A N I N T R O D U C T I O N TO T H E M A G I C A L L A N D O F A RT I F I C I A L I N T E L L I G E N C E
Machine Learning For Dummies
T H E D I G I TA L G R O W T H C O M PA N Y
Projekt
Unsere Services
Digital
Products
CRM
Data
Entrepreneur
Network
Creative
Consulting Customer
Digital
Marketing
OUR MINDSE T
The world is changing rapidly. We are
fascinated by these upheavals. We are tackling
digital transformation - together with our
partners. We enjoy experimentation and
demand to find the best digital solution for any
given Problem!
2011
Wurde die TOWA Digitalagentur gegründet.
28
Jahre ist unser Durchschnittsalter.
84
Mitarbeiter sind bei TOWA beschäftigt.
Philipp Freytag v. Loringhoven
D IRECTO R DATA CO NSULTING
Marketeer | Dev | Designer | Gamer | Chef
Artificial Intelligence
Machine Learning
What is Intelligence?
What defines Learning?
Intelligence
A mental capability, that involves the abilities:
▪ to reason,
▪ plan
▪ solve problems
▪ think abstractly
▪ comprehend complex ideas
▪ learn quickly
▪ learn from experience!
Learning
Learning is the transformative process of
taking in data and turning it into information
and knowledge – when internalized and
mixed with what we have experienced so far
– it changes what we „know“ and builds on
what we do.
It‘s based on input, proces and reflection.
Artificial Intelligence
Machine Learning
DEF INITIO N
It is the Science and engineering of
designing intelligent machines, especially
intelligent computer programs.
Miss by a faktor of 1000
2018: 33 Zettabytes 1
Until 2025: 175 Zettabytes 1
We just don’t know
1 https://de.statista.com/statistik/daten/studie/267974/umfrage/prognose-zum-weltweit-generierten-datenvolumen/
In Machine Learning, What is Better: More
Data or better Algorithms?
PE TER NORVIG , GOOGLE DIRECTO R RESE ARCH
We don’t have better algorithms. We just
have more data.
What he means is:
Better Data != More Data
The issue is that better data does not
mean more data. As a matter of fact,
sometimes it might mean less!
E XPERTS MAKE ERRORS ASWELL
The NASA lost 328 million $, because the
systems within a satellite did not use the
same units of measurement thoughout
Quelle: Wikipedia
E XPERTS MAKE ERRORS ASWELL
TARGET lost $5.4 billion in Canada, partly
because its inventory system was loaded
with incorrect data.
Quelle: CanadianBusiness
Data comes in all forms and sizes
A feature is an individual measurable
property or characteristic of a phenomenon
being observed
Feature Examples
ID Product
Catrgory
Name Color Size Sales
13 Clothing Pants Red 12 13
16 Shoes Wedges Blue 40 8
22 Shoes Ankle Boots Green 38 13
29 Accessories Necklace Yellow 3 3
„Big Data“
VOLME VERACIT YVARIE T Y VELOCIT Y
NATE SILVER , STATISTIKE R UND PUBLIZIST
Data is useless without context.
Machine Learning is a group of algorithms used to recognize structures in
data.
The concept assumes that it is possible to train a model (algorithm) with
data in such a way that it can make decisions.
So what does ML do?
Machine Learning
Teach (train) a model (algorithm) with
experience (data)
How does it work?
INPUT DATA
How does it work?
INPUT DATA
Model
ALGORITHM
How does it work?
INPUT DATA
Model
ALGORITHM OUTPUT DATA
Classification
supervised
▪ Labelled input
unsupervised
▪ Unlabelled input
x1
x2
x1
x2
Classification
supervised unsupervised
x1
x2
x1
x2
▪ Labelled input
▪ Classification and
Regression
▪ Unlabelled input
▪ Clustering and
dimension
reduction
Popular supervised Algorithms
▪ Nearest Neighbor
▪ Naive Bayes
▪ Decision Trees
▪ Linear Regression
▪ Support Vector Machines (SVM)
▪ Neural Networks
Popular unsupervised Algorithms
▪ k-means clustering
▪ Association Rules
What is deep learning?
▪ AI > Machine Learning > Deep Learning
Machine Learning
Artificial Neural Networks, we‘re copying our brain!
Artificial Neuron
Inputs Output
„Activation“
Artificial Neural Networks
Deep? = Amount of Layers (hidden)
between Input and Output
ROGER KIMBAL
Welcome to The Information Age.
Data, Data Everywhere,
But Nobody Knows a Thing!
Business ImpactTechnology
Data
Successful
AI Project
What makes a great AI Project?
Applications
Traffic Predictions
Generate Haikus from news
Colorization of Black and White Videos
Adding Sounds To Silent Movies
Object Classification and Detection
Smart answers in Google Inbox
Game Playing
Translations
Is Machine Learning the answer to
everything?
Lets get to work!
Tools for your Choice
▪ Microsoft: https://azure.microsoft.com/de-de/services/machine-learning-studio/
▪ IBM: https://www.ibm.com/de-de/cloud/watson-studio
▪ MlJar: https://mljar.com/
▪ BigML: https://bigml.com/
▪ Datoin: https://datoin.com/
Microsoft Machine Learning Studio
Predict survival on the Titanic
QUESTIO N
Let's discuss, what do you think is the
most important reasons passengers
survived the Titanic sinking?
