Hands on image recognition with scala spark and deep learning4j

Hands on image recognition with scala spark and deep learning4j

• 1.
  Hands-on Image Recognition withScala, Spark and DeepLearning4j Presented by Guglielmo Iozzia Kiev, April 20th 2018
• 2.
  Something About Me Big Data Delivery Lead at Optum (UHG)  Previously at IBM and FAO of the UN  Current fields of expertise are Big Data, ML/DL and DevOps  Past experience in JVM languages (Java, Groovy, Scala) development, test automation, CI/CD
• 3.
  Something About Me Author of the upcoming book “Hands-on Deep Learning with Apache Spark”  I love preparing home-made pizza
• 4.
  The Dublin TechHub
• 5.
  Agenda  Scala forData Science  Deep Learning  Definition  Multilayer Neural Networks  Apache Spark  Overview  DeepLearning4j  ETL  CNN  Image Recognition Example
• 6.
  Scala for DataScience  Most part of the systems/tools in the Big Data/ML space run on the JVM.  Robustness and performance when it comes to production system and large datasets.  Plenty of frameworks available:
• 7.
  Deep Learning  Itis a subset of machine learning that can solve particularly hard and large-scale problems in areas such as natural language processing or image classification.  It is based on Multilayered Neural Networks.
• 8.
  Deep Learning
• 9.
  Artificial Neural Networks
• 10.
  Multilayer Neural Networks The first layer, called the input layer, is where features are input.  The last layer is called the output layer.  Any layer that is not an input or output layer is a hidden layer.
• 11.
  Multilayer Neural Networks Different variations.  CNN: used in image recognition.  RNN: used in NLP.
• 12.
  Apache Spark  Itis an Open Source fast cluster-computing platform.  Data are loaded in distributed memory (RAM) over a cluster of machines.  Compared to Hadoop MapReduce, it runs programs up to 100x faster when the data fits in memory, or 10x faster on disk.  It provides support for Java, Scala, Python and R.
• 13.
  Apache Spark: jobexecution
• 14.
  Apache Spark: components
• 15.
  DeepLearning4j: Overview  Itis an Open Source distributed Deep Learning library for the JVM.  It is integrated with Hadoop and Spark.  It provides support for both CPUs and GPUs.  It allows the import of neural net models from the most major frameworks via Keras.
• 16.
  DeepLearning4j: libraries  Deeplearning4J:Neural Net Platform  ND4J: Numpy for the JVM  DataVec: Tool for Machine Learning ETL Operations  JavaCPP: The Bridge Between Java and Native C++  Arbiter: Evaluation Tool for Machine Learning Algorithms  RL4J: Deep Reinforcement Learning for the JVM.
• 17.
  ETL with DL4j LiveExample
• 18.
  DL4j CNN Live Example
• 19.
  Image Recognition Live Example
• 20.
  Useful Links Apache Spark:http://spark.apache.org/ DeepLearning4j: https://deeplearning4j.org/ Keras: https://keras.io/
• 21.
  Q & A
• 22.
  Wrap Up Linkedin: https://ie.linkedin.com/in/giozzia Twitter:@GuglielmoIozzia Blog: googlielmo.blogspot.com DZone: https://dzone.com/users/2532948/virtualramblas.html Hands-On Deep Learning with Apache Spark: https://www.packtpub.com/big-data-and-business-intelligence/hands- deep-learning-apache-spark