"New Neural" is an Open Source neural network engine project for the MetaTrader 5 platform. - page 18

 

In general, ideally, you should get something like this http://www.peltarion.com/products/synapse/comparison.php.

At least you can be guided by what inquisitive mind of a non-programmer user may want. :)

Product Comparison - Synapse vs. The World
Product Comparison - Synapse vs. The World
  • www.peltarion.com
Synapse was designed to blow away the competition and it is exactly what it does. It is the most advanced and most easy to use software of its kind on the market. Below is a short comparison with the more common software available. The full list of Synapse features go far beyond the ones listed below.
 
Mischek:

Ouch ! That's not what I know ?

Unless you can guess ... I think I've only told Victor so far.

Are you intriguing, bragging, inviting ? // underline as appropriate

per worker's request :)

I'll tell you right off the bat -- I don't think you'd be interested.

 
joo:

In general, ideally, it should be something like this http://www.peltarion.com/products/synapse/comparison.php

Mm-hmm. You're a killer. It's a 1000 grn license... You're looking up to me :)
 
TheXpert:
Uh-huh. You're a killer. The license costs 1,000 grivnas... You've got a good point :)
Compared to the capabilities of Neurosolutions and NSDT, you can say for nothing.
 
joo:

In general, ideally, you should get something like this http://www.peltarion.com/products/synapse/comparison.php.

At least you can be guided by what inquisitive mind of a non-programmer user may want. :)

Now explain me a lamer, which of these mats are neural network types, types of learning and what are trademarks:

Data input------------------

Text
Image
Signal generator
SQL database
User script
User custom built
Data output------------------
Text
SQL database
User custom built
Preprocessing---------------
Data rearrangement
PCA
Outlier removal
Query-based filtering
Statistical
User script
Interactive visualization
Postprocessing-------------
Error analysis and tracing
Confidece analysis
Sensitivity analysis
Real-time testing
Adaptive paradigms-------
Ststic neural networks
Dynamic neural networks
Kernel machines
Bayesian methods
LSTM
Fuzzy logic
Optimazation paradigms-




Genetic algorithms
Particle swarm
Monte carlo
Architectures---------------
MLP
Generalized MLP
Modular netwoks
Self-organizing maps
Neural gas
Competitive learning
Hebbian
FFCPA
Radial basis networks
LSTM
Time lagged recurrent
Partially recurrent
Wavelet networks
Fully recurrent
Neuro-fuzzy
Support Vector Machines
Custom architectures
System design---------------
Component based
Plug-in based xml-ph-NET compliant
User interface----------------
Graphical construction
Scalable user interface
Block embedded GUI
Deployment-------------------
Solutions are deployable
.NET compatible components
Reusable components
Scalable deployment
Royalty free use of deployed systems

 
Urain:

Now explain to me, as a lamer, which of these mate types of neural networks, types of learning and what are trademarks:

MLP
Generalized MLP
Modular netwoks -- ugh
Self-organising maps
Neural gas
Competitive learning
-- unclear
Hebbian -- dunno, but if what I think is interesting
FFCPA -- dunno
Radial basis networks
LSTM -- ugh
Time lagged recurrent
Partially recurrent
Fully recurrent
I wonder how one differs from the other.
Wavelet networks - I don't know
Neuro-fuzzy - this is probably what I was talking about.

Support Vector Machines - dunno

xz -- something I'm not familiar with

 

Learning/optimization algorithms:

Genetic algorithms





Particle swarm





Monte carlo


Network Types:

MLP





Generalized MLP





Modular netwoks





Self-organising maps





Neural gas





Competitive learning





Hebbian





FFCPA





Radial basis networks





LSTM





Time lagged recurrent





Partially recurrent





Wavelet networks





Fully recurrent





Neuro-fuzzy





Support Vector Machines





Custom architectures


 
We've already seen the possibilities of visual block programming in MQL5. So, everything is possible, it's up to the enthusiasts. :)
 

At least some kind of battle plan.

By the way, why don't they have BeckProp? Or did I misunderstand something.

There's only Genetics, Bee Swarm and Montecarlo?

 
joo:
By the way, just a side note. Most packages don't have genetics for networks at all. And in this list, it is only for checkboxes or if heuristics are missing (imho).