High Performance Machine Learning in R with H2O
28 likes8,510 views
This document summarizes a presentation by Erin LeDell from H2O.ai about machine learning using the H2O software. H2O is an open-source machine learning platform that provides APIs for R, Python, Scala and other languages. It allows distributed machine learning on large datasets across clusters. The presentation covers H2O's architecture, algorithms like random forests and deep learning, and how to use H2O within R including loading data, training models, and running grid searches. It also discusses H2O on Spark via Sparkling Water and real-world use cases with customers.
![H2O.ai
Machine Intelligence
Sparkling Shell
H2O Sparkling Water
• Sparkling Water is transparent integration of H2O
into the Spark ecosystem.
• H2O runs inside the Spark Executor JVM.
• Provides advanced machine learning algorithms
to Spark workflows.
• Sophisticated alternative to the default MLlib
library in Spark.
• Sparkling Shell is just a standard Spark shell with
additional Sparkling Water classes
• export MASTER=“local-cluster[3,2,1024]”
• spark-shell —jars sparkling-water.jar
Spark Integration
Benefits](https://image.slidesharecdn.com/h2ohpcconoct2015-151014181822-lva1-app6891/85/High-Performance-Machine-Learning-in-R-with-H2O-45-320.jpg)
High Performance Machine Learning in R with H2O
- 1. H2O.ai Machine Intelligence High Performance Machine Learning in R with H2O Erin LeDell Ph.D. ISM HPC on R Workshop Tokyo, Japan October 2015
- 2. H2O.ai Machine Intelligence Introduction • Statistician & Machine Learning Scientist at H2O.ai in Mountain View, California, USA • Ph.D. in Biostatistics with Designated Emphasis in Computational Science and Engineering from UC Berkeley (focus on Machine Learning) • Worked as a data scientist at several startups • Written a handful of machine learning R packages
- 3. H2O.ai Machine Intelligence Agenda • What/who is H2O.ai? • H2O Machine Learning Software • H2O Architecture • H2O in R & Demo • Sparking Water: H2O on Spark • Ensembles in H2O & Demo
- 4. H2O.ai Machine Intelligence H2O.ai H2O Company H2O Software • Team: 35. Founded in 2012, Mountain View, CA • Stanford Math & Systems Engineers • Open Source Software • Ease of Use via Web Interface • R, Python, Scala, Spark & Hadoop Interfaces • Distributed Algorithms Scale to Big Data
- 5. H2O.ai Machine Intelligence H2O.ai Founders SriSatish Ambati • CEO and Co-founder at H2O.ai • Past: Platfora, Cassandra, DataStax, Azul Systems, UC Berkeley • CTO and Co-founder at H2O.ai • Past: Azul Systems, Sun Microsystems • Developed the Java HotSpot Server Compiler at Sun • PhD in CS from Rice University Dr. Cliff Click
- 6. H2O.ai Machine Intelligence Scientific Advisory Council Dr. Trevor Hastie Dr. Rob Tibshirani Dr. Stephen Boyd • John A. Overdeck Professor of Mathematics, Stanford University • PhD in Statistics, Stanford University • Co-author, The Elements of Statistical Learning: Prediction, Inference and Data Mining • Co-author with John Chambers, Statistical Models in S • Co-author, Generalized Additive Models • 108,404 citations (via Google Scholar) • Professor of Statistics and Health Research and Policy, Stanford University • PhD in Statistics, Stanford University • COPPS Presidents’ Award recipient • Co-author, The Elements of Statistical Learning: Prediction, Inference and Data Mining • Author, Regression Shrinkage and Selection via the Lasso • Co-author, An Introduction to the Bootstrap • Professor of Electrical Engineering and Computer Science, Stanford University • PhD in Electrical Engineering and Computer Science, UC Berkeley • Co-author, Convex Optimization • Co-author, Linear Matrix Inequalities in System and Control Theory • Co-author, Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers
- 7. H2O.ai Machine Intelligence H2O Platform Part 1 of 7 High Performance ML in R with H2O
- 8. H2O.ai Machine Intelligence H2O Software H2O is an open source, distributed, Java machine learning library. APIs are available for: R, Python, Scala & JSON
- 9. H2O.ai Machine Intelligence H2O Overview Speed Matters! No Sampling Interactive UI Cutting-Edge Algorithms • Time is valuable • In-memory is faster • Distributed is faster • High speed AND accuracy • Scale to big data • Access data links • Use all data without sampling • Web-based modeling with H2O Flow • Model comparison • Suite of cutting-edge machine learning algorithms • Deep Learning & Ensembles • NanoFast Scoring Engine
- 10. H2O.ai Machine Intelligence Current Algorithm Overview Statistical Analysis • Linear Models (GLM) • Cox Proportional Hazards • Naïve Bayes Ensembles • Random Forest • Distributed Trees • Gradient Boosting Machine • R Package - Super Learner Ensembles Deep Neural Networks • Multi-layer Feed-Forward Neural Network • Auto-encoder • Anomaly Detection • Deep Features Clustering • K-Means Dimension Reduction • Principal Component Analysis • Generalized Low Rank Models Solvers & Optimization • Generalized ADMM Solver • L-BFGS (Quasi Newton Method) • Ordinary Least-Square Solver • Stochastic Gradient Descent Data Munging • Integrated R-Environment • Slice, Log Transform
