SlideShare a Scribd company logo
1 of 61
Download to read offline
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Information Extraction for Enterprise
Documents
A I M 4 0 7
Cyrus M Vahid
Principal Evangelist, AWS AI Labs
cyrusmv@amazon.com
Vivek Srivastava
ML Product Manager
Workday
Henry Zhang
ML Engineer
Workday
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
Apache MXNet and why we use it
Amazon SageMaker and
distributed training
Amazon SageMaker and DevOps
style deep learning
Business Case
Framework
Model and Algorithms
Results
Productization
Key Learnings
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Computational Dependency/Graph
• 𝑧 = 𝑥 ⋅ 𝑦
• 𝑘 = 𝑎 ⋅ 𝑏
• 𝑡 = 𝜆𝑧 + 𝑘
x y
𝑧
x
𝜆
𝑢
x
a
x
b
k
𝑡
+
1 1
2
3
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Computational Dependency/Graph
net = mx.sym.Variable('data')
net = mx.sym.FullyConnected(net, name='fc1', num_hidden=64)
net = mx.sym.Activation(net, name='relu1', act_type="relu")
net = mx.sym.FullyConnected(net, name='fc2', num_hidden=10)
net = mx.sym.SoftmaxOutput(net, name='softmax')
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Computational Dependency/Graph
net = mx.sym.Variable('data')
net = mx.sym.FullyConnected(net, name='fc1', num_hidden=64)
net = mx.sym.Activation(net, name='relu1', act_type="relu")
net = mx.sym.FullyConnected(net, name='fc2', num_hidden=10)
net = mx.sym.SoftmaxOutput(net, name='softmax')
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Training
import logging
logging.getLogger().setLevel(logging.DEBUG) # logging to stdout
# create a trainable module on compute context
mlp_model = mx.mod.Module(symbol=mlp, context=ctx)
mlp_model.fit(train_iter,
eval_data=val_iter,
optimizer='sgd',
optimizer_params={'learning_rate':0.1},
eval_metric='acc',
batch_end_callback = mx.callback.Speedometer(batch_size, 100),
num_epoch=10)
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Computational Dependency/Graph
net = mx.sym.Variable('data')
net = mx.sym.FullyConnected(net, name='fc1', num_hidden=64)
net = mx.sym.Activation(net, name='relu1', act_type="relu")
net = mx.sym.FullyConnected(net, name='fc2', num_hidden=10)
net = mx.sym.SoftmaxOutput(net, name='softmax')
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Training Efficiency – 92%
https://mxnet.incubator.apache.org/tutorials/vision/large_scale_classification.html
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
End-to-End
Machine Learning
Platform
Zero setup Flexible Model
Training
Pay by the second
$
Amazon SageMaker
Build, train, and deploy machine learning models at scale
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon SageMaker and Distributed Training
• Faster training through SageMaker Streaming for Custom Algorithms.
• Boilerplate code for your algorithms to train over a cluster.
PCA Bemchmark
if len(hosts) == 1:
kvstore = 'device' if num_gpus > 0 else 'local’
else:
kvstore = 'dist_device_sync' if num_gpus > 0 else 'dist_sync’
trainer = gluon.Trainer(net.collect_params(), 'sgd’,
{'learning_rate': learning_rate, 'momentum': momentum},
kvstore=kvstore)
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Training code
• Matrix Factorization
• Regression
• Principal Component Analysis
• K-Means Clustering
• Gradient Boosted Trees
• And More!
Amazon provided Algorithms
Bring Your Own Script (IM builds the Container)
Bring Your Own Algorithm (You build the Container)
Fetch Training data
Save Model Artifacts
Fully
managed –
Secured
–
Amazon ECR
Save Inference Image
IM Estimators in
Apache Spark
CPU GPU HPO
Distributed Training
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Automatic Model Tuning
Training code
• Factorization Machine
• Regression/classification
• Principal Component Analysis
• K-Means Clustering
• XGBoost
• DeepAR
• And More
SageMaker built-in Algorithms Bring Your Own Script (prebuilt containers) Bring Your Own Algorithm
Fetch Training data Save Model Artifacts
Fully
managed –
Secured–
Automatic Model Tuning
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Evolution of DL Frameworks
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Why Gluon
Simple, Easy-to-
Understand Code
Flexible, Imperative
Structure
Dynamic Graphs High Performance
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Gluon Code – Network Definition
net = gluon.nn.HybridSequential()
with net.name_scope():
net.add(gluon.nn.Dense(units=64, activation='relu'))
net.add(gluon.nn.Dense(units=10))
softmax_cross_entropy = gluon.loss.SoftmaxCrossEntropyLoss()
net.initialize(mx.init.Xavier(magnitude=2.24), ctx=ctx, force_reinit=True)
trainer = gluon.Trainer(net.collect_params(), 'sgd', {'learning_rate': 0.02})
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Gluon Code – Training
smoothing_constant = .01
for e in range(10):
cumulative_loss = 0
for i, (data, label) in enumerate(train_data):
data = data.as_in_context(model_ctx).reshape((-1, 784))
label = label.as_in_context(model_ctx)
with autograd.record():
output = net(data)
loss = softmax_cross_entropy(output, label)
loss.backward()
trainer.step(data.shape[0])
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
What’s New – GluonCV
• A Deep Learning Toolkit for Computer Vision
• Features:
• Training scripts that reproduces SOTA results reported in latest
papers
• A large set of pre-trained models
• Carefully designed APIs and easy-to-understand implementations
• Community support
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
What’s New - GluonNLP
• A deep learning toolkit for natural language processing
• Features:
• Training scripts to reproduce SOTA results reported in research
papers.
• Pre-trained models for common NLP tasks.
• Carefully designed APIs that greatly reduce the implementation
complexity.
• Community support.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
What’s New – Keras Backend
Instance Type GPUs Batch Size
Keras-MXNet
(img/sec)
Keras-
TensorFlow
(img/sec)
C5.18X Large 0 32df 13 4
P3.8X Large 1 32 194 184
P3.8X Large 4 128 764 393
P3.16X Large 8 256 1068 261
Instance Type GPUs Batch Size
Keras-MXNet
(img/sec)
Keras-
TensorFlow
(img/sec)
C5.X Large 0 32 5.79 3.27
C5.8X Large 0 32 27.9 18.2
https://github.com/awslabs/keras-apache-mxnet/tree/master/benchmark
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
What’s New - Sockeye
• A seq2seq toolkit based on Apache MXNet.
• Features:
• Beam search inference
• Easy assembling of multiple models
• Residual connections between RNN layers (Wu et al., 2016) [Deep LSTM with parallelism]
• Lexical biasing of output layer predictions (Arthur et al., 2016) [Low Frequency Words]
• Modeling coverage (Tu et al., 2016) [Keeping attention history to reduce over and under translation]
• Context gating (Tu et al., 2017) [Improving adequacy of translation by controlling rations of source
and target context]
• Cross-entropy label smoothing (e.g., Pereyra et al., 2017)
• Layer normalization (Ba et al, 2016) [improving training time]
• Multiple supported attention mechanisms [dot, mlp, bilinear, multihead-dot, encoder last state,
location]
• Multiple model architectures (Encoder-Decoder Wu et al., 2016, Convolutional Gehring et al, 2017,
Transformer Vaswani et al, 2017,)
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Inference Efficiency - TensorRT
Model Name Relative TRT Sppedup HArdware
Resnet 101 1.99x Titan V
Resnet 50 1.76x Titan V
Resnet 18 1.54x Jetson TX1
cifar_resnext29_16x64d 1.26x Titan V
cifar_resnet20_v2 1.21x Titan V
Resnet 18 1.8x Titan V
Alexnet 1.4x Titan V
https://cwiki.apache.org/confluence/display/MXNET/How+to+use+MXNet-TensorRT+integration
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Inference Efficiency – NNVM
https://aws.amazon.com/blogs/machine-learning/introducing-nnvm-compiler-a-new-open-end-to-end-compiler-for-ai-frameworks/
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Portability - NNVM
https://aws.amazon.com/blogs/machine-learning/introducing-nnvm-compiler-a-new-open-end-to-end-compiler-for-ai-frameworks/
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Portability - ONNX
Model Parameters
Hyper Parameters
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Deployment with Amazon SageMaker
I
ML Hosting Service
Amazon ECR
30 50
10 10
ProductionVariant
Model Artifacts
Inference Image
Model versions
Versions of the same
inference code saved in
inference containers.
Prod is the primary
one, 50% of the traffic
must be served there!
One-Click!
EndpointConfiguration
Inference Endpoint
Amazon Provided Algorithms
Amazon SageMaker
InstanceType: c3.4xlarge
InitialInstanceCount: 3
ModelName: prod
VariantName: primary
InitialVariantWeight: 50
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Recap
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
Business Case
Framework
Model and Algorithms
Results
Productization
Key Learnings
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Expense Report Automation
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AP Invoice Automation
Machine
Learning
Machine
Learning
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Document scanning
CV and NLU to extract and understand text in varied
documents and pre-populate the appropriate fields
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Framework Requirement
• Fast and agile prototyping and experiments
• Option to productize via Python and JVM language by exporting
models
• Best accuracy
• Ideally - active support
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Framework Choice (2017)
• Agile Python prototyping, production JVM language support
• MXNet
• Imperative
• ~50% faster training and inference speed + less memory consumption vs Tensorflow
• Better accuracy for our problem
• Sockeye (S2S)
• Easy to use
• Quick support
• Not Expected: Strong and active support from AWS MXNet team
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
MXNet vs. Tensorflow Accuracy (2017)
MXNet Tensorflow
Perfect
Matching Error
Rate
11.6% 21.36%
Edit Distance
Error Rate
4% 7.75%
• Note: For our problem and models
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Problems
Where are
the texts?
What do
the texts
mean?
What are
the texts?
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Where are the texts?
• Customized model based on You
Only Look Once (YOLO) network
• Added angle of tilt
• Used ResNet instead of CNN
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
What are the texts?
• Customized model based on
WaveNet (Originally intended for
speech)
• ResNet + Dilated Convolution
• Able to scan forward and
backward
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
What do the texts mean?
(Rules-based system)
• Rules-Based System
• Looks into surrounding context
• Fuzzy Regex
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Encoder-Decoder Seq2seq Approach
• Encoder-decoder model with
attention mechanism
• Character embedding to word
embedding
• Vocabulary built within training
data
• Future work to use embeddings
from FastText/Word2vec
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Benefit of Encoder-
Decoder Approach
• This approach is good at inferring
from context
• However, it requires you to have
seen the merchant in training
data
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
CNN-based Box Classification Approach
• CNN-based model to detect
probability of each box
• Leverages word2vec word
embedding
• This approach considers both
language context and
approximate location of each
field
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Ensemble Strategy
• Easier fields – rules based
only
• More difficult fields –
perplexity value decision
• Combined two 0.60 perfect
match model to get 72% on
test set with only 40k training
data
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Transferred Learning
• With limited real document
data, we trained on
generated data.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Sample Business Documents
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Results
• We have competitive results that’s at least on par, and on average
performs better when comparing to major public cloud providers on
Vision part of the model.
• Test Set metrics: ~75–85% accuracy
• Production metrics: ~95% accuracy
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Productization
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Key Learnings
• Value in building models specific for the product/use case
• Prototyping/Experimenting mindset vs. production mindset
• Importance of working with the framework team for challenges
Thank you!
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Vivek Srivastava
Henry Zhang
Cyrus M Vahid
cyrusmv@amazon.com
© 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.