What we‘ll do
1. Signup to Microsoft Machine Learning Studio
2. Download the Data: https://tinyurl.com/y6qg2qh8
3. Load the data into ML Studio
4. Let‘s Explore the Data
Data Features
▪ pclass: A proxy for socio-economic status (SES)
1st = Upper
2nd = Middle
3rd = Lower
▪ age: Age is fractional if less than 1. If the age is estimated, is it in the form of xx.5
▪ sibsp: The dataset defines family relations in this way...
Sibling = brother, sister, stepbrother, stepsister
Spouse = husband, wife (mistresses and fiancés were ignored)
▪ parch: The dataset defines family relations in this way...
Parent = mother, father
Child = daughter, son, stepdaughter, stepson
Some children travelled only with a nanny, therefore parch=0 for them.
What we‘ll do
1. Clean Data
▪ Drop Name,
▪ Ticket
▪ Cabin
2. Convert Text Data to Numerical Data
▪ Sex
▪ Pclass
▪ Embarked
Evaluation Meaning
▪ Mean Absolute Error (MAE): The mean value of the absolute errors. (An error is
the difference between the predicted value and the actual value).
▪ Root Mean Squared Error (RMSE): The square root of the average square of the
forecast errors for the test dataset.
▪ Relative Absolute Error: The mean value of the absolute errors relative to the
absolute difference between actual values and the average of all actual values.
▪ Relative Squared Error: The average of squared errors relative to the squared
difference between actual values and the average of all actual values.
▪ Coefficient of Determination: This value, also known as the R square, is a
statistical measure of how well a model fits the data.
In other words
Small Errors = Good!
High Coefficient of Determination = Good!
The closer Coefficient of Determination is to 1 the better!
Contact me in case of digital!
TO WA. THE D IGITAL GRO W TH CO MPANY
TOWA
▪ Instagram: www.instagram.com/towa.digital/
▪ LinkedIn: www.linkedin.com/company/2099786/
▪ Facebook: www.facebook.com/towa.digital
▪ Web: www.towa-digital.com
PHILIPP
▪ Instagram: instagram.com/ploringhoven/
▪ LinkedIn: linkedin.com/in/philipploringhoven/
▪ Mail: philipp.loringhoven@towa.at

Machine Learning for dummies!

  • 1.
    A N IN T R O D U C T I O N TO T H E M A G I C A L L A N D O F A RT I F I C I A L I N T E L L I G E N C E Machine Learning For Dummies
  • 3.
    T H ED I G I TA L G R O W T H C O M PA N Y
  • 4.
  • 5.
    OUR MINDSE T Theworld is changing rapidly. We are fascinated by these upheavals. We are tackling digital transformation - together with our partners. We enjoy experimentation and demand to find the best digital solution for any given Problem!
  • 6.
    2011 Wurde die TOWADigitalagentur gegründet. 28 Jahre ist unser Durchschnittsalter. 84 Mitarbeiter sind bei TOWA beschäftigt.
  • 7.
    Philipp Freytag v.Loringhoven D IRECTO R DATA CO NSULTING Marketeer | Dev | Designer | Gamer | Chef
  • 8.
  • 9.
    What is Intelligence? Whatdefines Learning?
  • 10.
    Intelligence A mental capability,that involves the abilities: ▪ to reason, ▪ plan ▪ solve problems ▪ think abstractly ▪ comprehend complex ideas ▪ learn quickly ▪ learn from experience!
  • 11.
    Learning Learning is thetransformative process of taking in data and turning it into information and knowledge – when internalized and mixed with what we have experienced so far – it changes what we „know“ and builds on what we do. It‘s based on input, proces and reflection.
  • 12.
  • 13.
    DEF INITIO N Itis the Science and engineering of designing intelligent machines, especially intelligent computer programs.
  • 15.
    Miss by afaktor of 1000 2018: 33 Zettabytes 1 Until 2025: 175 Zettabytes 1 We just don’t know 1 https://de.statista.com/statistik/daten/studie/267974/umfrage/prognose-zum-weltweit-generierten-datenvolumen/
  • 16.
    In Machine Learning,What is Better: More Data or better Algorithms?
  • 17.
    PE TER NORVIG, GOOGLE DIRECTO R RESE ARCH We don’t have better algorithms. We just have more data.
  • 18.
    What he meansis: Better Data != More Data The issue is that better data does not mean more data. As a matter of fact, sometimes it might mean less!
  • 19.
    E XPERTS MAKEERRORS ASWELL The NASA lost 328 million $, because the systems within a satellite did not use the same units of measurement thoughout Quelle: Wikipedia
  • 20.
    E XPERTS MAKEERRORS ASWELL TARGET lost $5.4 billion in Canada, partly because its inventory system was loaded with incorrect data. Quelle: CanadianBusiness
  • 21.
    Data comes inall forms and sizes
  • 22.
