Bringing Deep Learning into production
0 likes•855 views
- The document discusses deep learning frameworks and how to choose one for a given environment. It summarizes the strengths, weaknesses, opportunities and threats of popular frameworks like TensorFlow, Theano, Torch, Caffe, DeepLearning4J and H2O. - It recommends H2O as a good choice for enterprise environments due to its ease of use, scalability on big data, integration with Spark, Java/Scala support and commercial support. DeepLearning4J is also recommended for more advanced deep neural networks and multi-dimensional arrays. - The document proposes using Spark as a middleware to leverage multiple frameworks and avoid vendor lock-in, and describes Agile Lab's recommended stack for enterprises which combines H
Bringing Deep Learning into production
- 2. Brief introduction • CTO & co-founder of Agile Lab • Data & Tech addicted • Contributor of Spark Notebook • Spark early adopter • Certified Cassandra Architect • DeepLearning enthusiast
- 3. Who is Agile Lab ? GO BIG (data) or GO HOME http://www.meetup.com/it-IT/Torino-Scala-Programming-Big-Data-Meetup/
- 4. What we do Applications High scalability Decision Support Systems data engineering, data mining and data «meaning» Big Data Strategies Training Reactive, NoSQL, Big Data, Machine learning
- 6. Deep Learning is trending
- 7. What is Deep Learning • Deep learning is just another name for artificial neural networks • An algorithm is deep if the input is passed through several non-li nearity before being output • Deep learning is discovering the features that best represent the problem, rather than just a way to combine them
- 8. Deep Learning: Use cases
- 9. Do you want start with Deep Learning ? Let’s choose the right tools !!
- 10. Deep Learning Frameworks • Deeplearning4J • TensorFlow • Caffe • Theano • Torch • Spark ML MultilayerPerceptrons • H2O • CNTK • MatLab • maxDNN And many others
- 11. How to choose Background Target Environment Vision
- 12. Background Productivity !! • Scala • Java Big Data Engineer • Java • Python Math Engineer • R • Python Statistician
- 13. Target Environment Trained model should be deployable !! Trained Model Dev Env Prod Env
- 14. Target Environment Prod Env Dev Env Training Data Cleaning ETLScheduling ML Pipeline - Track model performance over time - Care about SLA - Continous tweaks
- 15. Enterprise Architecture HADOOP Online DataStore Enterprise Service BUS DataIntegrationLayer Data Integration Layer DataIntegrationLayer External Sources ANALYTICS VALUE ADDED SERVICES API SERVICES Internal Business Sources Internal System Sources DeepLearning
- 16. Easy Wins Training pipeline should run on Spark or Hadoop Trained Model should be represented in Java objects
- 17. Vision: keep in mind Scaling High Level dynamic languages are incredibly productive for prototyping and data exploration Scaling on larger data sets quickly runs into performance limitations Keep in mind scaling requirements from beginning
- 18. Vision: simplify the pipeline Copy & Sample data from Dev Env to Data Scientist Env Prototype in Python or R Train model Predict on validation Data Translate Model to match Prod Env Java, MapReduce, Spark Deploy training pipeline and model
- 19. Easy Wins Datascientists should work directly on distributed environment Datascientist and big data engineers should co-operate on the same platform
- 20. SWOT Analysis
- 21. Tensor Flow Strenghts: - Powered By Google - Nice UI Weaknesses: - Powered By Google - No support for “inline” matrix operations Slow Opportunities: - Awesome community Threats: - No Scala or Java integration - No commercial support
- 22. Theano Strenghts: - Grand Daddy of deep learning - RNN and CNN - Computational graph abstraction - Python Weaknesses: - No support for Hadoop or Spark - No plug & play nets Opportunities: - Great community Threats: - No Scala or Java integration - No commercial support
- 23. Torch Strenghts: - GPU support - Lots of pretrained models and packages - Easy to use Weaknesses: - Lua language Opportunities: - Backed by DeepMind and Facebook Threats: - No Scala or Java integration - No commercial support
- 24. Caffè Strenghts: - C++ & Python - Good Performance - GPU Support Weaknesses: - Focused on image processing Opportunities: - Backed by Yahoo for Spark integration - Gpu Clustering Threats: - No commercial support
- 25. DeepLearning4j Strenghts: - GPU support - Java and Scala - Full DNN set - Support Hadoop, Spark & Akka Weaknesses: - Not for dummies Opportunities: - Commercial support - SkyMind Threats: - Not so sexy for DataScientist because of Java/Scala
- 26. H2O • Easy to use Web UI • Multi language API • Run directly on HDFS or S3 • Model is Java PoJo • Big Data Ready • Really Fast • Compressed data • Regularization • Grid Search • GPU is still on roadmap • CNN and RNN too
- 27. H2O - Flow
- 28. H20 – Sparkling Water • Python, R and Scala API • Best Kagglers use H20 • Tons of tools for profiling and tu ning • Spark leverage • Best in class algorithms – battle tested • Regolarization • Grid search
- 29. H20 – Sparkling Water
- 30. Workflow POJO Java Training Set Embeddable in: • J2EE App • Spark Job • MR Job • DWH as UDF training
- 31. Spark as middleware Using Spark as middleware, you can leverage : • Deeplearning4J • H2O • TensorFlow ( Arimo Extension) • Caffe ( Yahoo Extension ) • ML MultilayerPerceptrons and future implementations NO tech provider Lock-in
- 32. Our Stack for Enterprise • Ready for Enterprise and Hadoop World • Deployable into Java Env • Notebook ( Flow ) • H2O for out of the box algorithms • DeepLearning 4J for advanced DNN and n-dimension array manipulation • Good usability for both DataScientists and Big Data Engineers • Enterprise Support along the whole stack