Smallest neural network of nature (2009)

Several molecules could exchange information among themselves during their random collisions. This event could be considered as an information exchange between molecules through wireless communication. Can we find any logic in the random motion of the molecules? We did find such logic. This is very interesting to see that molecules are spontaneously confined within a fixed space without any external interference and then they exchange information collectively. The appearence of a virtual boundary is also helpful because it electronically isolates a system.

Why a neural network arising out of a fuzzy system is important?

1. It enables us to find out fundamental parameters of a theoretical neural network. Moreover, we can find out some critical parameters which is impossible to find using theoretical simulations.

2. Historically, since 1943, all neural network models are basically inspired by the brain function and the architecture, it is also possible to verify the mathematical models proposed to explain the remarkable brain features.

3. In the 1930s, information theory came into being. For the first time it was defined "What is information?" Information theory was based on single bit switches and there were no provisions for the collective information exchange.

Once neural network based on collective information exchange has been established we can redefine the information theory for a multilevel logic system.

How one can define information if molecules are computing?

This is a very interesting topic for discussion since information is related to entropy and the sencond law of thermodynamics.

What can molecular neural network do that other parallel technologies can not?

1. Molecular neural network could be produced trillions in numbers identical and reproducible in its evolving properties.

2. Molecular neurons could be redesigned by atomic scale engineering. Adding potential features into the molecular wont be a singular event.

3. Wireless communication is very fast (basically its surface density wave modulated communication).

4. MNN can encode dynamic memory.

5. STM and LTM, collective memory all these concepts are very easy to realize if neuron is a molecule.

1. It is known that building two identical cells is challenging task and the neurons of a general artificial neural network has complex machinery. Encoding all levels of complexity into those cells appears nearly impossible.

2. Artificial neurons like swarms robots are not that flexible for redesigning.

3. Wireless communication is possible but extremely low.

4. Dynamic memory concept does not exist in the theory of general ANN.

5. This is nearly impossible to operate all the technologies parallely, we need an intelligent hardware attached with a neuron.