More Related Content

What's hot

Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018
Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018
Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018Amazon Web Services
 
Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018
Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018
Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018Amazon Web Services
 
Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...
Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...
Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...Amazon Web Services
 
Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...
Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...
Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...Amazon Web Services
 
Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...
Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...
Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...Amazon Web Services
 
Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...
Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...
Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...Amazon Web Services
 
Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...
Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...
Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...Amazon Web Services
 
Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...
Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...
Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...Amazon Web Services
 
Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...
Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...
Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...Amazon Web Services
 
Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...
Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...
Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...Amazon Web Services
 
Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...
Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...
Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...Amazon Web Services
 
Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018
Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018
Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018Amazon Web Services
 
Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...
Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...
Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...Amazon Web Services
 
Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018
Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018
Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018Amazon Web Services
 
Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018
Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018
Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018Amazon Web Services
 
New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...
New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...
New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...Amazon Web Services
 
Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...
Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...
Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...Amazon Web Services
 
Machine Learning at the Edge (AIM302) - AWS re:Invent 2018
Machine Learning at the Edge (AIM302) - AWS re:Invent 2018Machine Learning at the Edge (AIM302) - AWS re:Invent 2018
Machine Learning at the Edge (AIM302) - AWS re:Invent 2018Amazon Web Services
 
Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...
Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...
Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...Amazon Web Services
 
Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018
Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018
Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018Amazon Web Services
 

What's hot (20)

Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018
Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018
Unsupervised Learning with Amazon SageMaker (AIM333) - AWS re:Invent 2018
 
Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018
Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018
Alexa Everywhere: A Year in Review (ALX201) - AWS re:Invent 2018
 
Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...
Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...
Bring Your Own Apache MXNet and TensorFlow Scripts to Amazon SageMaker (AIM35...
 
Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...
Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...
Democratize Data Preparation for Analytics & Machine Learning A Hands-On Lab ...
 
Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...
Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...
Building, Training, and Deploying fast.ai Models Using Amazon SageMaker (AIM4...
 
Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...
Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...
Tailor Your Alexa Skill Responses to Deliver Truly Personal Experiences (ALX3...
 
Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...
Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...
Build an Intelligent Multi-Modal User Agent with Voice and NLU (AIM340) - AWS...
 
Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...
Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...
Hollywood's Cloud-Based Content Lakes: Modernized Media Archives (MAE203) - A...
 
Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...
Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...
Deep Learning Applications Using PyTorch, Featuring Facebook (AIM402-R) - AWS...
 
Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...
Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...
Build Human-in-the-Loop Systems with AWS Lambda and Mechanical Turk (AIM354) ...
 
Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...
Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...
Build, Train, and Deploy ML Models with Amazon SageMaker (AIM410-R2) - AWS re...
 
Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018
Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018
Run XGBoost with Amazon SageMaker (AIM335) - AWS re:Invent 2018
 
Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...
Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...
Deep Dive on Amazon Rekognition, ft. Tinder & News UK (AIM307-R) - AWS re:Inv...
 
Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018
Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018
Human-in-the-Loop for Machine Learning (AIM358-R1) - AWS re:Invent 2018
 
Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018
Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018
Robocar Rally 2018 (AIM206-R20) - AWS re:Invent 2018
 
New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...
New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...
New AI/ML Solutions with AWS DeepLens & Amazon SageMaker with ConocoPhillips ...
 
Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...
Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...
Transform the Modern Contact Center Using Machine Learning and Analytics (AIM...
 
Machine Learning at the Edge (AIM302) - AWS re:Invent 2018
Machine Learning at the Edge (AIM302) - AWS re:Invent 2018Machine Learning at the Edge (AIM302) - AWS re:Invent 2018
Machine Learning at the Edge (AIM302) - AWS re:Invent 2018
 
Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...
Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...
Detecting Financial Market Manipulation Using Machine Learning (AIM347) - AWS...
 
Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018
Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018
Detect Anomalies Using Amazon SageMaker (AIM420) - AWS re:Invent 2018
 

Similar to Build Deep Learning Applications Using Apache MXNet, Featuring Workday (AIM407-R) - AWS re:Invent 2018

Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências - MCL...
Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências -  MCL...Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências -  MCL...
Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências - MCL...Amazon Web Services
 
Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...
Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...
Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...Amazon Web Services
 
From Notebook to production with Amazon SageMaker
From Notebook to production with Amazon SageMakerFrom Notebook to production with Amazon SageMaker
From Notebook to production with Amazon SageMakerAmazon Web Services
 
Work with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS Summit
Work with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS SummitWork with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS Summit
Work with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS SummitAmazon Web Services
 
Amazon SageMaker (December 2018)
Amazon SageMaker (December 2018)Amazon SageMaker (December 2018)
Amazon SageMaker (December 2018)Julien SIMON
 
Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...
Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...
Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...Codiax
 
BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...
BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...
BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...Amazon Web Services
 
An Introduction to Amazon SageMaker (October 2018)
An Introduction to Amazon SageMaker (October 2018)An Introduction to Amazon SageMaker (October 2018)
An Introduction to Amazon SageMaker (October 2018)Julien SIMON
 
Time series modeling workd AMLD 2018 Lausanne
Time series modeling workd AMLD 2018 LausanneTime series modeling workd AMLD 2018 Lausanne
Time series modeling workd AMLD 2018 LausanneSunil Mallya
 
Apache MXNet and Gluon
Apache MXNet and GluonApache MXNet and Gluon
Apache MXNet and GluonSoji Adeshina
 
[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...
[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...
[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...Amazon Web Services
 
AWS re:Invent 2018 - ENT321 - SageMaker Workshop
AWS re:Invent 2018 - ENT321 - SageMaker WorkshopAWS re:Invent 2018 - ENT321 - SageMaker Workshop
AWS re:Invent 2018 - ENT321 - SageMaker WorkshopJulien SIMON
 
Introducing Amazon SageMaker - AWS Online Tech Talks
Introducing Amazon SageMaker - AWS Online Tech TalksIntroducing Amazon SageMaker - AWS Online Tech Talks
Introducing Amazon SageMaker - AWS Online Tech TalksAmazon Web Services
 
Introduction to Scalable Deep Learning on AWS with Apache MXNet
Introduction to Scalable Deep Learning on AWS with Apache MXNetIntroduction to Scalable Deep Learning on AWS with Apache MXNet
Introduction to Scalable Deep Learning on AWS with Apache MXNetAmazon Web Services
 
Supercharge your Machine Learning Solutions with Amazon SageMaker
Supercharge your Machine Learning Solutions with Amazon SageMakerSupercharge your Machine Learning Solutions with Amazon SageMaker
Supercharge your Machine Learning Solutions with Amazon SageMakerAmazon Web Services
 
Amazon AI/ML Overview
Amazon AI/ML OverviewAmazon AI/ML Overview
Amazon AI/ML OverviewBESPIN GLOBAL
 
Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...
Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...
Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...Amazon Web Services
 
Amazon SageMaker 內建機器學習演算法 (Level 400)
Amazon SageMaker 內建機器學習演算法 (Level 400)Amazon SageMaker 內建機器學習演算法 (Level 400)
Amazon SageMaker 內建機器學習演算法 (Level 400)Amazon Web Services
 
Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018
Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018
Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018Amazon Web Services
 
Quickly and easily build, train, and deploy machine learning models at any scale
Quickly and easily build, train, and deploy machine learning models at any scaleQuickly and easily build, train, and deploy machine learning models at any scale
Quickly and easily build, train, and deploy machine learning models at any scaleAWS Germany
 

Similar to Build Deep Learning Applications Using Apache MXNet, Featuring Workday (AIM407-R) - AWS re:Invent 2018 (20)

Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências - MCL...
Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências -  MCL...Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências -  MCL...
Machine Learning e Amazon SageMaker: Algoritmos, Modelos e Inferências - MCL...
 
Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...
Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...
Building Deep Learning Applications with TensorFlow and SageMaker on AWS - Te...
 
From Notebook to production with Amazon SageMaker
From Notebook to production with Amazon SageMakerFrom Notebook to production with Amazon SageMaker
From Notebook to production with Amazon SageMaker
 
Work with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS Summit
Work with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS SummitWork with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS Summit
Work with Machine Learning in Amazon SageMaker - BDA203 - Atlanta AWS Summit
 
Amazon SageMaker (December 2018)
Amazon SageMaker (December 2018)Amazon SageMaker (December 2018)
Amazon SageMaker (December 2018)
 
Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...
Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...
Julien Simon, Principal Technical Evangelist at Amazon - Machine Learning: Fr...
 
BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...
BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...
BDA301 Working with Machine Learning in Amazon SageMaker: Algorithms, Models,...
 
An Introduction to Amazon SageMaker (October 2018)
An Introduction to Amazon SageMaker (October 2018)An Introduction to Amazon SageMaker (October 2018)
An Introduction to Amazon SageMaker (October 2018)
 
Time series modeling workd AMLD 2018 Lausanne
Time series modeling workd AMLD 2018 LausanneTime series modeling workd AMLD 2018 Lausanne
Time series modeling workd AMLD 2018 Lausanne
 
Apache MXNet and Gluon
Apache MXNet and GluonApache MXNet and Gluon
Apache MXNet and Gluon
 
[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...
[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...
[NEW LAUNCH!] Introducing Amazon Elastic Inference: Reduce Deep Learning Infe...
 
