Deep Learning with Apache Spark - MasterClass!

Name: Deep Learning with Apache Spark - MasterClass!
Rating: 3.8 (6 reviews)

A fast-paced guide to implementing practical hands-on examples, streamlining Deep Learning with Apache Spark

Created byPackt Publishing

Last updated 5/2019

English

What you'll learn

Explore deep learning neural networks such as RBM, RNN, and DBN using some of the most popular industrial deep learning frameworks.
Learn how to leverage big data to solve real-world problems using deep learning.
Understand how to formulate real-world prediction problems as machine learning tasks, how to choose the right neural net architecture for a problem, and how to train neural nets using DL4J.
Configure a Convolutional Neural Network (CNN) to extract value from images.
Create a deep network with multiple layers to perform computer vision.
Classify speech and audio data.
Get up-and-running and gain an insight into the deep learning library DL4J and its practical uses.
Train and test neural networks to fit your data model.

Course content

3 sections • 86 lectures • 5h 32m total length

The Course Overview2:05
This video provides an overview of the entire course.
Review of Key Machine Learning Terminology and Fundamentals4:10
In this video, we will learn about machine learning terminology and fundamentals.
Learn about unsupervised learning
Learn about unsupervised learning
Understand when to use which
Fundamentals of Deep Networks: Feature Engineering3:10
In this video, we will learn the fundamentals of deep networks.
Understand what feature engineering is
Learn how to extract features from data
The Building Blocks of Deep Learning4:46
In this video, we will understand the building blocks of deep learning.
Understand what deep learning is
Learn when to use deep learning
Learn what a deep learning algorithm consists of
Learning Path for Deep Learning2:22
In this video, we will understand the learning path for deep learning.
Get an overview of neural network
Understand the recurrent neural network
Get an overview of the DL4J library
Deep Learning Use Cases3:08
In this video, we will learn about deep learning use cases.
Learn where it is best to use the deep learning approach
Look at the use cases
Pre-requisites and Installation3:44
In this video, we will learn a few pre-requisites and the installation steps for DL4J.
Understand the data set
Understand what you want to achieve with neural networks
Add DL4J to your application
Up and Running with DL4J on Spark4:11
In this video, we will jump right into DL4J.
Download the MNIST database in our model
Define the parameters of the input data set and the parameters of the neural network
Create ImageRecorder
Configuration and Test Run5:51
In this video, we will configure the neural network.
Configure and fit the neural network
Validate this against your data set
Run your code
Up and Running with TensorFlow on Spark from Yahoo4:00
In this video, we will dive deeper into TensorFlow on Spark.
Use TensorFlow via Python API
Fetch TensorFlow and Spark flow dependencies
Create similar neural network like in the previous video
Understanding the Basics of Deep Learning6:57
In this video, we will understand the basics of deep learning.
Add ND4J library
Create an instance of INDArray
Create a matrix using INDArray
ND4J for NumPy-like Arrays and Operations9:26
In this video, we will learn how to use ND4J for NumPy-like arrays.
Validate options to create arrays with pre-defined data
Fill INDArray with random data
Create more 3-dimensional arrays
Data.Vec for Data Preparation Pipelines4:54
In this video, we will learn about data preparation pipelines
Perform math operations on vectors
Compare two vectors using ND4J API
Perform ceil, floor and round operations on vectors
DL4J for Building Neural Network Architectures3:46
In this video, we will learn about neural network architectures.
Perform statistical operations on vectors
Calculate min and max
Calculate variance and standard deviation using ND4J
Understanding the Basics of GPU5:41
In this video, we will understand the basics of GPU.
Understand why we use GPU with deep learning
Learn what we can benefit from this
Learn why this is better than using standard CPU
Parallel Training with Multiple GPUs4:23
In this video, we will learn about multiple GPUs.
Add GPU dependencies to DL4J project
Configure CUDA execution environment
Designing a Basic CNN4:52
In this video, we will design a basic CNN.
Define the MNist data set problem
Configure a multi-layer network
Use ParallelWrapper for proxy processing to GPUs
Implement a Basic CNN on DL4J in Spark5:09
In this video, we will learn to implement a basic CNN in DL4J on Spark.
Perform classification using GPUs
Validate your results
Basics and Design of RNN2:44
In this video, we will learn the basics and design of RNN.
Understand what RNN is
Learn when to use RNN
Understand the problems solved by RNN
Implement a Basic RNN on DL4J in Spark4:52
In this video, we will implement a basic RNN in DL4J on Spark.
Create a neural network
Setup a recurrent neural network
Use the proper parameters
Design a Basic LSTM4:36
In this video, we will learn to design a basic LSTM.
Get an overview of the LSTM network
Understand how It differs from RNN
Understand why we need LSTM
Implement a Basic LSTM in Spark5:25
In this video, we will learn to implement a basic LSTM in Spark.
Train RNN with LSTM to guess subsequent characters in a sentence
Tweak the number of iterations
Validate your results
Test your knowledge

