Machine Learning What, how, why?

Transcription

1 Machine Learning What, how, why? Rémi Emonet Web En Vert

2 $ whoami

3 $ whoami Software Engineer Researcher: machine learning, computer vision Teacher: web technologies, computing literacy Geek: deck.js slides, isochrones, You are shrewd, skeptical and restrained. You are independent: you have a strong desire to have time to yourself. You are calm-seeking: you prefer activities that are quiet, calm, and safe. And you are philosophical: you are open to and intrigued by new ideas and love to explore them. Experiences that give a sense of prestige hold some appeal to you. You are relatively unconcerned with both tradition and taking pleasure in life. You care more about making your own path than following what others have done. And you prefer activities with a purpose greater than just personal enjoyment.

4 What is Machine Learning? 4 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

5 Machine Learning Basic Principle 5 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Given a dataset {y, x,, x } Optimize the likelihood function L = n(w, t, d)log p(w, t z)p(z, t d) Or using a sparse regularization L λ n i i1 ip i=1 a d w t a z sparse KL(U p(ts z, d)) d z t s r s By using a Gibbs Sampler ji ji N obs (W ji, rt ji, z ji) + η(w ji, rt ji) p(w ji, atji o ji = o, O ) = ji N (w, rt, z ) + η(w, rt ) w,rt ( obs ji )

6 Machine Learning in the Wild 6 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

7 Which One of These Services Uses Machine Learning? 7 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

8 Machine Learning in Future Tech? 8 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

9 What is Machine Learning? an example motivation 9 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

10 Challenge: Which Iris Species?

11 Feature Extraction 11 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Sepal length: 5.1 Sepal width: 2.5 Petal length: 4.2 Petal width: 1.0 Expected Label: Iris Setosa

12 Analysis and Program Writing

13 IFTTT...

14 Predictive Machine Learning 14 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Instead of writing a program that solves a task, We 1. collect labeled data: input/output pairs 2. automatically generate a program that computes an output for each new input 3. profit! The machine learns to generalize from a limited number of examples, like humans do.

15 Different Types of Tasks 15 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Supervised learning: some labels are known classification: find the label of an example regression: find the target value Unsupervised learning: no labels clustering: group things together pattern mining: find recurrent events anomaly detection: find outliers

16 The Principle of Overfitting 16 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

17 A Lot of Different Methods 17 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Also called models linear regression, logistic regression, SVM, kernel SVM, neural networks, k-means clustering, collaborative filtering, bayesian networks, expectation maximization, belief propagation, multiple kernel learning, metric learning, transfer learning, decision trees, gaussian processes, random forests, boosting,... For different contexts different tasks different nature of data different suppositions on the data different amount of data

18 Different Ways to Start 18 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Use a product that uses ML e.g. adwords, ibm bluemix, Use a prediction API send your data to the service get API to process new inputs e.g., google pred. API, prediction.io,... Dive into machine learning

19 Into Machine Learning 19 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Using libraries libraries exist in most languages most models already implemented test different methods with different parameters Learning machine learning many online courses get deeper understanding

20 Does Machine Learning Actually Matter? 20 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

21 Example: The Netflix Challenge 21 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

22 22 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? FAIR

23 Example: Facebook AI Research 23 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Director: Yann LeCun Scientific Leads Léon Bouttou Rob Fergus Florent Perronnin fr rest

24 Data, Data, Data 24 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

25 Data is Machine Learning's Fuel 25 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

26 data === power

27 Getting Data?

28 Getting Data? 28 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why? Collect from your services/applications Do it yourself Pay some people you know Use crowd-sourcing, e.g., Amazon Mechanical Turk (MTurk) Find existing datasets (open data, etc) Work for/with a data rich company Create your intermediation business

29 What Can It Do For Me 29 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

30 Search Google Search, Bing, etc 30 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

31 Advertising AdWords, etc 31 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

32 Recommendations Netflix, Amazon, Youtube, app Stores, etc 32 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

33 Text Translation 33 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

34 Optical Character Recognition (postcodes, checks, book scans, etc) 34 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

35 Visual Recognition (objects, plants, animals, etc) 35 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

36 Face Detection Smile Detection (embedded in Cameras) 36 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

37 Face Identification (Picasa, Facebook, etc) 37 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

38 Kinect Controller 38 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

39 Self Driving Cars 39 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

40 Voice Recognition and Synthesis (GoogleNow, Siri, Cortana) 40 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

41 Sound Recognition (birds, underwater sounds, safety, etc) 41 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

42 Fraud Detection (Banking, Websites, etc) 42 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

43 Automated Trading 43 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

44 Customer/Person Profiling BlueMix Watson, etc 44 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

45 Adaptive Websites (automated A/B testing) 45 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

46 The Big Data Hype 46 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

47 and much more...

48 Where Will it Stop? 48 / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

49 Singularity?

50 Thanks! Questions? web/ Recommended Links: comprehensive introduction to ML models scikit learn (python) / 50 Rémi Emonet (@remiemonet) Machine Learning What, how, why?

51 tatadbb

52 -Porsupah-

53 ali eminov

54 efilpera

55 jannekestaaks

56 Asa-moya

57 francoisjouffroy

58 JD Hancock