자료구조 1차 과제 - WordPress.com · Figure 10-5. Perceptron diagram . w. is the connection...

Post on 06-Aug-2020

0 views 0 download

transcript

2

3

4

Walren McCulloch & Walter Pitts proposed a very simple model of biological neurons

5

The Percepton is one of the simplest ANN architectures, invented in 1957 by Frank Rosenblatt.

6

7

Hebb’s rule : “Cells that fire together, wire together.”=> The connection between two neurons is increased whenever

they have the same output.

- The perceptron does not reinforce connections that lead to the wrong output. - It is fed one training instance at a time, and for each instance it makes its predictions.

8

9

10

11

12

13

14

TF.Learn offers a Scikit-Learn-compatible API.

15

MNIST dataset

The first step : construction phase

The second step : execution phase

16

17

18

19

20

Next Step : Fine-Tuning NN Hyperparameters

Random search, Grid search …

21

It has a actually been shown that an MLP with just one hidden layer can model even the most complex functions provided it has enough neurons.

But! It is the fact that deep networks have a much higher parameter efficiency than shallow ones: they can model complex functions using exponentially fewer neurons than shallow nets, making them much faster to train.