User:J. C. Martinez-Garcia/Notebook/HMS Activities/2008/11/06
|Inheritance systems, robustness and the model||Main project page|
Previous entry Next entry
Some extra comments on the artificial neural network
I tried some others neural network (the same architecture, but different number of neurons in the hidden layer). It seems that there is not an spectacular difference between the performance of the original network (10 neurons) and the reduced versions.
I continued with the lecture of the book by Jablonka and Lamb. I was reading about the way Behavioral Inheritance Systems (BIS) works. They argue that observed behavioral inheritance3 systems, like the one involving food preferences in young rabbits, works through the transmission of information (in this case the odor of the food that the mother ate, impregnated in the skin, in the milk, and even in the feces) which provides to the young rabbits with some clues. What is important is that the behavior of the young rabbits converges to the behavior of the mother (or close relatives), because of the given stimuli. There is not such a thing like a supervised learning (in this case). However, the behavior is essentially transmitted from one generation to the next one. The mechanism is similar to the inheritance of sustained loops in the case of epigenetic inheritance systems.
I continue with the mathematical modeling of the reduced version of the AKT signaling pathway. I will focus my effort in the desensitization mechanism and the phosphorilation process.
The Tzachi proposal