Kohonen maps or Self-Organizing Maps (SOM) or Self-Organizing Characteristic Maps (SOFM) is a teacherless machine studying method used to create a low dimensional (normally two-dimensional) area to symbolize a multivariate information set whereas preserving the topological construction of the information. For instance, have a dataset of p variables measured in n observations – these could be represented as clusters of observations with the identical values of the variables. These clusters can then be represented as a two-dimensional map, with observations in shut clusters having extra comparable values than these in distant clusters. This will simplify visualization and evaluation of multivariate information.
Kohonen maps have been developed by Finnish mathematician Teuvo Kohonen within the Eighties.
A Kohonen map consists of a grid of neurons which might be related to neighboring neurons. Throughout coaching, enter information is offered to the community and every neuron calculates its similarity to the enter information. The neuron with the very best similarity turns into the winner and its weights are adjusted to raised match the enter information.
Over time, neighboring neurons additionally alter their weights to change into extra just like the successful neuron, leading to a topological ordering of neurons within the map. This self-organization course of permits complicated relationships between inputs to be represented in a decrease dimensional area by way of Kohonen maps. This functionality could be very invaluable for information visualization and clustering.
The purpose of the self-organizing map algorithm is to get totally different elements of the community to reply in the identical method to sure enter patterns. That is partly motivated by the way in which visible, auditory, and different data is processed in sure elements of the human mind.
Let’s have a look at how this algorithm works as utilized to mathematical calculations and the MQL5 code.
The event of the algorithm consists of 4 fundamental steps:
Step 1 is initialization of weights . Random values can be utilized. Different parameters resembling studying fee and variety of clusters are additionally initialized on this step.
A self-organizing map is a kind of synthetic neural community that, in contrast to different sorts of synthetic neural networks that be taught by error, resembling again propagation with gradient descent, Kohonen maps use aggressive studying.
In aggressive studying, neurons in a Kohonen map compete towards one another to change into the “winner”. To do that, the neuron have to be probably the most just like the enter information.
Within the coaching section, every enter information level is represented on the Kohonen map. The similarity between the enter information and the burden vector of every neuron is then calculated. The neuron whose vector is most just like the enter information is set because the winner or the “finest matching unit” (BMU).
The BMU is chosen primarily based on the smallest Euclidean distance between the enter information and the neuron’s weight vector. The successful neuron then updates its weight vector, making it extra just like the enter information. The burden replace formulation used is the Kohonen studying rule, which strikes the burden vector of the successful neuron and its neighboring neurons nearer to the enter information.
Advantages of Kohonen Playing cards
Kohonen maps have a number of benefits:
They will seize non-linear relationships between the enter information and the output map, that’s, they’ll deal with complicated patterns and information buildings.
They will discover patterns on unlabeled information. That’s, they can be utilized when the labeled information is inadequate or tough to acquire.
By decreasing the dimensionality of the enter information, the computational complexity of subsequent duties resembling regression and classification is lowered.
Topological relations between the enter information and the output map are preserved: shut neurons within the map correspond to comparable areas within the enter area, which might help in information exploration and visualization.
Proof against noise and outliers within the enter information if the noise is simply too massive.
Kohonen maps or self-organizing maps are an modern strategy to buying and selling that may assist merchants navigate the markets. Utilizing unsupervised studying, Kohonen maps can establish patterns and buildings in market information, permitting merchants to make knowledgeable selections. Kohonen maps can establish non-linear relationships in information and cluster information into acceptable teams. Nonetheless, when working with them, one ought to pay attention to potential drawbacks resembling sensitivity to initialization, lack of formal convergence standards, and many others. Basically, Kohonen Playing cards present wonderful leads to buying and selling the markets, I’ve used Matrix and Kohonen Playing cards in my New EA Falcon Rig EA, which you’ll be able to check out in your buying and selling now.