Apache Spark and BigDL Deep Learning Training provides the developers to develop large datasets in very speedy way. The tackling of the program can be usually done with faster hardware (GPUs), Optimized codes and also some sort of parallelism.


  • Understanding the concept of training neural networks basically on spark cluster
  • Dealing with the Network related training for BigDL
  • Understanding the concept of using Spark local in conjunction with DL4J


  • Completely secure and safe for the Blockchain developers
  • Synchronization and Serialization overhead concepts are most overwhelmed
  • Complete enhancement in Training Performance
Duration: 1 Day

Course Content:
  1. Basic Overview of Apache Spark
  2. Understanding the features of Apache Spark
  3. Core prerequisites for Apache Spark
  4. Basic Configuration of TrainingMaster
  5. Dealing with basic dependencies for Training
  6. Understanding the concept of Spark Example Repository
  7. Using the concept of GPU on Spark

    • Dealing with YARN and GPUs
    • Dealing with Mesos and GPUs
  8. Memory Configuration for Spark on YARN

    • Concept of DeepLearning Managing Memory
    • Concept of YARN Handling Memory Management
    • Memory Configuration Deeplearning spark training
  9. Spark Locality Configuration for Improved Training Performance
  10. Understanding the concept of Performance Debugging
  11. Concept of Collecting Training Performance Information
  12. Dealing with Caching/Persisting RDD and RDD
  13. Working with Kyro Serialization
  14. Working with Intel MKL on Amazon Elastic MapReduce
  15. Basic Overview of BigDL
  16. Understanding the core concept of Distributed Deep Learning on Spark
  17. Basic Features of BigDL
  18. Reasons for Using BigDL
  19. Working of BigDL
  20. Practical Scenarios on BigDL