Introduction to Artificial Intelligence Notes Outline
0 Report
Introduction to Artificial Intelligence Notes Outline
Related Recommendations
Other works by the author
Outline/Content
See more
Machine learning
Definition
What is machine learning
Spam classification
Task T: Identify emails as spam and not spam
Performance P: Correctly identify emails as spam/not spam
Training Lesson E: Observe the emails you mark as spam or not spam
Handwriting recognition learning problem
Task T: Recognize and classify handwritten text in images
Performance P: classification accuracy
Training experience E: Handwritten text database with known classification
Classification of machine learning
supervised learning
K-nearest algorithm
decision tree
linear regression
support vector machine
unsupervised learning
clustering
autoencoder
weakly supervised learning
semi-supervised learning
transfer learning
source domain and target domain
core goals
Transfer knowledge from source domain to target domain
Implementation method
Sample migration
Feature transfer
Model migration
reinforcement learning
Target
Strategy
explore
use
Classification by learning outcome
return
Classification
clustering
Dimensionality reduction
Important concepts
Examples & Samples
Dataset
properties and characteristics
attribute space
eigenvector
learning and training
training data
training samples
training set
hypothesis
the truth
generalize
General process
Collect data
amount of data
Data quality
Distribution of data
Data preparation (preprocessing)
Possible problems with data
missing data
noise
inconsistent
redundant, imbalanced
repeat
normalization
feature engineering
Data set partitioning
Training set: training model
Test set: evaluate the final model
Validation set: used for tuning and selecting models
Data cleaning
delete
Manual filling
autofill
outlier
type conversion
Numerical type
continuous
Discrete
non-numeric type
There are categories
non-category
Choose an algorithm to build a model
Select model
Training model
Evaluation model
Performance metrics
precision
ROC
loss function
Swarm intelligence algorithm
Definition
Swarm intelligence
Swarm intelligence algorithm
Genetic algorithm
basic genetic operator
Select operator
crossover operator
mutation operator
five basic elements
Parameter encoding
binary encoding
One-dimensional chromosome encoding method
advantage
shortcoming
real encoding
Initial group settings
Generation of initial population
Determination of population size
Design of fitness function
deception problem
premature convergence
stagnation phenomenon
genetic manipulation design
Control parameter settings
cross
a little crossover
two point intersection
crossover probability
Mutations
concept
Randomly change some bits in the individual code
main purpose
Basic content
mutation probability
mutation method
site variation
reverse mutation
insertional mutation
Search technology
Search strategy
Blind search: no information guidance
heuristic search
blind searchB
depth first search
Basic idea
Breadth first search
Basic idea
Compare similarities, differences, advantages and disadvantages
Heuristic search
A algorithm
A* algorithm
Game search
a-b pruning algorithm
Monte Carlo Tree
choose
Expand
simulation
return
Confidence online decision-making method
Factor 1
Factor 2
0 Comments
Next Page