SlideShare ist ein Scribd-Unternehmen logo
1 von 46
Downloaden Sie, um offline zu lesen
A Gentle Introduction to
Machine Learning
Applications, Algorithms and Scalable Deployment
Eindhoven Developers Meetup
2015-04-29
Machine Learning:
everybody talks about it
Today let’s try to demystify machine learning
Focus less on glorifying the machine learning and
more on the technical details
Demystify and Glorify
The slides and the talk solely represent the speakers’
personal views
 In its general form, Machine Learning means
teaching to a computational machine the way of
solving a problem by giving examples
 Machine learning algorithms then automatically
infer rules to associate inputs to the correspondent
outputs
Machine Learning
Input Program
Output
General
programming
Input Output
Learned
Program
Machine
Learning
New Input
Inferred
Output
General Programming vs Machine Learning
TRAIN TEST
Lion
PandaGiraffe Tiger
Elephant
Input – Output Examples
Tiger
Real Example: Characters Recognition
 Letters have predefined shapes: we can measure and quantify
their relative proportions
 In real world, rules are hard to formulate
 Even if we might have a comprehensive set of rules, it would
be hard to scale and make them robust over different writing
styles
Handwritten Character Recognition in
the Real World
Input Output
Learned
Program
Machine
Learning
Machine Learning: Inputs and Outputs
 We will focus on Supervised Learning i.e.,
learning associations between Inputs and
Outputs
There exist other types of Learning
Unsupervised: learning association
between inputs only
Semi-supervised: learning from few
outputs and a large amount of inputs
Reinforcement: giving rewards for good
associations
 Inputs are generally represented by features
 characteristic and meaningful measures computed on raw-
data
 they provide domain information from human to machine
 they make the learning process easier
Machine Learning: Inputs
 Signal processing: from sound we can extract frequency, maximum
amplitude, power spectrum, etc ...
 Probability and Statistics: from text we can compute probability
distributions of usual words, words co-occurrences, etc …
Nevertheless, many inputs are already structured in feature (numeric)
format
For example: The customer information like income range, payment dues
etc. are the possible features for credit risk profiling.
What Features are?
Classification:
Predict a categorical output
Input Output
Feature1 Feature2 Class
3.2 4.1 1
1.6 1.7 0
1.9 2.8 2
3.2 45.0 0
1.4 11.5 2
2.7 22.0 1
Input Output
Feature1 Feature2
3.2 4.1 1.26
1.6 1.7 0.82
1.9 2.8 2.94
3.2 45.0 0.33
1.4 11.5 22.5
2.7 22.0 12.5
Machine Learning: Outputs
Regression:
Predict a numerical output
How Learning looks like in a Features
space
Using Features we can use Linear Algebra as main tool for the learning process
Signal Processing
Probability and Statistics
Linear Algebra
Inputs Outputs
New Input Predicted Output
Machine Learning Model
…the final ingredients
Let’s go more in depth ...
Separate 2 classes: Linear Classifier
Class 1
Class 2
….but this is also good
Class 1
Class 2
There exist infinite good separations
Class 1
Class 2
Linear SVM
Support
Vectors
Support Vector Machines (SVMs)
Class 1
Class 2
Some problems are not separable with a
line: XOR problem
Trick: let’s add a fake dimension
Non-linear SVMs: The Kernel Trick
Φ: x → φ(x)
The features space can always be mapped to some higher-dimensional
space where the training set is linearly separable
Generalization vs Overfitting
Another way to separate..
Between the infinite lines …
…let’s pick up one slightly good
…let’s give “bad” rewards to wrong
points
…and do it again
… and again
…until we get a complete separation
Face Detection with AdaBoost
Original
Training data
....D1
D2 Dt-1 Dt
D
Step 1:
Create Multiple
Data Sets
C1 C2 Ct -1 Ct
Step 2:
Build Multiple
Classifiers
C*
Step 3:
Combine
Classifiers
Ensemble Methods: General Idea
• Construct a set of classifiers from the training data
• Predict class label of previously unseen records by aggregating
predictions made by multiple classifiers
• Suppose there are 25 base classifiers
• Each classifier has error rate,  = 0.35
• Assume classifiers are independent
• Probability that the ensemble classifier makes a wrong
prediction:
25
i
æ
è
ç
ö
ø
÷ei
(1-e)25-i
= 0.06
i=1
25
å
Why does it work
This is the only formula in this presentation.
Enjoy it!!
X-box Kinect: Random Forest Ensemble
Deep Learning: the new Frontier
Brief history of Neural Nets: born in 70s, developed in 80s, dead
in 90s, forgotten in 2000, state of the art in 2010
How do we deploy these algos ?
 Real world problems: Unpredictable size
 The volume and velocity of the input data can be very big
 For example:-
One wearable device streams out more than a million
message per day
Scalable Deployment
 Distributed and Parallel Machine learning
Dealing with the Volume
http://dme.rwth-aachen.de/de/research/projects/mapreduce
Distributed Machine Learning:
Map-Reduce
http://opensource.com/life/14/8/intro-apache-hadoop-big-data
Map-Reduce Implementation: Hadoop
 Machine Learning Libraries available
 Distributed computation geared for machine learning
(Iterative computation)
Resources for Distributed Processing
 Cloud
 Amazon
 Elastic Compute Cloud(EC2)
 Elastic Map-Reduce (EMR)
 Azure
 HDInsight
 Your own server farm
How to Deploy Scalable Solutions
Ingesting the data
Amazon: Kinesis
 Microsoft: Event-hub
Scalable Ingestion
Data Ingestion
Storage
Consumption
Connecting the components

