But, recently the world of computation had changed by the introduction of AI and particularly artificial neuron layers. This enabled our system to handle a much larger amount of data compared to conventional algorithms. Why is that important? Well, we are living in a world of big data, everything is digitalised and thus we have a tremendous amount of data, produced in each second. This made a problem in computing these. Artificial neuron networks became a solution for this. This article is not aiming to connect the actual brain and these but is just to tell you some similarities that we could spot.

If you are considering artificial neural layers, one of its features is that it could analyse complex data. Before that let’s consider how it works. Each unit of such a layer is called a neuron. It could take a variable amount of inputs with different weights. There would be a function acting on this weighted sum. This function could be a sigmoid function, ReLU, etc. If it is giving a binary output sigmoid function is only chosen in the last layer as the slope of the function reduces the determines the rate of learning. In the deeper layers ReLU is preferred as it has a higher constant slope towards the ends compared to sigmoid function. When many such units are combined to form layers and many such layers are stacked together, we could call it as a neural network. And the efficiency and function depends on the number of layers.

Despite the technicality, the main advantage over the conventional algorithm, or one feature that resembles humans is that it could analyse unstructured data! That is consider facial recognition done by one such network. Let’s say it is five layers deep. At the first layer, there is no wonder, the units are looking for edges. Each unit would be looking at edges having a different orientation. Next layer would be looking at the possible shapes. But, when you are going to the final layers, it would be looking at the complete faces. One would be surprised at this point to see, how simple numerical calculations occurring at different layers are able to recognise unstructured data.

Can we find a similar structure in humans? This would be a rather interesting question to address. Well, we can consider our eye itself. If we are considering the receptive fields in the eye, it is actually sensitive to edges. There are on centred off surrounding receptive fields and also the one in the reverse order. Then how we are able to recognise the face of someone?

If you are considering the brain of mammals, there is an important process of corticalization. Evolutionarily, it is first observed in mammals and neuronal cell bodies are arranged in six different layers in the cortex and it is called neo-cortex. So, we have layers here also? Does it contribute to the complex functions of our brains?

I can’t say yes or no to this. In fact, we didn’t know what kind of logic is used in actual neurons. Despite the similarities and speculations, one could make, the very nature of logic used by the cells still pause a challenge to us. One speculation is that the artificial neuron layers are like a reduced model of a diverse picture of actual neurons, in which case the functions could be even more complex than we expect. In both cases, information is flowing from one layer to another and the strength of connections are changing depending on the type of feedback.

But, my friend, we are not yet at the end of this discussion to conclude something. We are discussing some similarities. And it is more likely that the actual brain could be a forest of diverse functions and it could pause more challenge to us.