AWS re:Invent 2018 - ENT321 - SageMaker Workshop
AWS re:Invent 2018 - ENT321 - SageMaker WorkshopAWS re:Invent 2018 - ENT321 - SageMaker Workshop
AWS re:Invent 2018 - ENT321 - SageMaker Workshop
 
Introducing Amazon SageMaker - AWS Online Tech Talks
Introducing Amazon SageMaker - AWS Online Tech TalksIntroducing Amazon SageMaker - AWS Online Tech Talks
Introducing Amazon SageMaker - AWS Online Tech Talks
 
Introduction to Scalable Deep Learning on AWS with Apache MXNet
Introduction to Scalable Deep Learning on AWS with Apache MXNetIntroduction to Scalable Deep Learning on AWS with Apache MXNet
Introduction to Scalable Deep Learning on AWS with Apache MXNet
 
Supercharge your Machine Learning Solutions with Amazon SageMaker
Supercharge your Machine Learning Solutions with Amazon SageMakerSupercharge your Machine Learning Solutions with Amazon SageMaker
Supercharge your Machine Learning Solutions with Amazon SageMaker
 
Amazon AI/ML Overview
Amazon AI/ML OverviewAmazon AI/ML Overview
Amazon AI/ML Overview
 
Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...
Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...
Build, Train, and Deploy Machine Learning for the Enterprise with Amazon Sage...
 
Amazon SageMaker 內建機器學習演算法 (Level 400)
Amazon SageMaker 內建機器學習演算法 (Level 400)Amazon SageMaker 內建機器學習演算法 (Level 400)
Amazon SageMaker 內建機器學習演算法 (Level 400)
 
Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018
Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018
Predicting the Future with Amazon SageMaker - AWS Summit Sydney 2018
 
Quickly and easily build, train, and deploy machine learning models at any scale
Quickly and easily build, train, and deploy machine learning models at any scaleQuickly and easily build, train, and deploy machine learning models at any scale
Quickly and easily build, train, and deploy machine learning models at any scale
 

More from Amazon Web Services

Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...Amazon Web Services
 
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...Amazon Web Services
 
Esegui pod serverless con Amazon EKS e AWS Fargate
Esegui pod serverless con Amazon EKS e AWS FargateEsegui pod serverless con Amazon EKS e AWS Fargate
Esegui pod serverless con Amazon EKS e AWS FargateAmazon Web Services
 
Costruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWSCostruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWSAmazon Web Services
 
Come spendere fino al 90% in meno con i container e le istanze spot
Come spendere fino al 90% in meno con i container e le istanze spot Come spendere fino al 90% in meno con i container e le istanze spot
Come spendere fino al 90% in meno con i container e le istanze spot Amazon Web Services
 
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...Amazon Web Services
 
OpsWorks Configuration Management: automatizza la gestione e i deployment del...
OpsWorks Configuration Management: automatizza la gestione e i deployment del...OpsWorks Configuration Management: automatizza la gestione e i deployment del...
OpsWorks Configuration Management: automatizza la gestione e i deployment del...Amazon Web Services
 
Microsoft Active Directory su AWS per supportare i tuoi Windows Workloads
Microsoft Active Directory su AWS per supportare i tuoi Windows WorkloadsMicrosoft Active Directory su AWS per supportare i tuoi Windows Workloads
Microsoft Active Directory su AWS per supportare i tuoi Windows WorkloadsAmazon Web Services
 
Database Oracle e VMware Cloud on AWS i miti da sfatare
Database Oracle e VMware Cloud on AWS i miti da sfatareDatabase Oracle e VMware Cloud on AWS i miti da sfatare
Database Oracle e VMware Cloud on AWS i miti da sfatareAmazon Web Services
 
Crea la tua prima serverless ledger-based app con QLDB e NodeJS
Crea la tua prima serverless ledger-based app con QLDB e NodeJSCrea la tua prima serverless ledger-based app con QLDB e NodeJS
Crea la tua prima serverless ledger-based app con QLDB e NodeJSAmazon Web Services
 
API moderne real-time per applicazioni mobili e web
API moderne real-time per applicazioni mobili e webAPI moderne real-time per applicazioni mobili e web
API moderne real-time per applicazioni mobili e webAmazon Web Services
 
Database Oracle e VMware Cloud™ on AWS: i miti da sfatare
Database Oracle e VMware Cloud™ on AWS: i miti da sfatareDatabase Oracle e VMware Cloud™ on AWS: i miti da sfatare
Database Oracle e VMware Cloud™ on AWS: i miti da sfatareAmazon Web Services
 
Tools for building your MVP on AWS
Tools for building your MVP on AWSTools for building your MVP on AWS
Tools for building your MVP on AWSAmazon Web Services
 
How to Build a Winning Pitch Deck
How to Build a Winning Pitch DeckHow to Build a Winning Pitch Deck
How to Build a Winning Pitch DeckAmazon Web Services
 
Building a web application without servers
Building a web application without serversBuilding a web application without servers
Building a web application without serversAmazon Web Services
 
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...Amazon Web Services
 
Introduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container ServiceIntroduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container ServiceAmazon Web Services
 

More from Amazon Web Services (20)

Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
Come costruire servizi di Forecasting sfruttando algoritmi di ML e deep learn...
 
