not that I am
> > against any of your choice, just would like to know more.
> >
> > Thanks,

Machine learning, and deep learning in particular is all the rage
right now (very popular), especially in Silicon Valley. People that
have any AI skills are being bought up very quickly for vast sums of
money. Start-ups are being bought for many millions of dollars, and
are fundraising millions of dollars also.

Recently on IRC someone mentioned that my project has enough
“buzzwords” that I could probably sell an unfinished product to
venture capitalist for 50 million. While I myself am not interested
in “selling out”, it is the general atmosphere in the silicon valley
IT world.

For instance Google bought Deep Mind (an AI company) for 500 million
Google Acquires Artificial Intelligence Startup DeepMind For More Than $500M

Recently Google released Tensorflow, which is now used by deepmind in
most of their products/services.
Other AI companies have raised many millions, for example Sentient
Technologies has raised over a hundred million

Those are both examples from 2014, more recently the Korean government
decided to dedicate 860 million to AI

Also in late 2015 high profile character Elon Musk, donated 10 million
to create OpenAI, which has a billion dollars in funding to start

So yeah, it’s pretty trendy stuff, lots of money floating around for it.

The biggest hype right now is deep learning neuro nets, but I think
it’s only a matter of time before the market gets saturated, and
people realize that it has serious limitations.
By contrast Watson (the IBM robot that won Jeopardy), is a composite
AI with many different kinds of AI working together, not just machine
learning (deep and shallow), but also expert systems, confidence
estimation, searches, hypothesis generation, and lots of other things
in parallel.

AI is a pretty huge field.

Genetic Programming is more relevant to the project I’m working on.
Recently two angel investors followed me on twitter. Since I’m
ethically opposed to closed source, only angel investors (such as
yourself) could really give money at this time.

By contrast many of the larger sum fundraising companies raised money
by giving people percentages of the private company. There is a way of
doing something similar with Open Source projects, and that is by
giving out project-centric crypto currency. For instance Etherium
raised 18 million dollars in bitcoins a year before they had any
software. MaidSafe raised 7 million in 2006, and then in 2014 with
still no product, they raised 10 more million. They are making some
noise on their blog to make it seem like they may have an initial
Alpha version available soonish.

Anyways, so my idea is SPELCoin, a cryptocurrency for the SPEL project,
people file bugs or feature requests, and then if they want to help
motivate them getting solved, they add some SPELCoin, which is given
to the people that fix the bug or make the feature. This way SPELCoin
has value, as it is like a freelance software marketplace, for various
products that are written in SPEL. It could also do related things
like translation.

Now humans programming and translating is all well and good, but
really by the mid 2020’s Ray Kurzweil predicts that as much computing
power as the human brain will be available for roughly $1000USD (1995
price adjusted). Currently the cheapest GFLOPS come from GPU’s, so
likely it will be GPU’s or similar architecture that will be providing
this huge amount of processing power. By 2030’s it will be only a few
hundred dollars to get as much computing power as a human brain.

The idea is to leverage this processing power. for instance the
Bitcoin network is currently the fastest super computer on the planet,
outranking the top 500 super-computers many times over.
It is currently running at 17 zetta flops.
by comparison the top super-computer is running at 33-55 peta-flops —
more than 300-500 times slower.
so the bitcoin network is faster than the top 500 super computers

Anyways so the point of the genetic programming is that people could
have their computers work for them, in writing code, solving bugs and
feature requests. For instance people could even request features
from their own computer, and if it has enough resources it could do it
itself, but if they want additional resources, they could put a
bounty/reward, and the network will help solve the feature request.

Feature requests don’t have to be limited to computer programs, as
with genetic algorithms, and other AI algorithms such as machine
learning that likely will be used in later stages, so it will be
possible to request just about anything, and then the computer will
set to work in finding solutions.

SPEL is the enabling factor that makes it possible, because the
specifications of bugs and feature requests have to be in a language
that both humans and computers can readily produce and understand.


> >
> > —–Original Message—– From: Logan Streondj [] Sent: May-11-16
> > 11:18 AM To: Angel Investors Subject: Re: Computer “Course”
> >
> > The machine learning course from google says it takes approximately
> > 420 hours to complete. So if I use all my spare-time on it (after
> > Caleb goes to bed, and before I do), then it will take
> > approximately 3-4 months or 600-800 USD.
> >
> > On 16-05-11 10:10 AM, Logan Streondj wrote:
>> >> There is another option, I can get a “nanodegree” in Machine
>> >> Learning, it is developed by google, and costs $200 a month.
>> >>–
> >
>> >>
>> >>
> >
> >
> > 09
> >
>> >> On 16-05-09 10:40 AM, Logan Streondj wrote:
>>> >>> Hi,
> >
>>> >>> I looked at the various course options that are available,
>>> >>> unfortunately most of them are very shallow, or teach things
>>> >>> which are already readily available to learn for free online.
> >
>>> >>> So I’ve found some good textbooks on Genetic Programming.
> >
>>> >>> Foundations of Genetic Programming Langdon Genetic systems
>>> >>> programming : theory and experiences. (ed.), Studies in
>>> >>> computational intelligence Nedjah, Nadia, Ajith Abraham and
>>> >>> Luiza de Macedo Mourelle: Genetic Programming and#8211; On the
>>> >>> Programming of Computers by Means of Natural Selection John R.
>>> >>> Koza
> >
>>> >>> combined they are $140CAD from abebooks (online used book
>>> >>> store).

Logan Streondj,
A dream of Gaia’s future.

Speakable Programming for Every Language:

You can use encrypted email with me,
how to:
key fingerprint:
BD7E 6E2A E625 6D47 F7ED 30EC 86D8 FC7C FAD7 2729


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s