Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Deep learning on Hadoop/Spark -NextML

Deep Learning on hadoop at galvanize for next ml

Deep learning on Hadoop/Spark -NextML

  1. 1. DEEP LEARNING ON HADOOP ADAM GIBSON // DL4J presentation@ Scale out Deep Learning on YARN Next.ML
  2. 2. deeplearning4j.org presentation@ ADAM / DL4J ● Adam: ○ DL specialist, in residence at Galvanize ○ Head programmer in DL4J project ● SkyMind ○ Commercial support arm of Deeplearning4j ● Deep Learning for Java (DL4J) ○ a distributed, open-source deep-learning framework for the JVM ○ has most major deep neural nets
  3. 3. deeplearning4j.org presentation@ OVERVIEW ● What is Deep Learning? ● DL major types of Neural Networks: a. Feed-forward NN b. Recurrent NN c. Convolutional NN d. Recursive NN ● Scaling DL4J with HADOOP
  4. 4. deeplearning4j.org What is DL? Machine Perception, Pattern Recognition
  5. 5. deeplearning4j.org presentation@ WHY IS DL HARD? We see this Machines see this… (Where’s the cat?)
  6. 6. deeplearning4j.org presentation@ WHAT’S IT GOOD FOR? ● Recommendation engines: anticipate what you will buy or click ● Anomaly detection: ○ fraud in e-commerce ○ tumors in X-rays ○ loans’ likely to default ● Signal processing: ○ customer lifetime value estimation ○ necessary inventory ○ approaching market crash ● Image
  7. 7. deeplearning4j.org presentation@ Google, Nov. 17 2014 “TWO PIZZAS SITTING ON A STOVETOP”
  8. 8. deeplearning4j.org presentation@ FACIAL RECOGNITION & FEATURE HIERARCHY DEEP means “high level abstractions achieved by using model architectures composed of multiple non-linear transformations”, typically traditional Machine Learning algorithms, mostly Neural-Network based (source: Wikipedia)
  9. 9. deeplearning4j.org presentation@ DL4J FACIAL RECONSTRUCTIONS
  10. 10. deeplearning4j.org presentation@ LEARNING PROPERTIES OF DL ● They can learn unsupervised data ● They save data scientists months of work ● They eat vectors and learn patterns based on occurrence statistics ● They can handle millions of parameters at a time ● After training, they can be deployed for fast classification
  11. 11. deeplearning4j.org MAJOR TYPES OF DL NETWORKS
  12. 12. deeplearning4j.org presentation@ FEED FORWARD NN ● Perceptron (1957) ● Back-prop learning ● No loops ● Can have different topologies, typically multi layered
  13. 13. deeplearning4j.org presentation@ RECURRENT NN ● Loops ● Temporal behavior ● Used for temporal series
  14. 14. deeplearning4j.org presentation@ CONVOLUTIONAL NN Specific type of Feed-Forward NN, impressive recent results ● Neocognitron (Fukushima’80) ● Improvement (LeCun’98) ● Google Brain (Ng’11)
  15. 15. deeplearning4j.org presentation@ RECURSIVE NN ● Recursive Autoencoder Deals with variable-length input, has a topological structure, good for NLP
  16. 16. deeplearning4j.org presentation@ RECURSIVE NN ● Recursive Neural Tensor Network
  17. 17. deeplearning4j.org SCALING WITH HADOOP AND YARN
  18. 18. deeplearning4j.org presentation@ SGD: Serial vs. Parallel Model Training Data Worker 1 Master Partial Model Global Model Worker 2 Partial Model Worker N Partial Model Split 1 Split 2 Split 3 …
  19. 19. deeplearning4j.org presentation@ WHAT’S ? ● Everything is in the cloud now, right?
  20. 20. deeplearning4j.org presentation@ MapReduce vs. Parallel iterative Input Output Processor Processor Processor Superstep 1 Processor Processor Superstep 2 . . . Processor
  21. 21. deeplearning4j.org DEMO TIME
  22. 22. deeplearning4j.org Thank you

    Be the first to comment

    Login to see the comments

  • PhilippeChevry

    Jan. 20, 2015
  • KenichiTakahashi1

    Apr. 15, 2015
  • KanSakamoto

    Apr. 19, 2015
  • kazuakey

    Apr. 20, 2015
  • netghost

    Apr. 20, 2015
  • caidong

    Jun. 10, 2015
  • keiichiroseida

    Jun. 30, 2015
  • YoshiakiAmano

    Jun. 30, 2015
  • naoki4277

    Jun. 30, 2015
  • darkseed

    Jul. 2, 2015
  • seralf

    Aug. 6, 2015
  • jayuloy

    Sep. 5, 2015
  • DongjoonHyun

    Sep. 23, 2015
  • BillChen12

    Oct. 4, 2015
  • SmrutiranjanSahu

    Oct. 6, 2015
  • vikramdutt

    Dec. 10, 2015

Deep Learning on hadoop at galvanize for next ml

Views

Total views

4,666

On Slideshare

0

From embeds

0

Number of embeds

428

Actions

Downloads

107

Shares

0

Comments

0

Likes

16

×