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
Big Data per le Startup: come creare applicazioni Big Data in modalità Server...
 
Esegui pod serverless con Amazon EKS e AWS Fargate
Esegui pod serverless con Amazon EKS e AWS FargateEsegui pod serverless con Amazon EKS e AWS Fargate
Esegui pod serverless con Amazon EKS e AWS Fargate
 
Costruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWSCostruire Applicazioni Moderne con AWS
Costruire Applicazioni Moderne con AWS
 
Come spendere fino al 90% in meno con i container e le istanze spot
Come spendere fino al 90% in meno con i container e le istanze spot Come spendere fino al 90% in meno con i container e le istanze spot
Come spendere fino al 90% in meno con i container e le istanze spot
 
Open banking as a service
Open banking as a serviceOpen banking as a service
Open banking as a service
 
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
Rendi unica l’offerta della tua startup sul mercato con i servizi Machine Lea...
 
OpsWorks Configuration Management: automatizza la gestione e i deployment del...
OpsWorks Configuration Management: automatizza la gestione e i deployment del...OpsWorks Configuration Management: automatizza la gestione e i deployment del...
OpsWorks Configuration Management: automatizza la gestione e i deployment del...
 
Microsoft Active Directory su AWS per supportare i tuoi Windows Workloads
Microsoft Active Directory su AWS per supportare i tuoi Windows WorkloadsMicrosoft Active Directory su AWS per supportare i tuoi Windows Workloads
Microsoft Active Directory su AWS per supportare i tuoi Windows Workloads
 
Computer Vision con AWS
Computer Vision con AWSComputer Vision con AWS
Computer Vision con AWS
 
Database Oracle e VMware Cloud on AWS i miti da sfatare
Database Oracle e VMware Cloud on AWS i miti da sfatareDatabase Oracle e VMware Cloud on AWS i miti da sfatare
Database Oracle e VMware Cloud on AWS i miti da sfatare
 
Crea la tua prima serverless ledger-based app con QLDB e NodeJS
Crea la tua prima serverless ledger-based app con QLDB e NodeJSCrea la tua prima serverless ledger-based app con QLDB e NodeJS
Crea la tua prima serverless ledger-based app con QLDB e NodeJS
 
API moderne real-time per applicazioni mobili e web
API moderne real-time per applicazioni mobili e webAPI moderne real-time per applicazioni mobili e web
API moderne real-time per applicazioni mobili e web
 
Database Oracle e VMware Cloud™ on AWS: i miti da sfatare
Database Oracle e VMware Cloud™ on AWS: i miti da sfatareDatabase Oracle e VMware Cloud™ on AWS: i miti da sfatare
Database Oracle e VMware Cloud™ on AWS: i miti da sfatare
 
Tools for building your MVP on AWS
Tools for building your MVP on AWSTools for building your MVP on AWS
Tools for building your MVP on AWS
 
How to Build a Winning Pitch Deck
How to Build a Winning Pitch DeckHow to Build a Winning Pitch Deck
How to Build a Winning Pitch Deck
 
Building a web application without servers
Building a web application without serversBuilding a web application without servers
Building a web application without servers
 
Fundraising Essentials
Fundraising EssentialsFundraising Essentials
Fundraising Essentials
 
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
AWS_HK_StartupDay_Building Interactive websites while automating for efficien...
 
Introduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container ServiceIntroduzione a Amazon Elastic Container Service
Introduzione a Amazon Elastic Container Service
 

Build Deep Learning Applications Using Apache MXNet, Featuring Workday (AIM407-R) - AWS re:Invent 2018

