Nexxworks bootcamp ML6 (27/09/2017)
- 1. Nexxworks Bootcamp Ghent - 27/09/2017 karel.dumon@ml6.eu - @kareldumon
- 2. Nexxworks Bootcamp Ghent - 27/09/2017 Building machine learning* solutions with Google Cloud*
- 3. Nexxworks Bootcamp Ghent - 27/09/2017 Team of Data Engineers, Data Scientist & Machine Learning Engineers Closing the gap between lots of data - lacking insights Robust & agile ML solutions through scalable APIs Premium Partner of Google Cloud
- 4. Nexxworks Bootcamp Ghent - 27/09/2017 Building machine learning* solutions with Google Cloud*
- 5. Nexxworks Bootcamp Ghent - 27/09/2017 Rise of ML Cases The Future Get started!
- 6. Nexxworks Bootcamp Ghent - 27/09/2017
- 8. Confidential & ProprietaryGoogle Cloud Platform 8 2012 20132002 2004 2006 2008 2010 GFS MapReduce Bigtable Colossus Dremel Flume Megastore Spanner Millwheel Pub/Sub F1 2016 Dataflow TensorFlow Google’s 15+ years innovation in data
- 9. Confidential & ProprietaryGoogle Cloud Platform 9 Decrease the innovation gap 2012 20132002 2004 2006 2008 2010 GCS Dataproc Bigtable GCS BigQuery Dataflow Datastore Spanner Dataflow Pub/Sub F1 2016 Dataflow Cloud ML NoSQL
- 11. 11 Rapidly Accelerating Use of Deep Learning at Google Number of projects using some form of deep learning 2012 2013 2014 2015 1500 1000 500 0 Used across products:
- 12. Nexxworks Bootcamp Ghent - 27/09/2017 What’s going on? The cloud is very good at handling, storing and manipulating large volumes of data. What we really care about now is understanding the data.
- 13. Nexxworks Bootcamp Ghent - 27/09/2017 MACHINE LEARNING MACHINE LEARNING EVERYWHERE
- 14. Keys to Successful ML Large Datasets Good Models Lots of Computation
- 15. 15 The Deep Learning Revolution: ImageNet 2012 Deep Learning on Google Trends
- 16. Nexxworks Bootcamp Ghent - 27/09/2017 16
- 17. Nexxworks Bootcamp Ghent - 27/09/2017 17 The old, algorithmic approach “apple” “orange” “banana” IF (round) THEN IF (orange AND coarse) THEN “orange” ELSE IF (green AND smooth) THEN “apple” ELSE IF ... ... ELSE IF … “banana”
- 18. Nexxworks Bootcamp Ghent - 27/09/2017 18 Let the machine find the rules “apple” “orange” “banana” ?
- 19. Nexxworks Bootcamp Ghent - 27/09/2017 19
- 22. + ‘The Next Big Thing’ + Sentient AI in the next 10 years (‘The Singularity’) + Will put humans out of a job + Foolproof MACHINE LEARNING in popular culture
- 23. + Been around for 60 years now + ‘Sentient next year’, every year, for the last 60 years + AI winters: 1970, 1990, … ? + Not foolproof MACHINE LEARNING reality
- 24. 24 A person on a beach flying a kite. A person skiing down a snow covered slope. A group of giraffe standing next to each other.
- 25. 25 A woman riding a horse on a dirt road. An airplane is parked on the tarmac at an airport. A group of people standing on top of a beach.
- 26. 26
- 27. 27
- 28. 28
- 29. Nexxworks Bootcamp Ghent - 27/09/2017 29
- 30. Confidential & ProprietaryGoogle Cloud Platform 30
- 31. Nexxworks Bootcamp Ghent - 27/09/2017 Use cases to illustrate what we actually do.
- 32. Nexxworks Bootcamp Ghent - 27/09/2017 Get inspired, but let’s start simple.
- 33. Nexxworks Bootcamp Ghent - 27/09/2017 33 Logistics optimisation Predicting the time packages stay in customs + Better predict airplane cargo loads for an international courier + For each parcel, predict the time under inspection by customs (0 min = not picked up) + Better prediction → Better planning Better planning → Fuller cargos Avoid sudden overload
- 34. Nexxworks Bootcamp Ghent - 27/09/2017 34 Training the deep learning network (simplified) A parcel’s attributes (weight, size, text, colours, origin ...) The predicted package delay in customs >100k examples Full case: 3 days!
