Answer the question
In order to leave comments, you need to log in
Is it true that a neural network in the case of classification is, roughly speaking, an ensemble of logistic regressions?
Just now I began to delve into deep learning, having more or less figured out the classical ML, and therefore I want to draw some analogies for myself in order to understand the essence, because The topic seems to be very difficult.
Answer the question
In order to leave comments, you need to log in
No.
An ensemble is always a generalization of several solutions into one - sequential, parallel, more sophisticated - but there are always several solutions on the basis of which one is somehow determined.
NN is always the search for one solution by optimizing the coefficients of the equation (well, the weights of the connections, if it's easier for you).
If you have understood the "classical ML", then you should understand that any task in this area is an optimization task. Sometimes explicitly so formulated, sometimes - a
little veiled. And NN is just another method for finding that very optimum. No more and no less. About any connection with "live" neurons, synapses and dendrites, etc. - forget immediately after reading.
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question