    A feature isan individual measurable property or characteristic of a phenomenon being observed
  • 23.
    Feature Examples ID Product Catrgory NameColor Size Sales 13 Clothing Pants Red 12 13 16 Shoes Wedges Blue 40 8 22 Shoes Ankle Boots Green 38 13 29 Accessories Necklace Yellow 3 3
  • 24.
    „Big Data“ VOLME VERACITYVARIE T Y VELOCIT Y
  • 25.
    NATE SILVER ,STATISTIKE R UND PUBLIZIST Data is useless without context.
  • 26.
    Machine Learning isa group of algorithms used to recognize structures in data. The concept assumes that it is possible to train a model (algorithm) with data in such a way that it can make decisions. So what does ML do?
  • 27.
  • 28.
    Teach (train) amodel (algorithm) with experience (data)
  • 29.
    How does itwork? INPUT DATA
  • 30.
    How does itwork? INPUT DATA Model ALGORITHM
  • 31.
    How does itwork? INPUT DATA Model ALGORITHM OUTPUT DATA
  • 32.
  • 33.
    Classification supervised unsupervised x1 x2 x1 x2 ▪ Labelledinput ▪ Classification and Regression ▪ Unlabelled input ▪ Clustering and dimension reduction
  • 34.
    Popular supervised Algorithms ▪Nearest Neighbor ▪ Naive Bayes ▪ Decision Trees ▪ Linear Regression ▪ Support Vector Machines (SVM) ▪ Neural Networks
  • 35.
    Popular unsupervised Algorithms ▪k-means clustering ▪ Association Rules
  • 36.
    What is deeplearning?
  • 37.
    ▪ AI >Machine Learning > Deep Learning Machine Learning
  • 38.
    Artificial Neural Networks,we‘re copying our brain!
  • 39.
  • 40.
  • 41.
    Deep? = Amountof Layers (hidden) between Input and Output
  • 42.
    ROGER KIMBAL Welcome toThe Information Age. Data, Data Everywhere, But Nobody Knows a Thing!
  • 43.
  • 44.
  • 45.
  • 46.
  • 47.
    Colorization of Blackand White Videos
  • 48.
    Adding Sounds ToSilent Movies
  • 49.
  • 50.
    Smart answers inGoogle Inbox
  • 51.
  • 52.
  • 53.
    Is Machine Learningthe answer to everything?
  • 55.
  • 56.
    Tools for yourChoice ▪ Microsoft: https://azure.microsoft.com/de-de/services/machine-learning-studio/ ▪ IBM: https://www.ibm.com/de-de/cloud/watson-studio ▪ MlJar: https://mljar.com/ ▪ BigML: https://bigml.com/ ▪ Datoin: https://datoin.com/
  • 57.
  • 58.
  • 59.
    QUESTIO N Let's discuss,what do you think is the most important reasons passengers survived the Titanic sinking?
  • 60.
    What we‘ll do 1.Signup to Microsoft Machine Learning Studio 2. Download the Data: https://tinyurl.com/y6qg2qh8 3. Load the data into ML Studio 4. Let‘s Explore the Data
  • 61.
    Data Features ▪ pclass:A proxy for socio-economic status (SES) 1st = Upper 2nd = Middle 3rd = Lower ▪ age: Age is fractional if less than 1. If the age is estimated, is it in the form of xx.5 ▪ sibsp: The dataset defines family relations in this way... Sibling = brother, sister, stepbrother, stepsister Spouse = husband, wife (mistresses and fiancés were ignored) ▪ parch: The dataset defines family relations in this way... Parent = mother, father Child = daughter, son, stepdaughter, stepson Some children travelled only with a nanny, therefore parch=0 for them.
  • 62.
    What we‘ll do 1.Clean Data ▪ Drop Name, ▪ Ticket ▪ Cabin 2. Convert Text Data to Numerical Data ▪ Sex ▪ Pclass ▪ Embarked
  • 63.
    Evaluation Meaning ▪ MeanAbsolute Error (MAE): The mean value of the absolute errors. (An error is the difference between the predicted value and the actual value). ▪ Root Mean Squared Error (RMSE): The square root of the average square of the forecast errors for the test dataset. ▪ Relative Absolute Error: The mean value of the absolute errors relative to the absolute difference between actual values and the average of all actual values. ▪ Relative Squared Error: The average of squared errors relative to the squared difference between actual values and the average of all actual values. ▪ Coefficient of Determination: This value, also known as the R square, is a statistical measure of how well a model fits the data.
  • 64.
    In other words SmallErrors = Good! High Coefficient of Determination = Good! The closer Coefficient of Determination is to 1 the better!
  • 65.
    Contact me incase of digital! TO WA. THE D IGITAL GRO W TH CO MPANY TOWA ▪ Instagram: www.instagram.com/towa.digital/ ▪ LinkedIn: www.linkedin.com/company/2099786/ ▪ Facebook: www.facebook.com/towa.digital ▪ Web: www.towa-digital.com PHILIPP ▪ Instagram: instagram.com/ploringhoven/ ▪ LinkedIn: linkedin.com/in/philipploringhoven/ ▪ Mail: philipp.loringhoven@towa.at