- 35. Nexxworks Bootcamp Ghent - 27/09/2017 35
- 36. Nexxworks Bootcamp Ghent - 27/09/2017 36 Predictive Maintenance: towards a highly dynamic Digital Twin Design Connect Predict theoretical model collect data create Digital Twin Testing generates theoretical life-time model - in-house testing - remaining life-time in ‘ideal conditions’ Machine Learning model predicts expected failure time Client receives personalized model by including custom environmental factors - usage, temperature, humidity… Cloud computing/storage STATIC MODEL HIGHLY DYNAMIC ML MODEL connected device Digital Twin ML
- 37. 37 Automated Customer service Building the smartest bots
- 38. 38 What did we do? + Shipped state-of-the-art system in under one month turnaround time + System is not an ‘Artificial Intelligence’ + System outperforms most solutions currently marketed as AI Why use Google Cloud Platform? + Access to massive computational resources to train our models + Able to dynamically adjust to changing number of requests + Leverage Google powered API’s The solution
- 39. 39 Architecture
- 40. 40
- 46. 46 The solution - semantic knowledge
- 48. 48 The problem
- 49. 49 The solution - context understanding
- 51. 51 The problem
- 52. 52 LSTM “How” “many” Word2Vec embedding “minutes” “left.”... Topic Long Short Term Memory (LSTM) network
- 53. 53 The solution - memory
- 54. 54 The problem
- 55. 55 Architecture
- 57. 57 Entity Extraction ● “I got charged 7 dollars for a Christmas Card that I never ordered. Plz remove this from my invoice, my telephone number is 472867486.” ● “My internet connection is no longer working. I have ADSL smart-50 and my address is NY city 52 A on 21st.” Entity Extraction Postprocessing 7 dollars $7.00 Christmas Card Christmas-card 472867486 +32472867486 ADSL smart-50 ADSL smart 50 NY city 52 A on 21st 52A, 21st avenue, New York
- 58. Nexxworks Bootcamp Ghent - 27/09/2017 58 Other ideas? smart cities? (garbage detection, road conditions, mobile camera’s for anomaly detection...)
- 59. Nexxworks Bootcamp Ghent - 27/09/2017 59
- 60. Medical imaging diagnostics with deep learning Input Output
- 61. MEDICAL IMAGE CLASSIFICATION TECHNICAL DETAILS - DATA Camelyon16: ISBI challenge on cancer metastasis detection in lymph node Training & Evaluation: 110 tumor slides, 160 normal slides, 130 evaluation slides Task 1: whole-slide level prediction binary classification problem Task 2: find metastasis location segmentation problem
- 63. MEDICAL IMAGE CLASSIFICATION TECHNICAL DETAILS - CHALLENGE BREAKDOWN Not just another standard image classifier… Network architecture Training set construction Computing environment Post-processing (classification and segmentation)
- 64. MEDICAL IMAGE CLASSIFICATION TECHNICAL DETAILS - TRAINING Patches are used to train a deep learning model - started with simple CIFAR10-model - ended up retraining Inception-v3 to experiment with synchronous updates across multiple GPUs Metastasis patches Normal patches Deep Learning model e.g. Inception-v3
- 65. MEDICAL IMAGE CLASSIFICATION TECHNICAL DETAILS - MULTIPLE GPUs Multiple GPUs were used to calculate model updates (based on TensorFlow tutorial) - individual model replica on each GPU - all GPUs process a batch of data and send update of model parameters to CPU, which performs a synchronized update Approach was extended to Cloud ML Engine GPUs during an external ackathon to build a dermatology POC in 1 day (by starting from a.o. pretrained VGG16)
- 68. Nexxworks Bootcamp Ghent - 27/09/2017 68
- 70. 70 Supervised: need large amounts of annotated training data Static inference machines Bad transfer learning capabilities to new tasks
- 71. 71 Policy
- 72. 72 Third person imitation learning
- 73. 73
- 74. 74
- 75. Nexxworks Bootcamp Ghent - 27/09/2017 75
- 76. 76
- 77. 77
- 78. 78 3:53
- 80. #cloudconf2016
- 81. Nexxworks Bootcamp Ghent - 27/09/2017 81 Use reinforcement learning to optimize energy usage Reinforcement learning allows for automatic parameter optimization: - e.g. energy optimization: let AI-agent control settings to optimize energy consumption in data center cooling - See also Google Deepmind
- 82. Ultimate Machine Learning with Google Cloud 82
- 83. Nexxworks Bootcamp Ghent - 27/09/2017 83 Fleet optimization w/ online reinforcement learning
- 85. Nexxworks Bootcamp Ghent - 27/09/2017 85 Advantages that AlphaGO can leverage 1. Fully deterministic: no noise in the game 2. Fully observed: each player has complete information and there are no hidden variables. (unlike Poker for example) 3. Discrete action space 4. Each game is relatively short (approximately 200 actions) 5. Target function is clear (win/lose) & fast to evaluate 6. Huge datasets of human gameplay are available to bootstrap the learning, so AlphaGo doesn’t have to start from scratch
- 86. Nexxworks Bootcamp Ghent - 27/09/2017 86
- 87. Nexxworks Bootcamp Ghent - 27/09/2017 87 “If a top 1% CEO today understands how software applications get built and how that changes the way she/he manages, a top 1% CEO in the near future will understand how models get built and how that changes the way she/he builds her/his organization.” James Cham, Bloomberg Data
- 88. Nexxworks Bootcamp Ghent - 27/09/2017 88 1. Educate yourself 2. AI strategy 3. TEST! (Get your hands dirty)
- 89. Nexxworks Bootcamp Ghent - 27/09/2017 How?
- 90. Plenty of problems to solve...
- 91. Plenty of solutions around...
- 92. Tensorflow Training THANKS! What my friends think I do What other computer scientists think I do What society thinks I do What mathematicians think I do What I think I do What I actually do
- 93. Nexxworks Bootcamp Ghent - 27/09/2017 karel.dumon@ml6.eu - @kareldumon