Download to read offline
Uploaded byBIWUG
Why you shouldn't probably care about Machine Learning
1) Machine learning involves using algorithms to automatically find patterns in data. It is used for tasks like classification, regression, clustering, and recommendation. 2) Machine learning can help with prediction, performing time-consuming tasks, and aiding human experts. However, there are also limitations, including that correlation does not imply causation, as well as ethical and privacy concerns. 3) Popular machine learning platforms and libraries include Microsoft Azure ML, Amazon SageMaker, TensorFlow, CNTK, and ONNX for model interchange. Pre-trained models and APIs are available through services like Microsoft Cognitive Services and Amazon Rekognition.
More Related Content
Similar to Why you shouldn't probably care about Machine Learning
Why you shouldn't probably care about Machine Learning
- 1. Why You Should(Not) Care About Machine Learning Edin Kapić | @ekapic SharePoint Saturday Belgium 2018 #SPSBE
- 2.
- 3. Patrick Tisseghem Founding fatherof BIWUG & true SharePoint master 17/10/1968 - 3/9/2008 Special Tribute Edition #SPSBE
- 4. Edin Kapić • SharePointMaster of Craft @ isolutions Barcelona • Cofounder and president of SUG.CAT • MVP for Office Apps and Services • Organizer for SPS Barcelona (October 27th) @ekapic
- 5.
- 8.
- 9.
- 10.
- 11. The Roadmap What WhyHow But
- 12. What Why HowBut
- 13.
- 14. Machine Learning • Automaticallylooking for patterns in the data
- 15. Machine Learning • Automaticallylooking for patterns in the data
- 16.
- 17. Machine Learning • Automaticallylooking for patterns in the data Alice Foo Baz Bar Bob Doo
- 18. Machine Learning vsProgramming Programming Machine Learning RESULTS RULES
- 19. Types of MachineLearning All examples are from https://docs.microsoft.com/en-us/azure/machine-learning/studio/algorithm-choice Clustering (what groups of customers I have?) Recommendation (what song should I listen to?) Regression (how many sales am I making?) Classification (what is in this picture?) Anomaly detection (is this account activity normal?) SUPERVISED UNSUPERVISED
- 20. Regression • Tries tofit observed values into a function curve • Predicts a value (price, size, number of customers…)
- 21. Classification • Splits thedecision space into discrete categories and predicts them
- 22. Anomaly Detection • Specialkind of classification that highlights outlier values regarding training data
- 23. Clustering • Group observeddata into “chunks”
- 24. Recommendation • Associate inputand output values to predict by similarity • Mix of classification and reinforcement graphs
- 25. Confused by now? •Azure ML Algorithm Cheat Sheet • Azure ML Basics Infographic with Examples bcned.in/AzureMLCheatsheet bcned.in/AzureMLInfographic
- 26. What Why HowBut
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.
- 33. What Why HowBut
- 35.
- 36. Self-service Machine LearningModels • Pre-trained models • API as a service
- 37.
- 38. SHAREPOINT PICTURE LIBRARY KEYWORDEXTRACTION
- 39. Microsoft ML Studio AmazonSageMaker
- 40. Azure ML Studio/ Azure ML Experimentation Service • Platform for ML models and experiments
- 41.
- 42.
- 44.
- 45. CNTK: Cognitive Toolkit •Open-source deep learning toolkit by Microsoft https://www.microsoft.com/en-us/cognitive-toolkit/
- 46. Windows ML • Evaluatespre-trained models on Windows platform https://docs.microsoft.com/en-us/windows/uwp/machine-learning/
- 47. TensorFlow • Open-source machinelearning framework https://www.tensorflow.org/
- 48. ONNX: Model InterchangeFormat • Open Neural Network Exchange Format • Developed by Microsoft, Facebook and Amazon • Gallery of models at https://github.com/onnx/models
- 49.
- 50. What Why HowBut
- 51.
- 52. Limitations of MachineLearning
- 53.
- 54.
- 55.
- 56.
- 57. What • Pattern recognition •Supervised • Unsupervised Why • Prediction • Realtime • Time-consuming How • Services • Platforms • Libraries But • Ethics • Privacy • Limitations Machine Learning
- 58.
- 59. SharePoint Saturday Belgium2018 #SPSBE
Editor's Notes
- #28 Predict some future behaviour Tell me what my factory workload is going to be next month.
- #29 Quickly do time-consuming classification Here are 100.000 animal photos. Please classify them into dogs, cats and parrots.
- #31 React to changes in real time I want the turbine blades on the airplane changed as soon as they show signs of wear.
- #32 Perform operations that are non-scriptable Here is the driving wheel, Tesla. Drive according to the traffic.
- #33 Metadata extraction Autoclassification Summarizing Recommendations Relating data
- #52 https://transcranial.github.io/keras-js/#/inception-v3 Normal cat: https://i.imgur.com/ZJR3S7F.jpg Hacked cat: https://i.imgur.com/vTKgnFR.png
- #53 Correlation is not causation
- #54 Uncertainty is always there
- #56 What biases are there in the source data? What information is used as an estimator? Is machine doing a human-relevant decision?
- #57 Big Data and user-generated data Ownership of data Transparency of models
- #58 Not a silver bullet Can be hacked