Deep Learning and NLP with Spark

Explore deep learning and natural language processing with Apache Spark in this 45-minute conference talk from Scala Days Berlin 2016. Discover how to implement a Spark-ready version of the Long Short-Term Memory (LSTM) neural network, a complex deep learning model used for challenging NLP tasks like automatic summarization, machine translation, and question answering. Learn about one-to-one relationships, word embeddings, image captioning, recurrent nets, and sentiment analysis. Witness a live LSTM demo featuring interactive, real-time visualizations using Spark Notebook and Spark Streaming. Gain insights into dependencies, data loading, memory management, and network building for multilayer architectures. Understand feature limitations and explore practical applications in California science.

Intro
Natural Language Processing
Neural Networks
OnetoOne Relationships
Word Tyvek
Image Captioning
Recurrent Nets
Sentiment Analysis
About Skymind
About Andys notebook
About Juju
Notebook
Idea
Dependencies
Load Data
Memory
Parameters
Multilayer Network Builder
Data Preparation
Feature
Limitations
California Science
References