2. A Flywheel For Building Organizations
Users Better Products
More Data Better Analytics
More
3. Machine Learning will empower
the business
Eventually everything connects
IoT/Mobile/Cameras/APIs/Datalake
Large amounts of data will be
processed in real-time
4. At Amazon, we’ve been making investments
in ML for the last 20 years…
Search &
Discovery
Fulfilment
& Logistics
Existing
Products
New
Initiatives
8. Recognize the discrete
anatomical structures of
the eye, such as the
retina, cornea or optic
nerve to identify diabetic
retinopathy
(leading cause of
blindness)
9. “The future is here.
It’s just not widely distributed yet.”
William Gibson
American-Canadian fiction writer
10. Put machine learning in the
hands of every developer
and data scientist
ML @ AWS:
O U R M I S S I O N
12. AWS has shipped over 100 major
new features and new services for
machine learning since re:Invent
2017
A m a z o n R e k o g n i t i o n I m a g e a n d V i d e o l a u n c h i n U S E a s t ( O h i o ) A d v a n c e d I n d e x i n g S u p p o r t A m a z o n L e x r e g i o n s u p p o r t A m a z o n L e x : D e f a u l t R e s p o n s e s
A m a z o n L e x : D T M F S u p p o r t
A m a z o n P o l l y W o r d P r e s s p l u g i n i n t e g r a t e s T r a n s l a t e A m a z o n R e k o g n i t i o n I m a g e a n d V i d e o l a u n c h i n T o k y o a n d S y d n e y A m a z o n T r a n s c r i b e - G A A m a z o n T r a n s l a t e - G A A n o m a l y D e t e c t i o n ( R a n d o m C u t F o r e s t ) A l g o r i t h m
A u t o m a t e r o l e c r e a t i o n d u r i n g s e t u p A u t o m a t i c M o d e l T u n i n g A u t o s c a l i n g c o n s o l e A W S D e e p L e n s d e v i c e s a v a i l a b l e f o r s a l e B l a z i n g T e x t A l g o r i t h m B u i l t - i n A l g o r i t h m s P i p e M o d e S u p p o r t C a f f e t o M X N e t c o d e t r a n s l a t o r
C h a i n e r p r e - b u i l t c o n t a i n e r C l o u d F o r m a t i o n s u p p o r t C l o u d T r a i l i n t e g r a t i o n f o r a u d i t l o g s C U D A 9 . 0 a n d 9 . 1 + c u D N N 7 . 1 . 4 + G P U D r i v e r 3 9 0 . 4 6 + N C C L 2 . 2 . 1 3 u p g r a d e C u s t o m e r V P C s u p p o r t f o r t r a i n i n g a n d h o s t i n g D C A : A W S D e e p L e a r n i n g A M I i n D C A
D e e p A R a l g o r i t h m D e e p L e a r n i n g A M I i n G o v C l o u d D e e p L e a r n i n g A M I : A d d C h a i n e r , M X N e t 1 . 1 s u p p o r t ( i n c l . B J S ) D e e p L e a r n i n g A M I : E C 2 O p t i m i z e d B i n a r i e s f o r T e n s o r F l o w D e e p L e a r n i n g A M I : O p t i m i z e d C h a i n e r 4 , C N T K 2 . 5
D e e p L e a r n i n g A M I : R e f r e s h f o r E C 2 o p t i m i z e d T e n s o r F l o w 1 . 8 D e e p L e a r n i n g A M I : R e f r e s h w i t h T F 1 . 5 a n d M X N e t 1 . 1 D e e p L e a r n i n g A M I : R e g i o n a l E x p a n s i o n i n 5 p u b l i c r e g i o n s : U S W e s t ( C a l i f o r n i a ) , C a n a d a , S a o P a u l o , L o n d o n , a n d P a r i s
D e e p L e a r n i n g A M I : R e l e a s e l a t e s t D L A M I v e r s i o n i n B J S D L A M I l a u n c h i n B J S a n d F R A , B O M , S I N
