Machine Learning 44444 ohne Hype 4444 Philipp Krenn 44 @xeraa
Slide 1
Slide 2
44444 Developer !
Slide 3
Slide 4
Machine Learning is going viral...
Slide 5
Slide 6
Slide 7
❝ Using #DeepLearning when all you needed was a few if statements. #MachineLearning #DataScience ❞ —https://twitter.com/randal_olson/status/927157485240311808
Slide 8
Slide 9
Slide 10
Slide 11
Agenda Machine Learning Domain Dataset
Slide 12
Machine Learning
Slide 13
Artificial Intelligence Machine Learning Deep Learning !
Slide 14
https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
Slide 15
General AI Human characteristics
Slide 16
AI Winter
Slide 17
Narrow AI Specific tasks
Slide 18
Slide 19
Facebook alt="Image may contain: ocean, sky, bridge, cloud, outdoor, water and nature"
Slide 20
Slide 21
Slide 22
PS: A lot of Chatbots are not AI
Slide 23
Slide 24
Slide 25
❝ Alice: I love stateless protocols! Bob: " ere has to be something bad about them. Alice: Bad about what? ❞ — https://twitter.com/znjp/status/933405548678021120
Slide 26
Machine Learning Algorithms parse data → learn from it → make a determination or prediction "Trained" machine
Slide 27
❝ Learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E. ❞
Slide 28
❝ "Machine Learning is an emerging tech!" Logistic regression 1958 Hidden Markov Model 1960 Support Vector Machine 1963 k-nearest neighbors 1967 Artificial Neural Networks 1975 Expectation Maximization 1977 Decision tree 1986 Q-learning 1989 Random forest 1995 ❞ — https://twitter.com/farbodsaraf/status/977916871000412160
Slide 29
https://twitter.com/ algorithmia/status/ 1009486664933052416
Slide 30
❝ But saying "powered by AI" is like saying you’re "powered by the internet" or "powered by computer code". By itself, it means nothing. ❞ — https://twitter.com/jensenharris/status/999119292086960128
Slide 31
Learning Regression Ranking Clustering
Slide 32
https://twitter.com/ShawnWildermuth/status/932724124237123584
Slide 33
Slide 34
For children and machines Watch your language
Slide 35
Statistics 101: Linear Regression
Slide 36
❝ We are leveraging machine learning. ❞
Slide 37
https://twitter.com/LesGuessing/status/997146590442799105
Slide 38
Supervised Learning Input features and output labels are defined
Slide 39
Unsupervised Learning Unlabeled dataset Discover hidden relationships
Slide 40
https://xkcd.com/882/
Slide 41
Slide 42
Slide 43
Reinforcement Learning Feedback loop to optimize some parameter
Slide 44
Deep Learning Neural network producing a probability vector Lots of training and parallelization
Slide 45
https://www.youtube.com/watch?v=bxe2T-V8XRs
Slide 46
Access to a unique data set is inherently valuable
Slide 47
Slide 48
❝ "What's the difference between AI and ML?" "It's AI when you're raising money, it's ML when you're trying to hire people." ❞ —https://twitter.com/WAWilsonIV/status/925599712849174528
Slide 49
Domain
Slide 50
Patterns Trend ( stationary ) Cyclical Seasonal Irregular
Slide 51
Anomaly Point Anomalies Contextual Anomalies Collective Anomalies
Slide 52
Breakouts Mean Shift Ramp Up
Slide 53
Anomaly Detection with Machine Learning Supervised Learning Unsupervised Learning
Slide 54
Examples IT operations: Spiking 500s Security analytics: Unusual DNS activity Business analytics: Rare log message
Slide 55
Visual Inspection Complex, fast moving data Humans not made to stare at graphs Easy to miss
Slide 56
Where is the Anomaly?
Slide 57
Static Rules Definition False positives & negatives Tuning and adjustment
Slide 58
Which threshold?
Slide 59
444444444444 Machine learning
Slide 60
❝ OH: "Do you run any CPU intensive application on your laptop? Like, machine learning, or Slack?" ! ❞ — https://twitter.com/jpetazzo/status/932464823530430464
Slide 61
Frameworks TensorFlow Keras SciKit ...
Slide 62
How to build ML pipelines? ETL Data storage Optimization algorithms
Slide 63
❝ I see you expected clean data. " at's cute. ❞
Slide 64
Slide 65
Model Baseline: What is normal?
Slide 66
Slide 67
Slide 68
Unsupervised
Slide 69
Evolves "Online" model learns continuously and ages out data
Slide 70
Single Time Series Example: Unusual tra ! c?
Slide 71
Multiple Time Series Multiple metrics or single metric split up Each series modeled independently Example: Unusual activity by country?
Slide 72
Slide 73
Dataset
Slide 74
nginx access log {
"source" : "/home/ec2-user/data/production-4/prod4elasticlog/_logs/access-logs541.log" ,
"beat" : {
"hostname" : "ip-172-31-5-206" ,
"name" : "ip-172-31-5-206" ,
"version" : "5.4.0" },
"@timestamp" : "2017-03-08T11:44:51.562Z" ,
"read_timestamp" : "2017-06-20T08:49:58.538Z" ,
"fileset" : {
"name" : "access" ,
"module" : "nginx" },
Slide 75
"nginx" : {
"access" : {
"body_sent" : {
"bytes" : "3262" },
"url" : "/assets/blt1afcb054f02e257c/logo-activision.svg" ,
"geoip" : {
"continent_name" : "Asia" ,
"country_iso_code" : "IN" ,
"location" : {
"lat" : 20 ,
"lon" : 77 } },
Slide 76
"response_code" : "200" ,
"user_agent" : {
"device" : "Other" ,
"os_name" : "Other" ,
"os" : "Other" ,
"name" : "Other" },
"http_version" : "1.1" ,
"method" : "GET" ,
"remote_ip" : "192.19.197.26" } },
"prospector" : {
"type" : "log" } }
Slide 77
Slide 78
Slide 79
Slide 80
Slide 81
Slide 82
Slide 83
Most of the internet went down
Slide 84
PS: When everything is on ! , nobody cares about your downloads
Slide 85
Slide 86
Counterfactual Reasoning Which host / IP / ... is involved in the anomaly
Slide 87
Slide 88
Slide 89
Slide 90
Combine Multiple Models
Slide 91
Slide 92
Correlation ≠ causation
Slide 93
https://xkcd.com/552/
Slide 94
https://xkcd.com/925/
Slide 95
Common problems Correlated features will mess up any model
Slide 96
Common problems ! row out most features if they are just noise
Slide 97
More features Future predictions
Slide 98
Slide 99
More features Clustering
Slide 100
Conclusion
Slide 101
Agenda Machine Learning Domain Dataset
Slide 102
Rules of Machine Learning: Best Practices for ML Engineering http://martin.zinkevich.org/rules_of_ml/ rules_of_ml.pdf
Slide 103
43 rules Rule #1: Don’t be afraid to launch a product without machine learning Rule #14: Starting with an interpretable model makes debugging easier Rule #16: Plan to launch and iterate
Slide 104
Machine Learning 44444 ohne Hype 4444 Philipp Krenn 44 @xeraa
