[ HOML ] Chapter 01 - The Fundamentals of Machine Learning

Posted on | In , ,
  • A famous up on YouTube recommand a book called \
  • The notes covers: Why use machine learning, types of machine learning system, Main Challenges of Machine Learning

Why Use Machine Learning

  • Problems for which existing solutions require a lot of hand-tuning or long lists of rules(spam filer).

  • Complex problems for which there is no good solution at all using a traditional approach(speech recognition).

  • Fluctuating environments: a Machine Learning system can adapt to new data.(speech recognition)

  • Getting insights about complex problems and large amounts of data(data mining)

Types

Supervised/Unsupervised Learning

Supervised Learning

  • Classification
  • Regression
  • Popular algorithms
    • KNN
    • Linear Regression
    • Logistic Regression
    • Support Vector Machines
    • Decision Trees and Random Forests
    • Neural networks(Partly)

Unsupervised learning

  • popular tasks:
    • cluster
    • visualization
    • dimensionality reduction
    • anomaly detection
  • Clustering
    • k-means
    • Hierarchical Cluster Analysis
    • Expectation Maximization
  • Visualization and dimensionality reduction
    • PCA
    • Kernel PCA
    • Locally-Linear Embedding
    • t-distributed Stochastic Neighbor Embedding
  • Association rule learning
    • Apriori
    • Eclat

Semisupervised learning

  • Tasks
    • photo-hosting services(cluster the same person and name the person)
  • Algorithms
    • deep belief networks

Reinforcement Learning

  • agent, the learning system observe the environment itself, select and perform actions, and get rewards or penalty in return.

Batch and Online Learning

Batch learning

  • Also called offline learning, as it is trained using all the available data and apply what is has learned.
  • training on the full set of data

Online learning

  • feeding it data instances sequentially, either individually or by small goups called mini-batches
  • good for limited computing resource: you can discard your data if the model is updated
  • Learning rate is an important parameter for an online learning system, deciding how fast they should adapt to changing data or how fast it forget old data.
  • Challenge: bad data feeding can greatly decline the performance of the system
    • You should monitor your system closely and promptly switch learning off.

Instance/Model Based Learning

Instance based learning

Model based learning

  • To measure the fitness of the mdoel:
    • you can specify a performance measure. define a utility function(fitness function)
    • or define a cost function define how bad the model is.
  • Code examples:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import sklearn

oecd_bli = pd.read_csv('datasets/lifesat/oecd_bli_2015.csv',thousands=',')
gdp_per_capita = pd.read_csv("datasets/lifesat/gdp_per_capita.csv",thousands=',',delimiter='\t',encoding='latin1',na_values='n/a')

#oecd_bli.INEQUALITY.drop_duplicates()
oecd_bli = oecd_bli[oecd_bli["INEQUALITY"]=="TOT"]
oecd_bli = oecd_bli.pivot(index="Country", columns="Indicator", values="Value")
oecd_bli

gdp_per_capita.rename(columns={"2015": "GDP per capita"}, inplace=True) #把原本叫2015的列重命名为gdp...
gdp_per_capita.set_index("Country", inplace=True)
full_country_stats = pd.merge(left=oecd_bli, right=gdp_per_capita, #index=True 使用index作为合并的公共键
                                  left_index=True, right_index=True)
full_country_stats.sort_values(by="GDP per capita", inplace=True)

remove_indices = [0, 1, 6, 8, 33, 34, 35]
keep_indices = list(set(range(36)) - set(remove_indices))
country_stats = full_country_stats[["GDP per capita", 'Life satisfaction']].iloc[keep_indices]

x = np.c_[country_stats["GDP per capita"]] #将数组拆开（一号轴）然后重组（二号轴）‘
y = np.c_[country_stats["Life satisfaction"]]
country_stats.plot(kind='scatter', x="GDP per capita", y='Life satisfaction')
plt.show()
# Select a linear model
model = sklearn.linear_model.LinearRegression()

# Train the model
model.fit(X, y)

# Make a prediction for Cyprus
X_new = [[22587]]  # Cyprus' GDP per capita
print(model.predict(X_new)) # outputs [[ 5.96242338]]

Main Challenges of ML

Insufficient Quantity of Training Data

  • The Unreasonable effectiveness of data

Nonrepresentative Training Data

  • Sample selection bias
  • US presidential election in 1936

Poor-Quality Data

  • Outliers
  • Missing a few features

Irrelevant features

  • Process Feature Engineering
    • feature selection
    • feature extraction
    • create new features by gathering new data

Overfitting the Training Data

  • Regularization : constraining a model to make it simpler and reduce the risk of overfitting

  • degrees of freedom: the number of parameters

  • Hyper-parameter to control for the amount of regularization

Underfitting the Training Data

  • The model is too simple