D e e p L e a r n i n g A M I : T e n s o r F l o w m u l t i - g p u o p t i m i z a t i o n s u s i n g H o r o v o d , N C C L a n d O p e n M P ID e e p L e a r n i n g A M I : T F 1 . 5 R C 0 , M M S , T F S e r v i n g , T e n s o r B o a r d ( i n c l u d i n g B J S )F P 1 6 s u p p o r t
D L A M I r e f r e s h w i t h M X N e t 1 . 0 , T F f o r C U D A 9 a n d P y T o r c h f o r C U D A 9
D e e p L e n s i n t e g r a t i o n w i t h A m a z o n K i n e s i s v i d e o s t r e a m E x p o r t / I m p o r t A m a z o n L e x b o t s c h e m a
F a c e R e c o g n i t i o n v 3 l a u n c h G D P R C o m p l i a n c e G r a d i e n t C o m p r e s s i o n
H I P A A c o m p l i a n c e
I m p o r t f r o m S a g e M a k e r t o D e e p L e n s I n c r e a s e C h a r a c t e r L i m i t I n t e r n e t - f r e e n o t e b o o k i n s t a n c e s K M S s u p p o r t f o r t r a i n i n g a n d h o s t i n g L i n e a r L e a r n e r I m p r o v e m e n t s - E a r l y S t o p p i n g , N e w L o s s F u n c t i o n s , a n d C l a s s W e i g h t s m l . p 3 . 2 x l a r g e n o t e b o o k i n s t a n c e s
M o d e l O p t i m i z e r M o r e i n s t a n c e t y p e s s u p p o r t M X N e t M o d e l S e r v e r M X N e t M o d e l S e r v e r M X N e t M o d e l S e r v e r v 0 . 2 M X N e t M o d e l S e r v e r v 0 . 3 M X N e t v 1 . 0 r e l e a s e M X N e t v 2 . 0 r e l e a s e n b e x a m p l e s u p p o r t i n S a g e M a k e r n o t e b o o k i n s t a n c e s
N C C L s u p p o r t
N E R , K e y p h r a s e , S e n t i m e n t B a t c h N e w B r e a t h S S M L t a g N e w F r e n c h V o i c e ( f e m a l e ) N e w P h o n a t i o n S S M L t a g N o t e b o o k b o o t s t r a p s c r i p t O N N X v 1 . 0 a n n o u n c e m e n t P C I D S S C o m p l i a n c e P o l l y i n G o v C l o u d P o l l y v o i c e s i n A l e x a S k i l l s
P r i v a t e L i n k s u p p o r t f o r S a g e M a k e r i n f e r e n c i n g A P I s P y T o r c h p r e - b u i l t c o n t a i n e r R e f r e s h D L A M I i n D C A R e g i o n e x p a n s i o n t o F R A R e g i o n e x p a n s i o n t o N R T R e g i o n e x p a n s i o n t o S y d n e y R e k o g n i t i o n H I P A A s u p p o r t R e k o g n i t i o n V i d e o i n A W S G o v C l o u d
S a g e M a k e r r e g i o n e x p a n s i o n t o I C N S t o r e t r a n s c r i p t i o n o u t p u t i n c u s t o m e r S 3 b u c k e t s S u p p o r t f o r K e r a s 2 . 0 S u p p o r t G l u o n m o d e l s T a g - b a s e d a c c e s s c o n t r o l T e n s o r F l o w 1 . 5 , M X N e t 1 . 0 , a n d C U D A 9 S u p p o r t T e n s o r F l o w 1 . 6 a n d M X N e t 1 . 1 C o n t a i n e r s
T e n s o r F l o w 1 . 7 C o n t a i n e r s T e n s o r F l o w 1 . 8 C o n t a i n e r T e n s o r f l o w a n d C a f f e s u p p o r t T e n s o r F l o w a n d M X N e t C o n t a i n e r s - O p e n S o u r c i n g a n d L o c a l M o d e T r a i n i n g J o b c l o n i n g i n c o n s o l e T r a n s c r i b e i n t e g r a t i o n w i t h C l o u d T r a i l T r a n s c r i b e i n t e g r a t i o n w i t h C l o u d W a t c h
A m a z o n P o l l y W o r d P r e s s P l u g i n