Weitere ähnliche Inhalte

Ähnlich wie Machine Learning: Demystify and Scale

MachinaFiesta: A Vision into Machine Learning 🚀
MachinaFiesta: A Vision into Machine Learning 🚀MachinaFiesta: A Vision into Machine Learning 🚀
MachinaFiesta: A Vision into Machine Learning 🚀GDSCNiT
 
Big Data & Machine Learning - TDC2013 Sao Paulo
Big Data & Machine Learning - TDC2013 Sao PauloBig Data & Machine Learning - TDC2013 Sao Paulo
Big Data & Machine Learning - TDC2013 Sao PauloOCTO Technology
 
Big Data & Machine Learning - TDC2013 São Paulo - 12/0713
Big Data & Machine Learning - TDC2013 São Paulo - 12/0713Big Data & Machine Learning - TDC2013 São Paulo - 12/0713
Big Data & Machine Learning - TDC2013 São Paulo - 12/0713Mathieu DESPRIEE
 
How to implement artificial intelligence solutions
How to implement artificial intelligence solutionsHow to implement artificial intelligence solutions
How to implement artificial intelligence solutionsCarlos Toxtli
 
Lessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at NetflixLessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at NetflixJustin Basilico
 
Strata London - Deep Learning 05-2015
Strata London - Deep Learning 05-2015Strata London - Deep Learning 05-2015
Strata London - Deep Learning 05-2015Turi, Inc.
 
Machine Learning.pptx
Machine Learning.pptxMachine Learning.pptx
Machine Learning.pptxchadhar227
 
AI and ML Skills for the Testing World Tutorial
AI and ML Skills for the Testing World TutorialAI and ML Skills for the Testing World Tutorial
AI and ML Skills for the Testing World TutorialTariq King
 
“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...
“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...
“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...Edge AI and Vision Alliance
 
2020 04 04 NetCoreConf - Machine Learning.Net
2020 04 04 NetCoreConf - Machine Learning.Net2020 04 04 NetCoreConf - Machine Learning.Net
2020 04 04 NetCoreConf - Machine Learning.NetBruno Capuano
 
Software Modeling and Artificial Intelligence: friends or foes?
Software Modeling and Artificial Intelligence: friends or foes?Software Modeling and Artificial Intelligence: friends or foes?
Software Modeling and Artificial Intelligence: friends or foes?Jordi Cabot
 
Intelligent Ruby + Machine Learning
Intelligent Ruby + Machine LearningIntelligent Ruby + Machine Learning
Intelligent Ruby + Machine LearningIlya Grigorik
 
Artificial Intelligence - Anna Uni -v1.pdf
Artificial Intelligence - Anna Uni -v1.pdfArtificial Intelligence - Anna Uni -v1.pdf
Artificial Intelligence - Anna Uni -v1.pdfJayanti Prasad Ph.D.
 