  • 1.
  • 2. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Information Extraction for Enterprise Documents A I M 4 0 7 Cyrus M Vahid Principal Evangelist, AWS AI Labs cyrusmv@amazon.com Vivek Srivastava ML Product Manager Workday Henry Zhang ML Engineer Workday
  • 3. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Agenda Apache MXNet and why we use it Amazon SageMaker and distributed training Amazon SageMaker and DevOps style deep learning Business Case Framework Model and Algorithms Results Productization Key Learnings
  • 4. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 5. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Computational Dependency/Graph • 𝑧 = 𝑥 ⋅ 𝑦 • 𝑘 = 𝑎 ⋅ 𝑏 • 𝑡 = 𝜆𝑧 + 𝑘 x y 𝑧 x 𝜆 𝑢 x a x b k 𝑡 + 1 1 2 3
  • 6. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Computational Dependency/Graph net = mx.sym.Variable('data') net = mx.sym.FullyConnected(net, name='fc1', num_hidden=64) net = mx.sym.Activation(net, name='relu1', act_type="relu") net = mx.sym.FullyConnected(net, name='fc2', num_hidden=10) net = mx.sym.SoftmaxOutput(net, name='softmax')
  • 7. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Computational Dependency/Graph net = mx.sym.Variable('data') net = mx.sym.FullyConnected(net, name='fc1', num_hidden=64) net = mx.sym.Activation(net, name='relu1', act_type="relu") net = mx.sym.FullyConnected(net, name='fc2', num_hidden=10) net = mx.sym.SoftmaxOutput(net, name='softmax')
  • 8. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Training import logging logging.getLogger().setLevel(logging.DEBUG) # logging to stdout # create a trainable module on compute context mlp_model = mx.mod.Module(symbol=mlp, context=ctx) mlp_model.fit(train_iter, eval_data=val_iter, optimizer='sgd', optimizer_params={'learning_rate':0.1}, eval_metric='acc', batch_end_callback = mx.callback.Speedometer(batch_size, 100), num_epoch=10)
  • 9. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Computational Dependency/Graph net = mx.sym.Variable('data') net = mx.sym.FullyConnected(net, name='fc1', num_hidden=64) net = mx.sym.Activation(net, name='relu1', act_type="relu") net = mx.sym.FullyConnected(net, name='fc2', num_hidden=10) net = mx.sym.SoftmaxOutput(net, name='softmax')
  • 10. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 11. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Training Efficiency – 92% https://mxnet.incubator.apache.org/tutorials/vision/large_scale_classification.html
  • 12. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. End-to-End Machine Learning Platform Zero setup Flexible Model Training Pay by the second $ Amazon SageMaker Build, train, and deploy machine learning models at scale
  • 13. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Amazon SageMaker and Distributed Training • Faster training through SageMaker Streaming for Custom Algorithms. • Boilerplate code for your algorithms to train over a cluster. PCA Bemchmark if len(hosts) == 1: kvstore = 'device' if num_gpus > 0 else 'local’ else: kvstore = 'dist_device_sync' if num_gpus > 0 else 'dist_sync’ trainer = gluon.Trainer(net.collect_params(), 'sgd’, {'learning_rate': learning_rate, 'momentum': momentum}, kvstore=kvstore)
  • 14. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Training code • Matrix Factorization • Regression • Principal Component Analysis • K-Means Clustering • Gradient Boosted Trees • And More! Amazon provided Algorithms Bring Your Own Script (IM builds the Container) Bring Your Own Algorithm (You build the Container) Fetch Training data Save Model Artifacts Fully managed – Secured – Amazon ECR Save Inference Image IM Estimators in Apache Spark CPU GPU HPO Distributed Training
  • 15. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Automatic Model Tuning Training code • Factorization Machine • Regression/classification • Principal Component Analysis • K-Means Clustering • XGBoost • DeepAR • And More SageMaker built-in Algorithms Bring Your Own Script (prebuilt containers) Bring Your Own Algorithm Fetch Training data Save Model Artifacts Fully managed – Secured– Automatic Model Tuning
  • 16. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 17. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Evolution of DL Frameworks
  • 18. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Why Gluon Simple, Easy-to- Understand Code Flexible, Imperative Structure Dynamic Graphs High Performance
  • 19. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Gluon Code – Network Definition net = gluon.nn.HybridSequential() with net.name_scope(): net.add(gluon.nn.Dense(units=64, activation='relu')) net.add(gluon.nn.Dense(units=10)) softmax_cross_entropy = gluon.loss.SoftmaxCrossEntropyLoss() net.initialize(mx.init.Xavier(magnitude=2.24), ctx=ctx, force_reinit=True) trainer = gluon.Trainer(net.collect_params(), 'sgd', {'learning_rate': 0.02})
  • 20. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Gluon Code – Training smoothing_constant = .01 for e in range(10): cumulative_loss = 0 for i, (data, label) in enumerate(train_data): data = data.as_in_context(model_ctx).reshape((-1, 784)) label = label.as_in_context(model_ctx) with autograd.record(): output = net(data) loss = softmax_cross_entropy(output, label) loss.backward() trainer.step(data.shape[0])
  • 21. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. What’s New – GluonCV • A Deep Learning Toolkit for Computer Vision • Features: • Training scripts that reproduces SOTA results reported in latest papers • A large set of pre-trained models • Carefully designed APIs and easy-to-understand implementations • Community support
  • 22. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. What’s New - GluonNLP • A deep learning toolkit for natural language processing • Features: • Training scripts to reproduce SOTA results reported in research papers. • Pre-trained models for common NLP tasks. • Carefully designed APIs that greatly reduce the implementation complexity. • Community support.
  • 23. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. What’s New – Keras Backend Instance Type GPUs Batch Size Keras-MXNet (img/sec) Keras- TensorFlow (img/sec) C5.18X Large 0 32df 13 4 P3.8X Large 1 32 194 184 P3.8X Large 4 128 764 393 P3.16X Large 8 256 1068 261 Instance Type GPUs Batch Size Keras-MXNet (img/sec) Keras- TensorFlow (img/sec) C5.X Large 0 32 5.79 3.27 C5.8X Large 0 32 27.9 18.2 https://github.com/awslabs/keras-apache-mxnet/tree/master/benchmark
