For this model, the mutual combination (quadratic term or cubic term) of each feature variable becomes a new feature variable, then each time there is a hig executive email list h-order term, the feature variable will become an order of magnitude larger, When executive email list there are more than thousands of feature variables, the regression algorithm is very slow. And we convert it to a neural network to solve it, it will be much simpler. Neural network model: 2. Why is it called a neural network? Let's take a look at how human neurons are connected: Human nerve cells are intricately connected.
One neuron receives information from multiple neurons, and after processing the information, it transmits the information to the next neuron. The design inspiration executive email list of the neural network comes from the information transfer between nerve cells. We can regard each circle in the neural network as a neuron, which accepts the input of the upper executive email list -level network, and then transmits the information after processing. to the next level network. 3. Why simulate a human neural network?
What's so magical about human neural networks? Neuroscientists have done an interesting experiment, cutting out the auditory cortex of an animal and executive email list transplanting it into the brain of another animal to replace its visual cortex, so that the signals received from the eyes will be transmitted to the transplanted auditory cortex, and finally The results executive email list showed that the transplanted auditory cortex also learned to "see". There are also many interesting experiments: The image above is an example of learning to "see" with the tongue, which helps blind people see