2. Outline
• Thales & Big Data
• On the difficulty of Sequence Learning
• Deep Learning for Sequence Learning
• Spark implementation of Deep Learning
• Use cases
– Predictive maintenance
– NLP
3. Thales & Big Data
Thales systems produce a huge quantity of data
Transportation systems (ticketing, supervision, …)
Security (radar traces, network logs, …)
Satellite (photos, videos, …)
which is often
Massive
Heterogeneous
Extremely dynamic
and where understanding the dynamics of the monitored phenomena
is mandatory Sequence Learning
4. What is sequence learning ?
Sequence learning refers to a set of ML tasks where a model has
to either deal with sequences as input, produce sequences as
output or both
Goal : Understand the dynamic of a sequence to
– Classify
– Predict
– Model
Typical applications
– Text
• Classify texts (sentiment analysis)
• Generate textual description of images (image captioning)
– Video
• Video classification
– Speech
• Speech to text
5. How is it typically handled ?
Taking into account the dynamic is difficult
– Often people do not bother
• E.g. text analysis using bag of word (one hot encoding)
– Problem for certain tasks such as sentiment classification (order of the words is important)
– Or use popular statistical approaches
• (Hidden) Markov model for prediction (and classification)
– Shortterm dependency (order 1) : 𝑃(𝑋$ = 𝑥 (𝑋$'( = 𝑥$'(,… , 𝑋$', = 𝑥$',)⁄ ) = 𝑃(𝑋$ = 𝑥$ 𝑋$'( = 𝑥$'()⁄
• Autoregressive approaches for time series forecasting
The chair is red 1 0 1 1 0 0 0 0
The cat is on a chair
The cat is young 1 1 0 0 1 1 0 0
1 1 1 0 0 1 1 1
The is chair red young cat on a
6. Link with artificial neural network ?
Artificial neural network is a set of statistical models inspired from the brain
– Transforms the input by applying at each layer (non linear) functions
– More layers equals more capabilities (≥ 2 hidden layers : Deep Learning)
• From manual features building to feature learning
Set of transformation and activation operations
– Affine : 𝒀 = 𝑾 𝒕
𝑿 + 𝒃, sigmoid activation :
𝟏
𝟏8𝐞𝐱𝐩 ('𝑿)
, tanh activation : 𝒀 = 𝐭𝐚𝐧𝐡 ( 𝑿)
• Only affine + activation layers = multi layer perceptron (available in Spark ML since 1.5.0)
– Convolutional : Apply a spatial convolution on the 1D/2D input (signal, image, …) : 𝐘 = 𝒄𝒐𝒏𝒗 𝑿, 𝑾 + 𝒃
• Learns spatial features used for classification (images) , prediction
– Recurrent : Introduces a recurrent part to learn dependencies between observations (features related to
the dynamic)
Objective
– Find the best weights W to minimize the difference between the predicted output and the desired one
(using back-propagation algorithm)
input
hidden
layers
output
7. Able to cope with varying size sequences either at the input or at the output
Recurrent Neural Network basics
One to many
(fixedsize input,
sequence output)
e.g. Image captioning
Many to many
(sequence input to sequence
output)
e.g. Speech to text
Many to one
(sequence input to fixedsize
output)
e.g. Text classification
Artificial neural networks with one or more recurrent layers
Classical neural network Recurrent neural network
𝒀 𝒌'𝟑 𝒀 𝒌'𝟐 𝒀 𝒌'𝟏 𝒀 𝒌
𝒀 𝒌
𝑿 𝒌'𝟑 𝑿 𝒌'𝟐 𝑿 𝒌'𝟏 𝑿 𝒌
𝒀 𝒌 = 𝒇(𝑾 𝒕 𝑿 𝒌 + 𝑯𝒀 𝒌'𝟏)
𝑿 𝒌𝑿
𝒀 𝒌 = 𝒇(𝑾 𝒕 𝑿 𝒌)
𝒀
Unrolled through time
𝒀 𝒌'𝟑 𝒀 𝒌'𝟐 𝒀 𝒌'𝟏 𝒀 𝒌
𝑿
𝒀 𝒌'𝟑 𝒀 𝒌'𝟐 𝒀 𝒌'𝟏 𝒀 𝒌
𝑿 𝒌'𝟑 𝑿 𝒌'𝟐 𝑿 𝒌'𝟏 𝑿 𝒌
𝑿 𝒌'𝟑 𝑿 𝒌'𝟐 𝑿 𝒌'𝟏 𝑿 𝒌
𝒀
8. On the difficulty of training recurrent networks
RNNs are (were) known to be difficult to learn
– More weights and more computational steps
• More computationally expensive (accelerator needed for matrix ops : Blas or GPU)
• More data needed to converge (scalability over Big Data architectures : Spark)
– Theano, Tensor Flow, Caffe do not have distributed versions
– Unable to learn long range dependencies (Graves & Al 2014)
• At a given time t, RNN does not remember the observations before 𝑋J',
⇒ New RNN architectures with memory preservation (more context)
𝑍$ = 𝑓 𝑊N
O
𝑋$ + 𝐻N 𝑌$'(
𝑅$ = 𝑓(𝑊S
O
𝑋$ + 𝐻S 𝑌$'()
𝐻T$ = tanh(𝑊YJZ[
O
𝑋$ + 𝑈 𝑌$'( o 𝑅$ )
𝑌$ = 1 − 𝑍$ 𝑌$'( + 𝑍$ 𝐻T$
LSTM GRU
9. Recurrent neural networks in Spark
Spark implementation of DL algorithms (data parallel)
– All the needed blocks
• Affine, convolutional, recurrent layers (Simple and GRU)
• Sigmoid, tanh, reLU activations
• SGD, rmsprop, adadelta optimizers
– CPU (and GPU backend)
– Fully compatible with existing DL library in Spark ML
Performance
– On 6 nodes cluster (CPU)
• 5.46 average speedup (some communication overhead)
– About the same speedup as MLP in Spark ML
Driver
Worker 1
Worker 2
Worker 3
Resulting gradients (2)
Model broadcast (1)
10. Use case 1 : predictive maintenance (1)
Context
– Thales and its clients build systems in different domains
• Transportation (ticketing, controlling)
• Defense (radar)
• Satellites
– Need better and more accurate maintenance services
• From planned maintenance (every x days) to an alert maintenance
• From expert detection to automatic failure prediction
• From whole subsystem changes to more localized reparations
Goal
– Detect early signs of a (sub)system failure using data coming
from sensors monitoring the health of a system (HUMS)
11. Use case 1 : predictive maintenance (2)
Example on a real system
– 20 sensors (20 values every 5 minutes), label (failure or not)
– Take 3 hours of data and predict the probability of failure in the next hour (fully
customizable)
Learning using MLLIB
12. Use case 1 : predictive maintenance (3)
Recurrent net learning
Impact of recurrent nets
– Logistic regression
• 70% detection with 70% accuracy
– Recurrent Neural Network
• 85% detection with 75% accuracy
13. Use case 2 : Sentiment analysis (1)
Context
– Social network analysis application developed at Thales (Twitter, Facebook,
blogs, forums)
• Analyze both the content of the texts and the relations (texts, actors)
– Multiple (big data) analysis
• Actor community detection
• Text clustering (themes)
• …
Focus on
– Sentiment analysis on the collected texts
• Classify texts based on their sentiment
14. Use case 2 : Sentiment analysis (2)
Learning dataset
– Sentiment140 + Kaggle challenge (1.5M labeled tweets)
– 50% positives, 50% negatives
Compare Bag of words + classifier approaches (Naïve Bayes, SVM, logistic
regression) versus RNN