Brief introduction to Machine Learning

36
Brief Introduction to Machine Learning Michael Krech, @Parsec

Transcript of Brief introduction to Machine Learning

Page 1: Brief introduction to Machine Learning

Brief Introduction to Machine Learning

Michael Krech, @Parsec

Page 2: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

OUR MISSION Chapter 1: History - Present Chapter 2: Overview Chapter 3: Computer can .. Chapter 4: Why now? Chapter 5: Future

Page 3: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

CHAPTER

1

Page 4: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

HISTORY

Page 5: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

PRESENT

Page 6: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

PRESENT

Page 7: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

PRESENT

What are my recommended movies ?

Page 8: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

PRESENT

What are my recommended friends?

Page 9: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

PRESENT

https://www.youtube.com/watch?v=YgYSv2KSyWgIn initial tests run during 2006 by David Ferrucci, the senior manager of IBM's Semantic Analysis and Integration department, Watson was given 500 clues from past Jeopardy! programs. While the best real-life competitors buzzed in half the time and responded correctly to as many as 95% of clues, Watson's first pass could get only about 15% correct. During 2007, the IBM team was given three to five years and a staff of 15 people to solve the problems. By 2008, the developers had advanced Watson such that it could compete with Jeopardy! Champions. By February 2010, Watson could beat human Jeopardy! contestants on a regular basis

Page 10: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

PRESENT

In June 2015, the team announced that their vehicles have now driven over 1 million miles, stating that this was "the equivalent of 75 years of typical U.S. adult driving", and that in the process they had encountered 200,000 stop signs, 600,000 traffic lights, and 180 million other vehicles. Google also announced its prototype vehicles were being road tested in Mountain View, California.During testing, the prototypes' speed cannot exceed 25 mph and will have safety drivers aboard the entire time.

Page 11: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

CHAPTER

2

Page 12: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

OVERVIEW

Page 13: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

2,5 KINDS OF MACHINE LEARNING ALGORITHMS

 Supervised Learning

Unsupervised Learning

 Reinforcement learning

Page 14: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

SUPERVISED-LEARNING

Supervised learning is the machine learning task of inferring a function from labeled training data. The training data consist of a set of training examples. In supervised learning, each example is a pair consisting of an input object (typically a vector) and a desired output value (also called the supervisory signal). A supervised learning algorithm analyzes the training data and produces an inferred function, which can be used for mapping new examples. An optimal scenario will allow for the algorithm to correctly determine the class labels for unseen instances. This requires the learning algorithm to generalize from the training data to unseen situations in a "reasonable" way

Page 15: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

UNSUPERVISED LEARNING

In machine learning, the problem of unsupervised learning is that of trying to find hidden structure in unlabeled data. Since the examples given to the learner are unlabeled, there is no error or reward signal to evaluate a potential solution. This distinguishes unsupervised learning from supervised learning and reinforcement learning. Unsupervised learning is closely related to the problem of density estimation in statistics. However unsupervised learning also encompasses many other techniques that seek to summarize and explain key features of the data. Many methods employed in unsupervised learning are based on data mining methods used to preprocess[citation needed] data. e.g.: Anomaly detection …

Page 16: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

REINFORMENT-LEARNING, (SEMI-SUPERVISED)

Reinforcement learning is an area of machine learning inspired by behaviorist psychology, concerned with how software agents ought to take actions in an environment so as to maximize some notion of cumulative reward. This is the most common form, humans learn.

Page 17: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

REINFORMENT-LEARNING, (SEMI-SUPERVISED)

http://cs.stanford.edu/people/karpathy/convnetjs/demo/rldemo.html

Page 18: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

WORKSPACES   Reporting

  Search

  Exploration

  Prediction   Classification

Page 19: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

CHAPTER

3

Page 20: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

COMPUTER CAN ….

?

Page 21: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

LISTEN, SPEAK

https://www.youtube.com/watch?v=Nu-nlQqFCKg

Speech Recognition Breakthrough for the Spoken, Translated Word

Page 22: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

SEE

http://benchmark.ini.rub.de/?section=gtsrb&subsection=resultsComputer better as humans

Page 23: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

RECOGNITION

Google can identify and transcribe all the views it has of street numbers in France in less than an hour, thanks to a neural network that’s just as good as human operators. http://www.technologyreview.com/view/523326/how-google-cracked-house-number-identification-in-street-view/ Wie lange und wie viele Menschen hätte dies benötigt?

Page 24: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

RECOGNITON AND CLUSTER

A selection of evaluation results, grouped by human rating. http://googleresearch.blogspot.in/2014/11/a-picture-is-worth-thousand-coherent.html

Page 25: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

SING

https://www.youtube.com/watch?v=dKUDHPw15m0

Page 26: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

READ

OCR http://de.wikipedia.org/wiki/Texterkennung

Page 27: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

UNDERSTAND

http://nlp.stanford.edu/sentiment/ http://nlp.stanford.edu/~socherr/EMNLP2013_RNTN.pdf

Page 28: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

COMPUTER CAN …

Read & Write

Listen, Speak Singen

See, Recognition Understand

Page 29: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

CHAPTER

4

Page 30: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

WHY THIS WORKS BETTER NOW

CLOUD / CLUSTER / CORES TOOLS Technologies (Forbes)

GPU, SaaS, PaaS, IaaS, Amazon Cloud Computing, Google Cloud Computing ….

Hadoop, Spark, Elasticsearch, MongoDB, Neo4j, Mahout, Python, R

Deep Learning, Clustering, Anomaly Detection, ConvNet, Reinforcement, Statistics

Page 31: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

BIOLOGIC PYRAMID

@mike_schatz

Page 32: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

CHAPTER

5

Page 33: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

Moores Law, complexity of IC’s double’s in 12-24 Month

Page 34: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

T-1000

http://en.wikipedia.org/wiki/T-1000

Page 35: Brief introduction to Machine Learning

@Parsec A brief introduction to Machine Learning

TURING TEST Back in 2002 Kurzweil (a scientist renowned for his accurate tech predictions), bet Mitch Kapor (founder of Lotus Development Corp., inventor of spreadsheet software) $20,000 that a computer would pass the Turing Test by 2029. He predicts a Singularity for the yeaar 2045…. Er prognostiziert für das Jahr 2045 eine exponentielle Zunahme der informationstechnologischen Entwicklung: Eine Singularität, die eine künstliche Intelligenz ermöglicht, mit welcher die Menschheit Unsterblichkeit erlangen kann. https://en.wikipedia.org/wiki/Predictions_made_by_Ray_Kurzweil#2045:_The_Singularity (Wikipedia)

Alan Turing Ray Kurzweil Raymond "Ray" Kurzweil is an American author, computer scientist, inventor, futurist, and is a director of engineering at Google. Aside from futurology, he is involved in fields such as optical character recognition (OCR), text-to-speech synthesis, speech recognition technology, and electronic keyboard instruments. He has written books on health, artificial intelligence (AI), transhumanism, the technological singularity, and futurism. Kurzweil is a public advocate for the futurist and transhumanist movements, as has been displayed in his vast collection of public talks, wherein he has shared his primarily optimistic outlooks on life extension technologies and the future of nanotechnology, robotics, and biotechnology.

The Turing test was introduced by Alan Turing in 1950. The Turing test is a test of a machine's ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human.

Page 36: Brief introduction to Machine Learning

THANK YOU