The foothold is in neuron networks and deep learning.

The foothold is reasoning, symbolic reasoning and machine reasoning.

The foothold is perception and action.

The foothold is behavior control, adaptive and evolutionary computing.

Additive relationship

relationship

The latter model must be built on top of the previous model

Independent

Function

Module

Number

Character string

import time

from time import *

import time as t

from time

K The larger the value, the easier the model is to overfit

K The larger the value, the smoother the segmentation surface of the classification

K Value is a hyperparameter

can k Value is set to 0

Atlas 800 model: 9000

Atlas 800 model: 3000

Atlas 800 model: 3010

TRUE

FALSE

True

False

Backpropagation update parameters

Forward the dream and update

Backpropagation calculation result

Forward calculation results

Recurrent neural network can be unfolded according to the time axis

LSTM Unable to solve the problem of vanishing gradient

LSTM It is also a recurrent neural network

Recurrent neural network can be abbreviated as RNN