The Course overview3:03
This video gives an overview of the entire course.
Creating a Dataframes in Pyspark3:36
Dataframes will serve as the framework for any and all data that will be used in building deep learning models. This video explains how we create our Spark cluster and configure our first Dataframes.
Configure our Dataframes
Manipulating Columns in a Pyspark Dataframes1:33
This video walks through the steps for the string conversion to a numeric value in the Dataframes.
Learn to update the Dataframes
Learn to convert Dataframes to an array
Converting a PySparkdataframe to an array3:08
In order to form the building blocks of the neural network, the PySpark dataframe must be converted into an array. Python has a very powerful library, numpy, that makes working with arrays simple.
Study the steps to convert the dataframe into an array
Visualizing an Array in a Scatterplot2:47
A powerful method for understanding the relationship between height, weight, and gender is by visualizing thedata points feeding the neural network.
Study the steps to visualize an array through scatter plot
Setting up Weights and Biases for Input into the Neural Network3:20
This video will focus on setting the weights that create the input which feeds into theactivation function.
Study the steps to set up weights and bias of neural network
Normalizing the Input Data for the Neural Network2:07
Neural networks work more efficiently when the inputs are normalized. This minimizes themagnitude of a particular input affecting the overall outcome over other potential inputsthat have lower values of magnitude. This video will normalize the height and weight inputs of the current individuals.
Learn the steps to normalize the height and weight
Validating Array for Optimal Neural Network Performance1:27
This video will walk through validating that our array has been normalized.
Study the steps for validation
Setting up the Activation Function with Sigmoid2:44
An activation function is used in a neural network to help determine the output, whether it is a yes or no, true or false, or in our case 0 or 1. This video walks through the steps of creating and plotting a sigmoid function with sample data.
Create and plot a sigmoid function
Creating the Sigmoid Derivative Function1:41
The sigmoid function is a unique function where the value of the derivative of the sigmoidfunction includes the value of the sigmoid function. This video walks through the steps to create a sigmoid derivative function.
Create a sigmoid derivative function
Calculating the Cost Function in a Neural Network3:24
Calculating the cost function is used by the neural network to determine howwell, the predicted outcome matched the original or actual outcome, given the 29 individualdata points that are currently available.
Study the steps to calculate the cost function
Predicting Gender based on Height and Weight1:56
A predictive model is only useful if it can actually predict based on new information. Thisis the case with a simple logistic or linear regression, or a more complex neural network model.
Study the steps to predict gender based on height and weight.
Visualizing Prediction Scores1:26
This video walks through the steps to visualize all of the predicted points in a graph.
Learn the steps to visualize the prediction score
Pain Point #1: Importing MNIST Images5:00
One of the most common datasets used for image classification is the MNIST dataset, which is composed of thousands of samples of handwritten digits.
Learn the steps of accessing the MNIST images
Pain Point #2: Visualizing MNIST Images2:42
Plotting images is often a major pain point when dealing with graphics within a Jupyter notebook. This video will walk through the steps to visualize the MNIST handwritten images in a Jupyter notebook.
Learn the steps to visualize handwritten images
Pain Point #3: Exporting MNIST Images as Files1:21
We often need to work within the image directly and not as an array vector. This video will guide us through converting our arrays to .png images.
Study the steps to convert the sample of MNIST arrays
Pain Point #4: Augmenting MNIST Images3:26
This video walks through the steps to increase the frequency and augmentation of our nine sample images.
See the steps of augmenting the MNIST images
Pain Point #5: Utilizing Alternate Sources for Trained Images2:36