  • 24. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. What’s New - Sockeye • A seq2seq toolkit based on Apache MXNet. • Features: • Beam search inference • Easy assembling of multiple models • Residual connections between RNN layers (Wu et al., 2016) [Deep LSTM with parallelism] • Lexical biasing of output layer predictions (Arthur et al., 2016) [Low Frequency Words] • Modeling coverage (Tu et al., 2016) [Keeping attention history to reduce over and under translation] • Context gating (Tu et al., 2017) [Improving adequacy of translation by controlling rations of source and target context] • Cross-entropy label smoothing (e.g., Pereyra et al., 2017) • Layer normalization (Ba et al, 2016) [improving training time] • Multiple supported attention mechanisms [dot, mlp, bilinear, multihead-dot, encoder last state, location] • Multiple model architectures (Encoder-Decoder Wu et al., 2016, Convolutional Gehring et al, 2017, Transformer Vaswani et al, 2017,)
  • 25. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 26. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Inference Efficiency - TensorRT Model Name Relative TRT Sppedup HArdware Resnet 101 1.99x Titan V Resnet 50 1.76x Titan V Resnet 18 1.54x Jetson TX1 cifar_resnext29_16x64d 1.26x Titan V cifar_resnet20_v2 1.21x Titan V Resnet 18 1.8x Titan V Alexnet 1.4x Titan V https://cwiki.apache.org/confluence/display/MXNET/How+to+use+MXNet-TensorRT+integration
  • 27. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Inference Efficiency – NNVM https://aws.amazon.com/blogs/machine-learning/introducing-nnvm-compiler-a-new-open-end-to-end-compiler-for-ai-frameworks/
  • 28. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 29. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Portability - NNVM https://aws.amazon.com/blogs/machine-learning/introducing-nnvm-compiler-a-new-open-end-to-end-compiler-for-ai-frameworks/
  • 30. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Portability - ONNX Model Parameters Hyper Parameters
  • 31. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Deployment with Amazon SageMaker I ML Hosting Service Amazon ECR 30 50 10 10 ProductionVariant Model Artifacts Inference Image Model versions Versions of the same inference code saved in inference containers. Prod is the primary one, 50% of the traffic must be served there! One-Click! EndpointConfiguration Inference Endpoint Amazon Provided Algorithms Amazon SageMaker InstanceType: c3.4xlarge InitialInstanceCount: 3 ModelName: prod VariantName: primary InitialVariantWeight: 50
  • 32. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Recap
  • 33. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Agenda Business Case Framework Model and Algorithms Results Productization Key Learnings
  • 34. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 35. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Expense Report Automation
  • 36. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. AP Invoice Automation Machine Learning Machine Learning
  • 37. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 38. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Document scanning CV and NLU to extract and understand text in varied documents and pre-populate the appropriate fields
  • 39. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 40. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Framework Requirement • Fast and agile prototyping and experiments • Option to productize via Python and JVM language by exporting models • Best accuracy • Ideally - active support
  • 41. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Framework Choice (2017) • Agile Python prototyping, production JVM language support • MXNet • Imperative • ~50% faster training and inference speed + less memory consumption vs Tensorflow • Better accuracy for our problem • Sockeye (S2S) • Easy to use • Quick support • Not Expected: Strong and active support from AWS MXNet team
  • 42. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. MXNet vs. Tensorflow Accuracy (2017) MXNet Tensorflow Perfect Matching Error Rate 11.6% 21.36% Edit Distance Error Rate 4% 7.75% • Note: For our problem and models
  • 43. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 44. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Problems Where are the texts? What do the texts mean? What are the texts?
  • 45. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Where are the texts? • Customized model based on You Only Look Once (YOLO) network • Added angle of tilt • Used ResNet instead of CNN
  • 46. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. What are the texts? • Customized model based on WaveNet (Originally intended for speech) • ResNet + Dilated Convolution • Able to scan forward and backward
  • 47. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. What do the texts mean? (Rules-based system) • Rules-Based System • Looks into surrounding context • Fuzzy Regex
  • 48. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Encoder-Decoder Seq2seq Approach • Encoder-decoder model with attention mechanism • Character embedding to word embedding • Vocabulary built within training data • Future work to use embeddings from FastText/Word2vec
  • 49. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Benefit of Encoder- Decoder Approach • This approach is good at inferring from context • However, it requires you to have seen the merchant in training data
  • 50. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. CNN-based Box Classification Approach • CNN-based model to detect probability of each box • Leverages word2vec word embedding • This approach considers both language context and approximate location of each field
  • 51. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Ensemble Strategy • Easier fields – rules based only • More difficult fields – perplexity value decision • Combined two 0.60 perfect match model to get 72% on test set with only 40k training data
  • 52. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Transferred Learning • With limited real document data, we trained on generated data.
  • 53. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Sample Business Documents
  • 54. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 55. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Results • We have competitive results that’s at least on par, and on average performs better when comparing to major public cloud providers on Vision part of the model. • Test Set metrics: ~75–85% accuracy • Production metrics: ~95% accuracy
  • 56. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 57. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Productization
  • 58. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.
  • 59. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Key Learnings • Value in building models specific for the product/use case • Prototyping/Experimenting mindset vs. production mindset • Importance of working with the framework team for challenges
  • 60. Thank you! © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved. Vivek Srivastava Henry Zhang Cyrus M Vahid cyrusmv@amazon.com
  • 61. © 2018, Amazon Web Services, Inc. or its affiliates. All rights reserved.