EnDe neural network (modified)

[Prakash]

EnDe neural network

This model comprises methods like encoding and decoding. It involves encoding information in one network and decoding the reduced information in another network. The model consists of layers such as input, intermediate, and output. The input layer receives information, the intermediate layer stores the reduced information, and the output layer extracts information from the intermediate layer. [Untitled33_20231011142323] Input layer These layers receive inputs in the form of values associate…

This is a unmodified concept of this model. Let me share the modified one in this thread.

• Begin by creating a set of layers.

• Within these layers, randomly connect neurons to other layers of neurons. The value of these randomly connected neurons should be set at 1/4th of the average value of the input values, for example, 0.2 for an input value of 0.10.

• Design the randomly connected neurons in a way that approximately 20 of them attach to the highest values in the input sequence, while the remaining 7 connect to the lowest values.

• Add placeholders in the neurons positioned in layers where these randomly connected neurons will be positioned during input processing. Typically, 3 or 4 placeholders are created for each neuron in the layer, except for the input and output neurons, where 6 or 7 placeholders are connected.

• In the encoding network, implement a function with a fixed range of analysis that seeks the highest values among 3 or 2 neurons. Only the values from the highest neurons are allowed to pass into the next layer.

• In the decoding network, include a function with a fixed range of analysis that searches for the lowest values in 3 or 2 neurons. This design choice is made because, in the encoding network, randomly connected neurons add 0.2 to the input values, while in the decoding network, they subtract 0.2 from the input values. This encoding and decoding process helps maintain the context of the input information.

• Even in the decoding network, the randomly connected neurons are configured to target the highest and lowest values in the input sequence.

• The range of these functions should gradually decrease and become more stable as the input data passes through the layers. For example, in the first two layers, the range of functions is set at 2, and in the subsequent two layers, it becomes 3. Beyond that point, the range of functions remains fixed at 4. This range may be adjusted based on the complexity of the input information.

First network - Encoding the information

Second network - Decoding the information

Feel free to comment your thoughts.