]]>

more...We continue to explore reinforcement learning methods. In previous articles we discussed methods for approximating the Q-learning Reward function and the policy gradient function learning. Each method has its own advantages and disadvantages. It would be great to use the maximum of their advantages when building and training models. When trying to find methods minimizing the shortcomings of the algorithms used, we often try to build certain conglomerates from various known algorithms and methods. In this article, we will talk about a way of combining the above two algorithms into a single model training method, which is called.Advantage Actor-Critic algorithm)

---------------------

- 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
- Neural networks made easy (Part 15): Data clustering using MQL5
- Neural networks made easy (Part 16): Practical use of clustering
- Neural networks made easy (Part 17): Dimensionality reduction
- Neural networks made easy (Part 18): Association rules
- Neural networks made easy (Part 19): Association rules using MQL5
- Neural networks made easy (Part 20): Autoencoders
- Neural networks made easy (Part 21): Variational autoencoders (VAE)
- Neural networks made easy (Part 22): Unsupervised learning of recurrent models .....
- Neural networks made easy (Part 23): Building a tool for Transfer Learning
- Neural networks made easy (Part 24): Improving the tool for Transfer Learning
- Neural networks made easy (Part 25): Practicing Transfer Learning
- Neural networks made easy (Part 26): Reinforcement Learning
- Neural networks made easy (Part 27): Deep Q-Learning (DQN)
- Neural networks made easy (Part 28): Policy gradient algorithm
- Neural networks made easy (Part 29): Advantage Actor-Critic algorithm

]]>more...In this article, I will try to find out if the algorithm is actually good for solving complex problems. In the classical version of the algorithm and in many of its modifications, there are significant limitations associated with the fact that the optimized function must be smooth and continuous, which means it is completely unsuitable for discrete functions. However, in line with the series of articles, all the algorithms under consideration will be changed in such a way (if there are any restrictions) in order to eliminate the shortcomings, at least to make the algorithms work at least purely technically. In other words, all algorithms must be indifferent to the smoothness of functions (such as in traders' problems) and have no restrictions on the argument step.

]]>

more...We continue studying different reinforcement learning methods.

]]>

more...K-Nearest Neighbors Algorithm is a non-parametric supervised learning classifier that uses proximity to make classifications or predictions about the grouping of an individual data point. While this algorithm is mostly used for classification problems, It can be used for solving a regression problem too, It is often used as a classification algorithm due to its assumption that similar points in the dataset can be found near one another; k-nearest neighbors algorithm is one of the simplest algorithms in supervised machine learning. We will build our algorithm in this article as a classifier.

]]>

more...In the previous article, we started exploring reinforcement learning methods and built our first cross-entropy trainable model. In this article, we continue to study reinforcement learning methods. We will proceed to the Deep Q-Learning method.

]]>

more...In the previous articles of this series, we have already seen supervised and unsupervised learning algorithms. This article opens another machine learning chapter: Reinforcement Learning.

]]>

more...We all as traders draw many lines while trading to help us to observe some important levels to take a suitable trading decision based on them. So, these lines are very important for us as traders and we may wonder if we have a method that can be used to draw these lines or take a suitable decision based on them automatically because I think that it will help us a lot. The answer is yes, we have a method to do that by MQL5 (MetaQuotes Language 5).

]]>more...We continue to study the Transfer Learning technology. In the previous two articles, we created a tool for creating and editing neural network models. This tool will help us transfer part of the pre-trained model to a new model and supplement it with new decision layers. We will also check the usability of our tool.

---------------------

- 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
- Neural networks made easy (Part 15): Data clustering using MQL5
- Neural networks made easy (Part 16): Practical use of clustering
- Neural networks made easy (Part 17): Dimensionality reduction
- Neural networks made easy (Part 18): Association rules
- Neural networks made easy (Part 19): Association rules using MQL5
- Neural networks made easy (Part 20): Autoencoders
- Neural networks made easy (Part 21): Variational autoencoders (VAE)
- Neural networks made easy (Part 22): Unsupervised learning of recurrent models .....
- Neural networks made easy (Part 23): Building a tool for Transfer Learning
- Neural networks made easy (Part 24): Improving the tool for Transfer Learning
- Neural networks made easy (Part 25): Practicing Transfer Learning

]]>

more...Is a Machine learning paradigm for problems where the available data consists of unlabeled examples. Unlike supervised learning techniques such as regression methods, SVM, decision trees, neural networks, and many others discussed in this article series, where we always have labeled datasets that we fit our models upon. In unsupervised learning, the data is unlabeled so, it's up to the algorithm to figure out the relationship and everything else on itself.

Examples of unsupervised learning tasks are clustering, dimension reduction, and density estimation.

]]>

more...Any technical indicator is based on a certain algorithm for processing market information. As a rule, prices are used as initial data. Speaking in the language of math, the indicator is a function that converts prices into some final result. In this article, I will consider linear functions that can be used to build an indicator.

]]>more...In the previous article in this series, we have created a tool to take advantage of the Transfer Learning technology. As a result of the work done, we got a tool that allows the editing of already trained models.

Furthermore, the created tool allows not only editing trained models. It also allows creating completely new ones.

Such a useful toll should also be as user friendly as possible. Thus, in this article, we will try to improve its usability.

---------------------

- 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
- Neural networks made easy (Part 15): Data clustering using MQL5
- Neural networks made easy (Part 16): Practical use of clustering
- Neural networks made easy (Part 17): Dimensionality reduction
- Neural networks made easy (Part 18): Association rules
- Neural networks made easy (Part 19): Association rules using MQL5
- Neural networks made easy (Part 20): Autoencoders
- Neural networks made easy (Part 21): Variational autoencoders (VAE)
- Neural networks made easy (Part 22): Unsupervised learning of recurrent models .....
- Neural networks made easy (Part 23): Building a tool for Transfer Learning
- Neural networks made easy (Part 24): Improving the tool for Transfer Learning

]]>

more...A new article with a new technical indicator in our series as we will learn how to design a trading system based on one of the most popular technical indicators that's the Fractals indicator.

]]>

more...After reading this article, you will receive a complete and working mathematical model, as well as learn to understand and correctly calculate everything related to orders.

]]>more...When optimizing trading systems, the most exciting things are metaheuristic optimization algorithms. They do not require knowledge of the formula of the function being optimized. Population algorithms involve the simultaneous handling of several options for solving the optimization problem and represent an alternative to classical algorithms based on motion trajectories whose search area has only one candidate evolving when solving the problem.

]]>

more...We continue our immersion in the world of artificial intelligence. What is Transfer Learning and why do we need it? Transfer Learning is a machine learning method in which the knowledge of a model trained to solve one problem is reused as a basis for solving new problems. Of course, to solve new problems, the model is preliminarily additionally trained on new data. In the general case, with a properly selected donor model, additional training runs much faster and with better results than training a similar model from scratch.

---------------------

- 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
- Neural networks made easy (Part 15): Data clustering using MQL5
- Neural networks made easy (Part 16): Practical use of clustering
- Neural networks made easy (Part 17): Dimensionality reduction
- Neural networks made easy (Part 18): Association rules
- Neural networks made easy (Part 19): Association rules using MQL5
- Neural networks made easy (Part 20): Autoencoders
- Neural networks made easy (Part 21): Variational autoencoders (VAE)
- Neural networks made easy (Part 22): Unsupervised learning of recurrent models .....
- Neural networks made easy (Part 23): Building a tool for Transfer Learning