GABRIELA DE QUEIROZ R E A D Y T O U S E D E E P L E A R N I N G M O D E L S : A L L Y O U N E E D I S 5 M I N U T E S SENI OR DEVELOPER ADVOCATE (ML /DL/A I) @ IBM @gdequeiroz | www.k-roz.com | [email protected]
GABRIELA DE QUEIROZ ‣ Sr Developer Advocate, IBM ‣ Founder, R-Ladies ‣ Lead Data Scientist ‣ Data Scientist ‣ Statistician/Epidemiologist/ Researcher About me Data + Community + Mentor + Advocate @gdequeiroz | www.k-roz.com | [email protected]
Center for Open Source Data and AI Technologies (CODAIT) CODAIT aims to make AI solutions easier to create, deploy, and manage in the enterprise Relaunch of the Spark Technology Center (STC) to reflect expanded mission 30+ open source developers! Watson West Building 505 Howard St. San Francisco, California Improving Enterprise AI lifecycle in Open Source Gather Data Analyze Data Machine Learning Deep Learning Deploy Model Maintain Model Python Data Science Stack Fabric for Deep Learning (FfDL) Mleap + PFA Scikit-Learn Pandas Apache Spark Apache Spark Jupyter Model Asset eXchange Keras + Tensorflow CODAIT codait.org
Improving Enterprise AI lifecycle in Open Source Gather Data Analyze Data Machine Learning Deep Learning Deploy Model Maintain Model Python Data Science Stack Fabric for Deep Learning (FfDL) Mleap + PFA Scikit-Learn Pandas Apache Spark Apache Spark Jupyter Model Asset eXchange Keras + Tensorflow CODAIT: Enabling End-to-End AI in the Enterprise
The Model Asset eXchange enables domain experts to use deep learning in the enterprise. Q: What is deep learning? A: Machine learning using deep neural networks.
Q: What is a deep neural network? A: A neural network with multiple hidden layers.
What is a neural network? Second layer of linear regressions Multilayer Perceptron Neural Network Dense (3×4) Dense (4×2) Input (3) Output (2) Same network in a more compact notation
What is a deep neural network? A neural network with multiple hidden layers Dense (3×8) Dense (8×6) Input (3) Output (2) Dense (6×4) Dense (4×2) “toy" deep neural network
Applying Deep Learning: Perception Data ??? Train model ??? $$$ Get model ??? Deploy model ??? $$$ Training – Data Scientist Consumption – App Developer, Domain Expert
Applying Deep Learning: Reality Find model Get code Test, verify, fix Train model Deploy model Use model Discovery Execution Consumption 1 2 3 4A 4B 5 @gdequeiroz | www.k-roz.com | [email protected]
Step 2: Get the code Is there a good implementation available? … that does what you need … that is free to use … that is performant enough TensorFlow code to build ResNet50 neural network graph
Or … Step 2: Get the pre-trained weights Is there a good pre-trained model available? … that does what you need … that is free to use … that is performant enough
Step 3: Verify the model you found Check … … that does what you need … that is free to use (license) … that is performant enough (computation & accuracy)
Step 4 (b): Figure out how to deploy your model … adjust inference code (or write from scratch) … package inference code and model code, and pre-trained weigths together … deploy your package
Model Asset eXchange • Wide variety of domains (text, audio, image, etc) • Multiple deep learning frameworks • Vetted and tested code and IP • Build and deploy a model web service in seconds • Training on Fabric for Deep Learning (FfDL) or Watson Machine Learning in minutes ibm.biz/model-exchange
DATA MODEL COMPUTER RESOURCES EXPERTISE Input/Output Processing Pre-Trained Model REST API Deep Learning Asset on Model Asset Exchange ibm.biz/model-exchange Deployable Models
Swagger Specification Deep Learning Asset on Model Asset Exchange ibm.biz/model-exchange Deployable Models Deploy Inference Endpoint Metada Endpoint Microservice REST API
Highlights • Image, audio, text, healthcare, time-series and more • Pre- / post-processing & inference wrapped up in Docker container • Generic API framework code - Flask RESTPlus • Swagger specification for API • One-line deployment locally and on a Kubernetes cluster