Neural networks made easy (Part 14): Data clustering
Quote:
In this series of articles, we have already made a substantial progress in studying various neural network algorithms. But all previously considered algorithms were based on supervised model learning principles. It means that we input some historical data into the model and optimized weights so that the model returned values very close to reference results.
There is another approach to Artificial Intelligence learning methods — unsupervised learning. This method enables model training only using the original data, without the need to provide reference values.
In this article, you will not see the previously used vertical structure of a neural network consisting of several neural layers. But first things first. Let's consider the possible algorithms and see how they can be used in trading.
more...
---------------------
- Neural networks made easy
- Neural networks made easy (Part 2): Network training and testing
- Neural networks made easy (Part 3): Convolutional networks
- Neural networks made easy (Part 4): Recurrent networks
- Neural networks made easy (Part 5): Multithreaded calculations in OpenCL
- Neural networks made easy (Part 6): Experimenting with the neural network learning rate
- Neural networks made easy (Part 7): Adaptive optimization methods
- Neural networks made easy (Part 8): Attention mechanisms
- Neural networks made easy (Part 9): Documenting the work
- Neural networks made easy (Part 10): Multi-Head Attention
- Neural networks made easy (Part 11): A take on GPT
- Neural networks made easy (Part 12): Dropout
- Neural networks made easy (Part 13): Batch Normalization
- Neural networks made easy (Part 14): Data clustering
1 Attachment(s)
MQL5 Wizard techniques you should know (Part 02): Kohonen Maps
Attachment 46120
Quote:
Continuing with this series on the MQL5 wizard, we will delve into Kohonen-Maps for this one.
more...
Developing a trading Expert Advisor from scratch (Part 14): Adding Volume At Price (II)
Quote:
Our EA already has some resources which assist in trading — we have added them in our previous articles. However, this EA has some issues with visualization and resizing. They do not interfere with trading, but at some points this leads to a mess on the screen until you force it to refresh. In addition, there are still some things missing that would provide valuable information for us. These are specific things, but the information can be necessary.
So, let's start implementing these new improvements. This interesting article can provide some new ideas and methods of presenting information. At the same time, it can assist in fixing minor flaws in projects.
-------------------
more...
Developing a trading Expert Advisor from scratch (Part 15): Accessing data on the web (I)
Quote:
There are a lot of websites and places on the web, featuring a huge amount information. What you need to know is where to look and how best to use this information. And the better you are informed during the right period, the better it is for trading. However, if you are going to use a browser, whatever it is: you will find that it is very difficult to filter certain information well, that you are forced to look at many screens and monitors, and in the end, although the information is there, it is impossible to use it.
But thanks to MQL5, which is very close to C/C++, programmers can do more than just work with a chart as is: we can search, filter, analyze data on the web and thus perform operations in a much more consistent way, than most traders because we are going to use all the computing power in our favor.
-------------------
more...
Neural networks made easy (Part 15): Data clustering using MQL5
1 Attachment(s)
The price movement model and its main provisions (Part 1): The simplest model version and its applications
Attachment 46241
Quote:
Up to the present, we have not had any mathematically rigorous price movement theory. Instead, we have had to deal with experience-based assumptions stating that the price moves in a certain way after a certain pattern. Of course, these assumptions have been supported neither by statistics, nor by theory. The article provides the concepts and foundations of the mathematical apparatus of the rigorous theory price movement theory.
more...
Developing a trading Expert Advisor from scratch (Part 16): Accessing data on the web (II)
Quote:
In the previous article "
Developing a trading Expert Advisor from scratch (Part 15): Accessing data on the web (I)", we presented the entire logic and ideas behind the methods of using the MetaTrader 5 platform to access marked data from specialized websites.
In that article, we considered how to access these sites and how to find and retrieve information from them in order to use it in the platform. But it doesn't end there, as simply capturing data doesn't make much sense. The most interesting part is to learn how to take this data from the platform and use it in an Expert Advisor. The method to do this is not so obvious and so it is hard to implement without knowing and understanding all the functions available in MetaTrader 5.
-------------------
more...
Neural networks made easy (Part 16): Practical use of clustering
Data Science and Machine Learning (Part 06): Gradient Descent
Quote:
The idea is to take repeated steps in the opposite direction of the gradient (or approximate gradient) of the function at the current point, because this is the direction of steepest descent. Conversely, stepping in the direction of the gradient will lead to a local maximum of that function; the procedure is then known as gradient ascent.
----------------
more...
Developing a trading Expert Advisor from scratch (Part 17): Accessing data on the web (III)
Quote:
In the previous article
Developing a trading Expert Advisor from scratch (Part 16): Accessing data on the web (II), we talked about the problems and consequences of data capturing from the web. We also considered how to use it in an Expert Advisor and discussed three possible solutions each having their pros and cons. However, we have a new problem: the system of terminal's global variables only allows the use of the double type. Many do not know how to avoid this, and therefore they pass various information, such as a piece of text, through the channel provided by MetaTrader 5. In this article, we will discuss how to get around this limitation.
-------------------
more...