- 12. H2O.ai Machine Intelligence H2O Flow Interface
- 15. H2O.ai Machine Intelligence H2O Architecture Part 2 of 7 High Performance ML in R with H2O
- 16. H2O.ai Machine Intelligence Distributed Key Value Store H2O Frame H2O Components • Multi-node cluster with shared memory model. • All computations in memory. • Each node sees only some rows of the data. • No limit on cluster size. • Objects in the H2O cluster such as data frames, models and results are all referenced by key. • Any node in the cluster can access any object in the cluster by key. • Distributed data frames (collection of vectors). • Columns are distributed (across nodes) arrays. • Each node must be able to see the entire dataset (achieved using HDFS, S3, or multiple copies of the data if it is a CSV file). H2O Cluster
- 17. H2O.ai Machine Intelligence Key’s Home Node Distributed K/V Store • The H2O K/V Store is a classic peer-to-peer distributed hash table. • There is no “name-node” nor central key dictionary. • Each key has a home-node, but the homes are picked pseudo-randomly per-key. • This allows us to force keys to “home” to different nodes (usually for load-balance reasons). • A key's “home” is solely responsible for breaking ties in racing writes and is the “source of truth.” • Keys can be cached anywhere, and both reads & writes can be cached (although a write is not complete until it reaches “home”.) Peer-to-Peer Pseudo-Random Hash
- 18. H2O.ai Machine Intelligence Data Shape Data in H2O • We read data fully parallelized from: HDFS, NFS, Amazon S3, URLs, URIs, CSV, SVMLight. • Data is highly compressed (about 2-4 times smaller than gzip). • Memory bound, not CPU bound. • If data accessed linearly, as fast as C or Fortran. • Speed = data volume / memory bandwidth • ~50GB / sec (varies by hardware). • Table width: <1k fast, <10k works, <100k slow • Table length: Limited only by memory • We have tested 10’s of billions of rows (TBs) Highly Compressed Speed
- 19. H2O.ai Machine Intelligence Distributed H2O Frame Diagram of distributed arrays. An “H2O Frame” is a collection of distributed arrays.
- 20. H2O.ai Machine Intelligence Optimizations Communication in H2O • H2O requires network communication to JVMs in unrelated process or machine memory spaces. • That network communication can be fast or slow, or may drop packets & sockets (even TCP can silently fail), and may need to be retried. • H2O implements a reliable RPC which retries failed communications at the RPC level. • We can pull cables from a running cluster, and plug them back in, and the cluster will recover. • Message data is compressed in a variety of ways (because CPU is cheaper than network). • Short messages are sent via 1 or 2 UDP packets; larger message use TCP for congestion control. Network Communication Reliable RPC
- 21. H2O.ai Machine Intelligence Ease of Use Data Processing in H2O • Map/Reduce is a nice way to write blatantly parallel code (although not the only way), and we support a particularly fast and efficient flavor. • Distributed fork/join and parallel map: within each node, classic fork / join • We have a GroupBy operator running at scale (called ddply in the R community). • GroupBy can handle millions of groups on billions of rows, and runs Map/Reduce tasks on the group members. • H2O has overloaded all the basic data frame manipulation functions in R and Python. • Tasks such as imputation and one-hot encoding of categoricals is performed inside the algorithms. Map Reduce Group By
- 22. H2O.ai Machine Intelligence H2O on Amazon Part 3 of 7 High Performance ML in R with H2O
- 23. H2O.ai Machine Intelligence H2O on Amazon EC2 H2O can easily be deployed on an Amazon EC2 cluster. The GitHub repository contains example scripts that help to automate the cluster deployment.
- 25. H2O.ai Machine Intelligence NERSC Supercomputers Edison is ranked 34 and Hopper is 62 on TOP500. We hope to be running H2O at NERSC soon… :-)
- 26. H2O.ai Machine Intelligence H2O in R Part 4 of 7 High Performance ML in R with H2O
- 27. H2O.ai Machine Intelligence Design “h2o” R package on CRAN • The only requirement to run the “h2o” R package is R >=3.1.0 and Java 7 or later. • Tested on many versions of Linux, OS X and Windows. • The easiest way to install the “h2o” R package is to install directly from CRAN. • Latest version: http://h2o.ai/download • No computation is ever performed in R. • All computations are performed (in highly optimized Java code) in the H2O cluster and initiated by REST calls from R. Requirements Installation
- 28. H2O.ai Machine Intelligence Start H2O Cluster from R
- 29. H2O.ai Machine Intelligence H2O in R: Load Data R code example: Load data
- 30. H2O.ai Machine Intelligence H2O in R: Train & Test R code example: Train and Test a GBM
- 31. H2O.ai Machine Intelligence H2O in R: Plotting plot(fit) plots scoring history over time.
