deep water - bringing tensorflow, caffe, mxnet to h2o
TRANSCRIPT
DeepWaterOr:BringingTensorFlowetal.toH2O
ArnoCandel,PhDChiefArchitect,Physicist&Hacker,H2O.ai
@ArnoCandelJuly182016
Overview
Machine Learning (ML)
Artificial Intelligence (A.I.)
Computer Science (CS)
H2O.ai
Deep Learning (DL)hot hot hot hot hot
ASimpleDeepLearningModel:ArtificialNeuralNetwork
heartbeat
blood pressure
oxygensend to regular care
send to intensivecare unit (ICU)
IN: data OUT: prediction
nodes : neuron activations (real numbers) — represent features arrows : connecting weights (real numbers) — learned during training
: non-linearity x -> f(x) — adds model complexity
from 1970s, now rebranded as DL
• conceptuallysimple• nonlinear• highlyflexibleandconfigurable
• learnedfeaturescanbeextracted• canbefine-tunedwithmoredata• efficientformulti-classproblems• world-classatpatternrecognition
DeepLearningProsandCons
Pros:• hardtointerpret• theorynotwellunderstood• slowtotrainandscore• overfits,needsregularization• manyhyper-parameters• inefficientforcategoricalvariables• verydatahungry,learnsslowly
Cons:
DeepLearninggotboostedrecentlybyfastercomputers
BriefHistoryofA.I.,MLandDL
John McCarthyPrinceton, Bell Labs, Dartmouth, later: MIT, Stanford
1955: “A proposal for the Dartmouth summer research project on Artificial Intelligence”
with Marvin Minsky (MIT), Claude Shannon (Bell Labs) and Nathaniel Rochester (IBM)
http://www.asiapacific-mathnews.com/04/0403/0015_0020.pdf
A step back: A.I. was coined over 60 years ago
1955proposalfortheDartmouthsummerresearchprojectonA.I.
“We propose that a 2-month, 10-man study of artificial intelligence be carried out during the summer of 1956 at Dartmouth College in Hanover, New Hampshire. The study is to proceed on the basis of the conjecture that every aspect of learning and any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it. An attempt will be made to find how to make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves. We think that a significant advance can be made in one or more of these problems if a carefully selected group of scientists work on it together for one summer.”
Step1:GreatAlgorithms+FastComputers
http://nautil.us/issue/18/genius/why-the-chess-computer-deep-blue-played-like-a-human
1997: Playing Chess (IBM Deep Blue beats Kasparov)
ComputerScience 30customCPUs,60billionmovesin3mins
“No computer will ever beat me at playing chess.”
Step2:MoreData+Real-TimeProcessing
http://cs.stanford.edu/group/roadrunner/old/presskit.html
2005: Self-driving CarsDARPA Grand Challenge, 132 miles (won by Stanford A.I. lab*)
Sensors&ComputerSciencevideo,radar,laser,GPS,7Pentiumcomputers
“No computer will ever drive a car!?”
*A.I. lab was established by McCarthy et al. in the early 60s
Step3:BigData+In-MemoryClusters
2011: Jeopardy (IBM Watson)
In-MemoryAnalytics/ML4TBofdata(incl.wikipedia),90servers,16TBRAM,Hadoop,6millionlogicrules
https://www.youtube.com/watch?v=P18EdAKuC1U https://en.wikipedia.org/wiki/Watson_(computer)
Note: IBM Watson received the question in electronic written form, and was often able to press the answer button faster than the competing humans.
“No computer will ever answer random questions!?”
“No computer will ever understand my language!?”
2014: Google(acquired Quest Visual)
DeepLearning ConvolutionalandRecurrent
NeuralNetworks,withtrainingdatafromusers
Step4:DeepLearning
• Translate between 103 languages by typing • Instant camera translation: Use your camera to translate text instantly in 29 languages • Camera Mode: Take pictures of text for higher-quality translations in 37 languages • Conversation Mode: Two-way instant speech translation in 32 languages • Handwriting: Draw characters instead of using the keyboard in 93 languages
Step5:AugmentedDeepLearning
2014: Atari Games (DeepMind)
2016: AlphaGo (Google DeepMind)
DeepLearning +reinforcementlearning,treesearch,
MonteCarlo,GPUs,playingagainstitself,…
Go board has approx. 200,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 (2E170) possible positions.
trainedfromrawpixelvalues,nohumanrules
“No computer will ever beat the best Go master!?”
