Restricted Boltzmann Machine
Agenda
❖ History of RBM
❖ Difference between RBM & Autoencoders
❖ Introduction to RBMs
❖ Energy Based Model & Probabilistic Model
❖ Training of RBMs
❖ Example: Collaborative Filtering
Restricted Boltzmann Machine
Restricted Boltzmann Machine
Restricted Boltzmann Machines (RBMs) are neural
networks that belong to Energy Based Models. These
are parameterized generative models representing a
probability distribution.
Introduction to RBMs
edureka!
Restricted Boltzmann MachineAutoencoders
Autoencoders vs RBM
Visible Layer Hidden Layer
Layers of RBMs
edureka!
Visible Layer Hidden Layer
Layers of RBMs
activation f((weight w * input x) + bias b ) = output a
Working of RBMs
X – input
W – weight
a – activation function
w
Visible Layer Hidden Layer
x
+ b = a
x
x
x
+ b = a
+ b = a
edureka!
Reconstruction of RBM
edureka!
RBM: Energy Based Model
RBM: Energy Based Model
Visible Layer
Hidden Layer
RBM: Probabilistic Model
RBM: Probabilistic Model
Partition Function
RBM: Probabilistic Model
Partition Function
RBM: Probabilistic Model
RBM:
Probabilistic
Model
MNIST’s handwritten numerals
Headshots found in Labelled Faces
edureka!
RBM Training
Copyright © 2017, edureka and/or its affiliates. All rights reserved.
1
2
RBM Training
Gibbs Sampling
Contrastive Divergence
Copyright © 2017, edureka and/or its affiliates. All rights reserved.
1
2
RBM Training
Gibbs Sampling
Contrastive Divergence
Copyright © 2017, edureka and/or its affiliates. All rights reserved.
1
2
RBM Training
Gibbs Sampling
Contrastive Divergence
Copyright © 2017, edureka and/or its affiliates. All rights reserved.
1
2
RBM Training
Gibbs Sampling
Contrastive Divergence
RBM: Training to Prediction
Step 1
Step 3
Step 4Train the network on
the data of all users
During inference time
take the training data
of a specific user
Use this data to
obtain the activations
of hidden neurons
Use the hidden neuron
values to get the activations
of input neurons
The new values of input
neurons show the rating
the user would give
Step 2
Step 5
edureka!
RBM: Example
Copyright © 2017, edureka and/or its affiliates. All rights reserved.
1
2
RBM: Collaborative Filtering
Recognizing Latent Factors in
The Data
Using Latent Factors for
Prediction
0
P = 0.838
0
P = 0.354
1
P = 0.838
1 0 0 1 0 -1
Drama Fantasy Science Fiction
Hidden nodes
Visible nodes
Lord of the Rings The Matrix Fight Club Harry Potter Titanic The Hobbit
Copyright © 2017, edureka and/or its affiliates. All rights reserved.
1
2
Recognizing Latent Factors in
The Data
Using Latent Factors for
Prediction
00 1
1
P=0.985
0
P=0.171
0
P=0.290
1
P=0.857
0
P=0.214
-1
P=0.910
Drama Fantasy Science Fiction
Hidden nodes
Visible nodes
Lord of the Rings The Matrix Fight Club Harry Potter Titanic The Hobbit
RBM: Collaborative Filtering
Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutorial | Edureka

Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutorial | Edureka