Daniel Sarbe
Development Manager, BigData and Machine Translation
SDL Research
Twitter: @danielsarbe
Machine Learning in the age
of Big Data
Agenda
• Machine Learning overview
• Big Data overview
• Why Machine Learning gain more importance in BigData age?
• Demo
• Q&A
What is Machine Learning?
• Machine learning
• field of study that gives computers the ability to learn without being
explicitly programmed
• Arthur Samuel in 1959 - chess program
• make intelligent decisions and predictions
based on your data
• ML algorithms
• essentially, just probabilities and statistics
• are math, they are not magic
• can be done on paper but it takes too much
• machines can do it really well
Machine Learning Styles
1. Supervised Learning - Learning from labeled data
• Regression
• Predicting house price
• Weight based on height
• Classification
• Spam filtering
• OCR
2. Unsupervised learning - Learning from unlabeled data
• Clustering
• Recommendation systems (e.g. Amazon/Netflix)
• Grouping related web news (e.g. Google News)
Traditional vs ML
Typical ML process
“It's not who has the best algorithm that wins. It's who has the
most data.”
Andrew Ng
What is BigData?
What is BigData?
ML and BigData
• ML works better on Big data
• We don’t need lots of things to learn, if we have a huge data
• What roles plays human in this?
Batch vs Online Learning Algorithms
• Batch
• has access to the entire training data set
• Online
• algorithm receives feedback about each prediction
• feedback is used to improve the accuracy on subsequent predictions.
• has to make predictions continuously (ad-hoc learning)
Language Learning Demo
• What is the problem that we are trying to solve?
Machine Learning in the age of Big Data
Machine Learning in the age of Big Data

Machine Learning in the age of Big Data

  • 1.
    Daniel Sarbe Development Manager,BigData and Machine Translation SDL Research Twitter: @danielsarbe Machine Learning in the age of Big Data
  • 2.
    Agenda • Machine Learningoverview • Big Data overview • Why Machine Learning gain more importance in BigData age? • Demo • Q&A
  • 3.
    What is MachineLearning? • Machine learning • field of study that gives computers the ability to learn without being explicitly programmed • Arthur Samuel in 1959 - chess program • make intelligent decisions and predictions based on your data • ML algorithms • essentially, just probabilities and statistics • are math, they are not magic • can be done on paper but it takes too much • machines can do it really well
  • 4.
    Machine Learning Styles 1.Supervised Learning - Learning from labeled data • Regression • Predicting house price • Weight based on height • Classification • Spam filtering • OCR 2. Unsupervised learning - Learning from unlabeled data • Clustering • Recommendation systems (e.g. Amazon/Netflix) • Grouping related web news (e.g. Google News)
  • 5.
  • 6.
  • 7.
    “It's not whohas the best algorithm that wins. It's who has the most data.” Andrew Ng
  • 8.
  • 9.
  • 10.
    ML and BigData •ML works better on Big data • We don’t need lots of things to learn, if we have a huge data • What roles plays human in this?
  • 11.
    Batch vs OnlineLearning Algorithms • Batch • has access to the entire training data set • Online • algorithm receives feedback about each prediction • feedback is used to improve the accuracy on subsequent predictions. • has to make predictions continuously (ad-hoc learning)
  • 12.
    Language Learning Demo •What is the problem that we are trying to solve?