Microsoft had won the Visual Recognition challenge: http://image-net.org/challenges/LSVRC/2015/
Step6:A.I.ChatbotshaveOpinionstoo!
WhatWillChange?
Today Tomorrow
BetterData—BetterModels—BetterResults
Example:Fraud
Prediction
WhataboutJobs?
Anything that can be automated will be automated.
Jobs of the past: assembly line work, teller (ATMs), taxi-firm receptionist
Jobs being automated away now: resume matching, driving, language translation, education
Jobs being automated away soon: healthcare, arts & crafts, entertainment, design, decoration, software engineering, politics, management, professional gaming, financial planning, auditing, real estate agent
Jobs of the future: professional sports, food & wine reviewer
Maybewe’llfinallygettheeatingmachine?
https://www.youtube.com/watch?v=n_1apYo6-Ow
1936 Modern Times (Charlie Chaplin)
LiveH2ODeepLearningDemo:PredictAirplaneDelays
10 nodes: all 320 cores busy
real-time, interactive model inspection in Flow
116M rows, 6GB CSV file 800+ predictors (numeric + categorical)
model trained in <1 min: 2M+ samples/second
Deep Learning Model
H2O Elastic Net (GLM): 10 secs alpha=0.5, lambda=1.379e-4 (auto)
H2O Deep Learning: 45 secs 4 hidden ReLU layers of 20 neurons, 1 epoch
Features have non-linear impact
Chicago, Atlanta, Dallas: often delayed
SignificantPerformanceGainswithDeepLearning
Predict departure delay (Y/N) on 20 years of airline flight data (116M rows, 12 cols, categorical + numerical data with missing values)
WATCH NOW
AUC: 0.656
AUC: 0.703 (higher is better, ranges from 0.5 to 1)
Feature importances
10 nodes: Dual E5-2650 (8 cores, 2.6GHz), 10GbE
READ MORE
Kaggle challenge 2nd place winner Colin Priest
READ MORE“FormyfinalcompetitionsubmissionIusedanensembleofmodels,including3deeplearningmodelsbuiltwithRandh2o.”
“IdidreallylikeH2O’sdeeplearningimplementationinR,though-theinterfacewasgreat,thebackendextremelyeasytounderstand,anditwasscalableandflexible.DefinitelyatoolI’llbegoingbackto.”
H2ODeepLearningCommunityQuotes
H2ODeepLearningCommunityQuotes
READ MORE WATCH NOW
“H2ODeepLearningmodelsoutperformotherGleasonpredictingmodels.”
READ MORE
“…combineADAMandApacheSparkwithH2O’sdeeplearningcapabilitiestopredictanindividual’spopulationgroupbasedonhisorhergenomicdata.Ourresultsdemonstratethatwecanpredicttheseverywell,withmorethan99%accuracy.”
H2ODeepLearningCommunityQuotes
KDNuggetsPollaboutDeepLearningTools&Platforms
http://www.kdnuggets.com
H2OandTensorFlowaretied
usageofDeepLearningtoolsinpastyear
TensorFlow+H2O+ApacheSpark=Anythingispossible
https://github.com/h2oai/sparkling-water/blob/master/py/examples/notebooks/TensorFlowDeepLearning.ipynb https://www.youtube.com/watch?v=62TFK641gG8
IntegrationwithexistingGPUbackends
Leverageopen-sourcetoolsandresearchforTensorFlow,Caffe,mxnet,Theano,etc.
ScalabilityandEaseofUse/DeploymentofH2O
Distributedtraining,real-timemodelinspectionFlow,R,Python,Spark/Scala,Java,REST,POJO,Steam
ConvolutionalNeuralNetworks
Image,video,speechrecognition,etc.
RecurrentNeuralNetworks
Sequences,timeseries,etc.NLP:naturallanguageprocessing
HybridNeuralNetworksArchitectures
Speechtotexttranslation,imagecaptioning,sceneparsing,etc.
DeepWater:Next-GenDeepLearninginH2O
EnterpriseDeepLearningforBusinessTransformation
DeepWater:Next-GenDeepLearninginH2O
Don’tmissourDeepLearningsessionstomorrowafternoon!
TensorFlow mxnetCaffe H2O