This video will explain to us how the process of transfer learning works as well as when applied to the MNIST dataset.
Study alternative sources for trained images
Pain Point #6: Prioritizing High-Level Libraries for CNNs7:07
In this video, we will be working with Keras to train a model for recognizing handwritten images from MNIST. This video walks through the steps to build a model to recognize handwritten images from MNIST.
Study the steps to build a model to recognize handwritten images from MNIST.
Downloading the San Francisco Fire Department Calls Dataset3:56
The City of San Francisco does a great job of collecting fire department calls for services across their area. This video will walk through the steps to download and import the .csv file to our Jupyter notebook.
Learn the steps to import the dataset
Identifying the Target Variable of the Logistic Regression Model3:48
A logistic regression model operates as a classification algorithm aiming to predict a binary outcome. In this video, we will specify the best column within the dataset to predict whether an incoming call to the operator is related to fire or non-fire incidents.
Study the steps to visualize the data
Preparing Feature Variables for the Logistic Regression Model3:49
This video will focus on identifying all of the features that will best help the model identify what the target should be.
Study the steps to prepare the feature variables
Applying the Logistic Regression Model3:44
This video will focus on applying a very common classification model called logistic regression.
Learn the steps to apply the model and evaluate the results
Evaluating the Accuracy of the Logistic Regression Model2:43
We will evaluate the performance of predicting whether a call was correctly classified as a fire incident.
Study the steps to evaluate the model performance
Downloading and Analyzing the Therapy Bot Session Dataset4:12
This video will focus on downloading and setting up the dataset that will be used for NLP. It is always important to first analyze any dataset before applying models on that same dataset.
Learn the steps for downloading the chatbot data
Study the steps to profile the text data
Visualizing Word Counts in the Dataset1:35
This video walks through the steps to convert the Spark Dataframes into a visualization that can be seen in the Jupyter notebook
Learn the steps to calculate sentiment analysis of text
Calculating Sentiment Analysis of Text3:22
Sentiment analysis is the ability to derive tone and feeling behind a word or series of words. This section will utilize techniques in python to calculate a sentiment analysis score from the 100 transactions in our dataset.
Study the steps to apply sentiment score to a dataset
Removing Stop Words from the Text2:52
This video walks through the steps to remove stop words.
Learn the steps to remove stop words
Training and Evaluating TF-IDF Model Performance5:30
We will train our TF-IDF NLP model and see if we can classify these transactions as either escalate or do_not_escalate.
Study the steps to train the TF-IDF model
Comparing Model Performance to a Baseline Score1:35
This video will walk through the steps to calculate the baseline accuracy.
Learn the steps to calculate the baseline accuracy
Downloading Stock Market Data for Apple3:36
This video walks through the steps for downloading historical stock market data for Apple.
Learn the steps to download the data
Exploring and Visualizing Stock Market Data for Apple4:25
Before any modeling and predictions are performed on the data, it is important to first explore and visualize the data at hand for any hidden gems.
Study the steps to explore and visualize the stock market data
Preparing Stock Data for Model Performance5:25
In the video, we will perform transformation and visualization. This video walks through the steps for preparing the stock market data for our model.
Study the steps to prepare the stock data
Building the LSTM Model2:29
We will perform model management and hyperparameter tuning of our model. This video walks through the steps to setting up and tuning the LSTM model.
Learn the steps to set up the LSTM model
Evaluating the Model1:58
This video walks through visualizing and calculating the predicted vs. actual stock quotes for Apple in 2017 and 2018.
Study the steps visualize and calculate the predicted versus actual stock data
Test your knowledge