T r a n s l a t e - - a d d t o P o l l y W o r d P r e s s p l u g i n T r a n s l a t e - - G D P R c o m p l i a n c e
T r a n s l a t e - - G o v C l o u d T r a n s l a t e - - M o b i l e S D K
X G B o o s t I n s t a n c e W e i g h t s
16. Deep learning-based image recognition service
Search, verify, and organize millions of images
Object and
scene
detection
Facial
analysis
Text in
image
Face search
and match
Celebrity
recognition
Unsafe
image
detection
Amazon Rekognition
17. Face Comparison
Measure the likelihood that faces in two images are of
the same person
• Add face verification to
applications and devices
• Extend physical security
controls
• Provide guest access to
VIP-only facilities
• Verify users for online
exams and polls
18. Amazon Rekognition API
var compareParams = {
SimilarityThreshold: 90,
SourceImage: { ... },
TargetImage: { ... }
}};
rekognition.compareFaces(compareParams, function(err, data) {
if (err) {
console.log(err, err.stack); // an error occurred
} else {
if (data.FaceMatches.length > 0) {
//get item in dynamo
console.log("Similarity: " + data.FaceMatches[0].Similarity);
dynamodb.getItem(paramsItem, function(err, data) { ... }
}
});
21. Deep Learning – Neural Network
Output
Neural Network
Input
Hidden layers
Computing systems inspired by the biological neural networks
which learn to do tasks by considering examples, generally
without task-specific programming
27. … or Apache Spark
through EMR and
the Amazon
SageMaker Spark
SDK...
UX
Use Amazon
SageMaker‘s
hosted Notebook
Instances...
... or Amazon
SageMaker‘s
Console for a point
and click
experience...
... or your own
device (EC2,
laptop, etc.)
33. XGBoost, FM,
Linear, and
Forecasting for
supervised
learning
Kmeans, PCA,
and BlazingText
(Word2Vec) for
clustering and
pre-processing
Image
classification with
convolutional
neural networks
LDA and
NTM for topic
modeling,
seq2seq for
translation
Built-in algorithms
Random Cut
Forest for
anomaly
detection
35. … explore and
refine models in a
single Notebook
Instance
TensorFlow & MXNet containers
… deploy to
production
Sample your
data…
Use the same code
to train on the full
dataset in a cluster
of GPU
instances…
39. Hyperparameter Optimization
Run a large set of training
jobs with varying
hyperparameters...
... and search the
hyperparameter space for
improved accuracy.
41. Amazon Polly: Life-like Speech Service
Converts text
to life-like
speech
53 voices 25 languages Low latency,
real time
Fully managed
42. TEXT
Market grew by > 20%.
WORDSPHONEMES
{
{
{
{
{
ˈtwɛn.ti
pɚ.ˈsɛnt
ˈmɑɹ.kət ˈgɹu baɪ ˈmoʊɹ
ˈðæn
PROSODY CONTOURUNIT SELECTION AND ADAPTATION
TEXT PROCESSING
PROSODY MODIFICATIONSTREAMING
Market grew by more
than
twenty
percent
Speech units
inventory
52. Join us for our first-ever Amazon Web Services Summit
in Ottawa on October 29, 2018
15 sessions featuring various
management and technical
topics
Connect with AWS & our
Canadian public sector partners
in the Solutions Expo
Meet and mingle with other public
sector customers from government,
education, and nonprofits
Register today!
aws.amazon.com/summits/ottawa-public-sector
53. We value your feedback!
Please share your feedback on the
AWS Public Sector Summit survey.
Survey will be emailed 24-48 hours following event.