- 32. H2O.ai Machine Intelligence H2O in R: Grid Search R code example: Execute a DL Grid Search
- 33. H2O.ai Machine Intelligence Live H2O Demo! https://gist.github.com/ledell Install H2O (stable): install_h2o_slater.R Demo: h2o_higgs_simple_demo.R
- 34. H2O.ai Machine Intelligence H2O Ensemble Part 5 of 7 High Performance ML in R with H2O
- 35. H2O.ai Machine Intelligence What is Ensemble Learning? What it is: ✤ “Ensemble methods use multiple learning algorithms to obtain better predictive performance that could be obtained from any of the constituent learning algorithms.” (Wikipedia, 2015) ✤ Random Forests and Gradient Boosting Machines (GBM) are both ensembles of decision trees. ✤ Stacking, or Super Learning, is technique for combining various learners into a single, powerful learner using a second-level metalearning algorithm. Ensembles typically achieve superior model performance over singular methods. However, this comes at a price — computation time. What it’s not:
- 36. H2O.ai Machine Intelligence Super Learner Why ensembles? H2O Ensemble Overview • Regression • Binary Classification / Ranking • Coming soon: Support for multi-class • H2O Ensemble implements the Super Learner algorithm. • The Super Learner algorithm finds the optimal (based on defined loss) combination of a collection of base learning algorithms. • When a single algorithm does not approximate the true prediction function well. • When model performance is the most important factor (over training speed and interpretability). ML Tasks
- 37. H2O.ai Machine Intelligence Super Learner: The setup
- 38. H2O.ai Machine Intelligence Super Learner: The Algorithm
- 39. H2O.ai Machine Intelligence H2O Ensemble R Interface R code example: Set up an H2O Ensemble
- 40. H2O.ai Machine Intelligence H2O Ensemble R Interface R code example: Train and Test an Ensemble
- 41. H2O.ai Machine Intelligence Live H2O Demo! https://gist.github.com/ledell Install H2O Ensemble: install_h2oEnsemble.R Demo: lending_club_bad_loans_ensemble.R
- 42. H2O.ai Machine Intelligence Sparkling Water Part 6 of 7 High Performance ML in R with H2O
- 43. H2O.ai Machine Intelligence SparkR Apache Spark and SparkR • Apache Spark is an open source in-memory processing engine built around speed. • It was originally developed at UC Berkeley in 2009. • Spark is commonly used on commodity clusters (such as Amazon EC2). • CRAY has been working with Spark community to optimize Spark for CRAY supercomputers. • Spark is written in Scala, but APIs exist for Python and R. • “SparkR” is the R API and has been part of Spark since Spark 1.4 (June, 2015). Apache Spark Spark for HPC
- 44. H2O.ai Machine Intelligence Distributed Frames H2O vs SparkR • Major difference is that SparkR creates a collection of slave R instances. • H2O uses a single R session and communicates to the H2O Java cluster via REST calls. • In Spark 1.5 (latest release), only GLM is accessible in R via SparkR. • All H2O algorithms are available via R. • Both H2O and Spark use distributed data frames. • SparkR is most useful for data processing on distributed data frames. Architecture Machine Learning Algorithms
- 45. H2O.ai Machine Intelligence Sparkling Shell H2O Sparkling Water • Sparkling Water is transparent integration of H2O into the Spark ecosystem. • H2O runs inside the Spark Executor JVM. • Provides advanced machine learning algorithms to Spark workflows. • Sophisticated alternative to the default MLlib library in Spark. • Sparkling Shell is just a standard Spark shell with additional Sparkling Water classes • export MASTER=“local-cluster[3,2,1024]” • spark-shell —jars sparkling-water.jar Spark Integration Benefits
- 46. H2O.ai Machine Intelligence H2O in Action Part 7 of 7 High Performance ML in R with H2O
- 47. H2O.ai Machine Intelligence Actual Customer Use Cases Ad optimization (200% CPA lift with H2O) Fraud detection — 11% higher accuracy with H2O Deep Learning (saves millions of dollars) Propensity to Buy model factory (60,000 models, 15x faster with H2O)
- 48. H2O.ai Machine Intelligence H2O on https://www.kaggle.com/mlandry • H2O starter scripts available on Kaggle • H2O is used in many competitions on Kaggle • Mark Landry, H2O Data Scientist and Competitive Kaggler
- 49. H2O.ai Machine Intelligence Where to learn more? • H2O Online Training (free): http://learn.h2o.ai • H2O Slidedecks: http://www.slideshare.net/0xdata • H2O Video Presentations: https://www.youtube.com/user/0xdata • H2O Community Events & Meetups: http://h2o.ai/events • Machine Learning & Data Science courses: http://coursebuffet.com
- 51. H2O.ai Machine IntelligenceCustomers • Community • Evangelists 35 Speakers Training 2-Full Days Nov. 9 - 11 http://world.h2o.ai 20% Discount code: h2ocommunity
- 52. H2O.ai Machine Intelligence ありがとう @ledell on Twitter, GitHub erin@h2o.ai http://www.stat.berkeley.edu/~ledell