The Course Overview2:20
This video will give you an overview about the course.
Analyzing Input Text Data That Will Need to Be Classified3:31
In this video, we will analyze input text data to be classified.
Delve into business domain speech data
Analyze texts from finance, health, and science
Load data into ML model
Configuring Word Vectors That Will Be Used in Our Network3:56
In this video, we will configure word vectors that will be used in our network.
Load data into paragraph vectors API construct
Set the tokenizer
Create the model
Adding Layers to Deep Neural Network3:54
In this video, we will add layers to deep neural network.
Build classification model
Leverage labelled data
Load unlabeled data will be used to validate model
Asserting Classification of Input Sentences4:08
In this video, we will assert classification of input sentences.
Transform unlabeled data into feature vector
Assign document into classes
Validate results
Generating Input Video Data4:47
In this video, we will generate input video data.
Generate input video
Create MP4 files for different kinds of shapes
Add text labels per frame
Creating a Neural Network for Video Classification3:56
In this video, we will create a neural network for video classification.
Configure multi-layer
Adapt to labelled input data
Create last layer that produces proper number of classes
Adding RNN and LSTMs to Network to Perform a Task Better4:17
In this video, we will add RNN and LSTM to network to perform a task better.
Validate neural network parameters
Configure LSTM layer
Start training
Testing and Validating Deep Learning Model5:39
In this video, we will test and validate our deep learning model.
Write code for cross-validation
Start code
Validate the video frames which are assigned to proper classes
Creating Paragraph Vectors2:36
In this video, we will create paragraph vectors.
Analyze input text data
Transform input data into feature vector
Adding Labels to Non-Labelled Data4:10
In this video, we will learn to add labels to non-labelled data.
Parameterize neural network
Build RNN layers
Normalize output of neurons
Finding Similarity between Vectors3:59
In this video, we will learn to find similarity between vectors.
Initialize the model
Create training data
Creating a Model That Can Guess the Meaning of The Word3:24
In this video, we will create a model that can guess the meaning of the word.
Train the model
Validate the model
Perform actual training and testing
Anomaly Detection Problem Explained4:14
In this video, we will learn about the anomaly detection problem.
Analyze “normal” input data set
Look at the feature vector
Understand what an anomaly is
Extracting Features from Input Data Using Multi-Layer Approach5:25
In this video, we will extract features from input data using multi-layer approach.
Create iterator that is fetching anomaly data
Create test set iterator
Create training set iterator
Adding Layer That Finds an Actual Anomaly4:27
In this video, we will find a layer that finds an actual anomaly.
Create model that encodes multiple features
Create multiple LSTM layers
Initialize the neural network
Testing and Validating Results from Our Deep Learning Model5:33
In this video, we will test and validate results from our deep learning model.
Fit the model with labelled anomaly data
Write a method to find outliers
Creating Data Generator for GAN3:33
In this video, we will create a data generator for GAN.
Analyze labelled input news data
Look at the test set
Look at the training set
Adding Discriminator for Our Data6:06
In this video, we will add discriminator for our data.
Tokenize input data
Leverage Word2Vec algorithm
Write output feature vector to a file
Create Classifier for Generated Data4:55
In this video, we will create a classifier for generated data.
Create dataset iterator
Load news data into DataSetIterator
Configure multi-layer
Performing Validation of Our Model4:27
In this video, we will learn to validate our model.
Start model
Assign news data into classes
Validate our model against training data
Configuring Spark for High Data Distribution5:12
In this video, we will configure Spark for high data distribution.
Look at data partitioning in Spark engine
Understand partitioning
Implement custom partitioner
Fetching Input Set into Distributed Data Set Using Spark API4:26
In this video, we will fetch an input set into distributed data set using Spark API.
Fulfill DataSet with data
Perform operations on DataSets
Join DataSets
Creating Training Master That Supervise Computations on the Workers4:55
In this video, we will create a training master that will supervise computations on the workers.
Understand which part of code executes on the master
Understand Spark lazy nature
Use actions and transformations
Evaluating Speed of Distributed Training Using Spark3:03
In this video, we will evaluate the speed of distributed training.
Understand the nature of action and immutability
Measure the time of actions
Use Spark RDD without re-usability
Monitoring of Models Using Spark UI4:29
In this video, we will monitor models using Spark UI.
Start Spark flow
Open Spark UI
Monitor flow via Spark UI
Speeding Up Computations by Employing Caching3:48
In this video, we will learn to speed up computations by employing caching.
Create Spark expensive operations
Measure the time taken for these operations
Add cache and compare the time of results with caching and non-caching
Partitioning Deep Learning Data into Several Workers3:54
In this video, we will partition deep learning data into several workers.
Understand Spark partitioning
Look at pre-defined partitioners
Choose a partitioner explicitly
Tweaking Spark Workers Configuration4:19
In this video, we will learn to tweak Spark workers configuration.
Understand Spark-submit
Understand Spark-submit parameters
Tweak parameters