Start machine learning in 5 simple steps
Start machine learning in 5 simple stepsStart machine learning in 5 simple steps
Start machine learning in 5 simple stepsRenjith M P
 
Keynote at IWLS 2017
Keynote at IWLS 2017Keynote at IWLS 2017
Keynote at IWLS 2017Manish Pandey
 

Ähnlich wie Machine Learning: Demystify and Scale (20)

MachinaFiesta: A Vision into Machine Learning 🚀
MachinaFiesta: A Vision into Machine Learning 🚀MachinaFiesta: A Vision into Machine Learning 🚀
MachinaFiesta: A Vision into Machine Learning 🚀
 
Big Data & Machine Learning - TDC2013 Sao Paulo
Big Data & Machine Learning - TDC2013 Sao PauloBig Data & Machine Learning - TDC2013 Sao Paulo
Big Data & Machine Learning - TDC2013 Sao Paulo
 
Big Data & Machine Learning - TDC2013 São Paulo - 12/0713
Big Data & Machine Learning - TDC2013 São Paulo - 12/0713Big Data & Machine Learning - TDC2013 São Paulo - 12/0713
Big Data & Machine Learning - TDC2013 São Paulo - 12/0713
 
How to implement artificial intelligence solutions
How to implement artificial intelligence solutionsHow to implement artificial intelligence solutions
How to implement artificial intelligence solutions
 
Raising the Bar
Raising the BarRaising the Bar
Raising the Bar
 
Lessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at NetflixLessons Learned from Building Machine Learning Software at Netflix
Lessons Learned from Building Machine Learning Software at Netflix
 
Strata London - Deep Learning 05-2015
Strata London - Deep Learning 05-2015Strata London - Deep Learning 05-2015
Strata London - Deep Learning 05-2015
 
Machine Learning.pptx
Machine Learning.pptxMachine Learning.pptx
Machine Learning.pptx
 
AI and ML Skills for the Testing World Tutorial
AI and ML Skills for the Testing World TutorialAI and ML Skills for the Testing World Tutorial
AI and ML Skills for the Testing World Tutorial
 
“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...
“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...
“Introducing Machine Learning and How to Teach Machines to See,” a Presentati...
 
LR2. Summary Day 2
LR2. Summary Day 2LR2. Summary Day 2
LR2. Summary Day 2
 
Lecture-6-7.pptx
Lecture-6-7.pptxLecture-6-7.pptx
Lecture-6-7.pptx
 
Introduction to r
Introduction to rIntroduction to r
Introduction to r
 
2020 04 04 NetCoreConf - Machine Learning.Net
2020 04 04 NetCoreConf - Machine Learning.Net2020 04 04 NetCoreConf - Machine Learning.Net
2020 04 04 NetCoreConf - Machine Learning.Net
 
Software Modeling and Artificial Intelligence: friends or foes?
Software Modeling and Artificial Intelligence: friends or foes?Software Modeling and Artificial Intelligence: friends or foes?
Software Modeling and Artificial Intelligence: friends or foes?
 
Intelligent Ruby + Machine Learning
Intelligent Ruby + Machine LearningIntelligent Ruby + Machine Learning
Intelligent Ruby + Machine Learning
 
3 algorithm-and-flowchart
3 algorithm-and-flowchart3 algorithm-and-flowchart
3 algorithm-and-flowchart
 
Artificial Intelligence - Anna Uni -v1.pdf
Artificial Intelligence - Anna Uni -v1.pdfArtificial Intelligence - Anna Uni -v1.pdf
Artificial Intelligence - Anna Uni -v1.pdf
 
Start machine learning in 5 simple steps
Start machine learning in 5 simple stepsStart machine learning in 5 simple steps
Start machine learning in 5 simple steps
 
Keynote at IWLS 2017
Keynote at IWLS 2017Keynote at IWLS 2017
Keynote at IWLS 2017
 

Machine Learning: Demystify and Scale