Requirements

Basic knowledge of Machine Learning and Big Data concepts is assumed.

Description

Deep learning has solved tons of interesting real-world problems in recent years. Apache Spark has emerged as the most important and promising Machine Learning tool and currently a stronger challenger of the Hadoop ecosystem. In this course, you’ll learn about the major branches of AI and get familiar with several core models of Deep Learning in its natural way.

This comprehensive 3-in-1 course is a fast-paced guide to implementing practical hands-on examples, streamlining Deep Learning with Apache Spark. You’ll begin by exploring Deep Learning Neural Networks using some of the most popular industrial Deep Learning frameworks. You’ll apply built-in Machine Learning libraries within Spark, also explore libraries that are compatible with TensorFlow and Keras. Next, you’ll create a deep network with multiple layers to perform computer vision and improve cybersecurity with Deep Reinforcement Learning. Finally, you’ll use a generative adversarial network for training and create highly distributed algorithms using Spark.

By the end of this course, you'll develop fast, efficient distributed Deep Learning models with Apache Spark.

Contents and Overview

This training program includes 3 complete courses, carefully chosen to give you the most comprehensive training possible.

The first course, Deep Learning with Apache Spark, covers deploying efficient deep learning models with Apache Spark. The tutorial begins by explaining the fundamentals of Apache Spark and deep learning. You will set up a Spark environment to perform deep learning and learn about the different types of neural net and the principles of distributed modeling (model- and data-parallelism, and more). You will then implement deep learning models (such as CNN, RNN, LTSMs) on Spark, acquire hands-on experience of what it takes, and get a general feeling for the complexity we are dealing with. You will also see how you can use libraries such as Deeplearning4j to perform deep learning on a distributed CPU and GPU setup. By the end of this course, you'll have gained experience by implementing models for applications such as object recognition, text analysis, and voice recognition. You will even have designed human expert games.

The second course, Apache Spark Deep Learning Recipes, covers over 35 recipes that streamline eep learning with Apache Spark. This video course starts offs by explaining the process of developing a neural network from scratch using deep learning libraries such as Tensorflow or Keras. It focuses on the pain points of convolution neural networks. We’ll predict fire department calls with Spark ML and Apple stock market cost with LSTM. We’ll walk you through the steps to classify chatbot conversation data for escalation. By the end of the video course, you'll have all the basic knowledge about apache spark.

The third course, Mastering Deep Learning using Apache Spark, covers designing Deep Learning models to edge industrial-grade apps. You’ll begin with building deep learning networks to deal with speech data and explore tricks to solve NLP problems and classify video frames using RNN and LSTMs. You’ll also learn to implement the anomaly detection model that leverages reinforcement learning techniques to improve cybersecurity. Moving on, you’ll explore some more advanced topics by performing prediction classification on image data using the GAN encoder and decoder. Then you’ll configure Spark to use multiple workers and CPUs to distribute your Neural Network training. Finally, you’ll track progress, solve the most common problems in your neural network, and debug your models that run within the distributed Spark engine.

By the end of this course, you'll develop fast, efficient distributed Deep Learning models with Apache Spark.

About the Authors

● Tomasz Lelek is a Software Engineer, programming mostly in Java and Scala. He has been working with the Spark and ML APIs for the past 5 years with production experience in processing petabytes of data. He is passionate about nearly everything associated with software development and believes that we should always try to consider different solutions and approaches before solving a problem. Recently he was a speaker at conferences in Poland, Confitura and JDD (Java Developers Day), and at Krakow Scala User Group. He has also conducted a live coding session at Geecon Conference. He is a co-founder of initlearn, an e-learning platform that was built with the Java language. He has also written articles about everything related to the Java world.

Who this course is for:

Data Scientist, Data Analysts, Big Data Architects, Anyone with a basic understanding of Deep Learning and Big Data concepts interested in developing fast, efficient distributed Deep Learning models with Apache Spark

Deep Learning with Apache Spark - MasterClass!

What you'll learn

Explore related topics

Course content

Deep Learning with Apache Spark22 lectures • 1hr 40min

Apache Spark Deep Learning Recipes35 lectures • 1hr 49min

Mastering Deep Learning using Apache Spark29 lectures • 2hr 3min

Requirements

Description

Who this course is for: