Erte
Pan
Wireless Eng. Group
Advisor: Dr. Han
Department of Electrical and Computer Engineering
University of Houston, Houston, TX.
Deep Belief Nets and Restricted Boltzmann Machines
Graphical Model
hidden
i
j
visible
hidden
i
j
visible
Undirected graphical model
:
links have
no directional
significance
inference(infer the states of unobserved variables) is easy
learning(adjust weights between variables to make the network
more likely to generate the observed data) and generating
processes are tricky
Directed graphical model
:
links have
a particular directionality indicated by
arrows
inference is difficult
learning and generating processes are simple
Generative model: graphical model captures the causal process
by which the observed data was generated, so it is also called
generative model.
Boltzmann Machine Model
Boltzmann Machine
:
one input layer and one hidden layer
typically binary states for every unit
stochastic (vs. deterministic)
recurrent (vs. feed

forward)
generative model (vs. discriminative): estimate the
distribution of observations(say p(image)), while
traditional discriminative networks only estimate the
labels(say p(
labelimage
))
defined
Energy
of the network and
Probability
of a
unit’s state(scalar T is referred to as the “temperature”):
Boltzmann Machine
Restricted Boltzmann
Machine Model
Restricted Boltzmann Machine
Restricted
Boltzmann Machine
:
a bipartite graph: no
intralayer
connections, feed

forward
RBM does not have T factor, the rest are the same as
BM
one important feature of RBM is that the visible units
and hidden units are conditionally independent, which
will lead to a beautiful result later on:
Stochastic Search
Why BM?
Different optimization criteria in traditional networks and RBM for
optimization purpose
:
Traditional: Error criterion. BP method strictly goes along the gradient
descent direction. Any direction that enlarge error is NOT acceptable. Easy to
get stuck in local minima.
BM: associate the network with “Energy”. Simulated Annealing enables the
energy to grow under certain probability.
Simulated Annealing
Simulated Annealing
for BM:
1.
Create initial solution
S (global states of the network)
Initialize temperature
T
>>1
2.
Repeat
until
T =T

lower

bound
Repeat
until
thermal equilibrium is reached at
the current T
•
Generate a random transition from
S
to
S
’
•
Let
E
=
E
(
S
’
)
E
(
S
)
•
if
E
< 0
then
S
=
S
’
•
else if
exp
[
E
/
T
] >
rand
(0,1)
then
S
=
S
’
Reduce temperature
T
according to the
cooling schedule
3.
Return
S
This term allows “thermal
disturbance” which facilitate
finding global minimum
Restricted Boltzmann Machine
Two characters to define a Restricted
Boltzmann Machine:
states
of all the units: obtained through probability distribution.
weights
of the network: obtained through training(Contrastive Divergence).
As mentioned before, the objective of RBM is to estimate the distribution of input data.
And this goal is fully determined by the weights, given the input.
Energy
defined for the RBM:
Restricted Boltzmann Machine
Distribution of visible layer
of the RBM(Boltzmann
Distribution):
Z is the partition function defined as the sum of over all possible
configurations of {
v,h
}
Probability that unit i is on(binary state 1):
σ
(.) is
the logistic/sigmoid function
Restricted Boltzmann Machine
the <.>
0
denotes an average
w.r.t
. the data distribution
the gradient is then computed as:
the <.>
∞
denotes an average
w.r.t
. the model distribution
Given
i.i.d
. samples , The objective
is to maximize the
average log

likelihood:
Training for RBM:
Maximum Likelihood learning
the probability over a
vector x with
parameter W(weights) is:
Restricted Boltzmann Machine
Then the update of weights, W, can be computed as:
the <.>
0
term can be computed using the input samples
the <.>
∞
term can be solved by MCMC but very slow and suffering from large
variance of estimated gradient
Solution:
Contrastive Divergence
maximizing the log probability of the data is the same as minimizing the
KL
divergence
, (
define CD to be: where n is the small
number that we run the MC
use
CDn
multiplied by learning rate as the update of the weights
Note:
this update direction is NOT the gradient of ANY function, yet it is
successful in application…
Restricted Boltzmann Machine
Summarized algorithm for training RBM:
1)
take a training sample v, compute the probabilities of the hidden units and sample a hidden
activation vector h
from this probability distribution.
2)
compute the expectation
of
vh
and call this the
positive gradient
. (clamped phase, or positive
phase)
3)
From
h , sample a reconstruction
v’ of the visible units, then resample the hidden activations
h’
from this.
4)
Compute the
expectation of
v’h
’
and call this the
negative gradient
.(free phase, or negative phase)
5)
Let the weight update
Wij
to
be the positive gradient minus the negative gradient, times some
learning rate.
In RBM, the previous equations then become(for calculating a particular
weight between two units):
these two equations are obtained by substituting the energy function into
the learning rule.
General Deep Belief Nets
Problem with DBNs:
Since DBNs are directed graph model,
g
iven input data, the
posterior of hidden units is intractable due to the “explaining
away” effect.
Solution:
Complementary Priors
to ensure the posterior of
hidden units are under the independent constraints.
truck hits house
earthquake
house
jumps
20
20

20

10

10
General Deep Belief Nets
Explaining Away Effect
p(1,1)=.0001
p(1,0)=.4999
p(0,1)=.4999
p(0,0)=.0001
posterior
Explaining Away Effect
Brief summary for explaining away effect:
Given the observations, the posterior of associated hidden
variables are actually NOT independent(the probability
that one hidden variable is on or off influences the states
of others), even though the hidden variables are assumed
to be independent in their priors.
The reason is that we have non

independence in the
likelihood term:
Posterior(non

indep
) = prior(
indep
.) * likelihood (non

indep
.)
Eliminate
Explaining Away by
Complementary Priors
Add
extra hidden layers to create CP that has opposite
correlations with the likelihood term, so (when likelihood
is multiplied by the prior),
posterior will
become factorial
v
1
h
1
v
0
h
0
v
2
h
2
etc.
+
+
+
+
Complementary Priors
Definition of
Complementary Priors
:
Consider observations x and hidden variables y, for a given likelihood
function P(
xy
), the priors of y, P(y) is called the complementary priors of
P(
xy
), provided that P(
x,y
)=P(
xy
) P(y) leads to the posteriors P(
yx
) that
exactly
factorises
.
Infinite directed model with tied weights & Complementary Priors & Gibbs
sampling:
Recall that the RBMs have the property
The definition of energy function of RBM makes it proper model that has two
sets of conditional independencies(complementary priors for both v and h)
Since we need to estimate the distribution of data, P(v), we can perform
Gibbs sampling alternatively from P(
v,h
) for infinite times. This procedure is
analogous to unroll the single RBM into
infinite directed stacks of RBMs
with
tied weights(due to “complementary priors”) where each RBM takes input
from the hidden layer of the lower level RBM.
DBNs based on RBMs
DBNs based on stacks of RBMs:
The top two hidden layers form an
undirected associative
memory(regarded as a shorthand for
infinite stacks) and the remained
hidden layers form a directed acyclic
graph.
h2
data
h1
h3
RBM
RBM
RBM
The red arrows are NOT part of the
generative model. They are just for
inference purpose
Training Deep Belief Nets
Previous discussion gives an intuition of training stacks of RBMs one layer at
a time.
This greedy learning algorithm is proved to be efficient in the sense of
expected variance by Hinton.
First, learn all the weights tied.
Learn as a single
RBM
Training Deep Belief Nets
Then freeze bottom layer and
relearn all the other layers.
Then freeze bottom two layers
and relearn all the other layers.
Learn as a single
RBM
Learn as a single
RBM
Fine

tuning Deep Belief Nets
Each time we learn a new layer, the inference at the lower layers will
become incorrect, but the
variational
bound on the log probability of the
data improves, proved by Hinton.
Since the inference at lower layers becomes incorrect, Hinton uses a fine

tuning procedure to adjust the weights, called wake

sleep algorithm.
Wake

sleep algorithm
:
wake phase: do a down

top pass, sample
h using the recognition weight based
on input v for each RBM, and then adjust
the generative weight by the RBM learning
rule.
sleep phase: do a top

down pass, start by
a random state of h at the top layer and
generate v. Then the recognition weights
are modified.
h2
data
h1
h3
RBM
RBM
RBM
Deep Belief Nets
Analogs for wake

sleep algorithm:
wake phase: if the reality is different with the imagination, then modify the
generative weights to make what is imagined as close as the reality.
sleep phase: if the illusions produced by the concepts learned during wake
phase are different with the concepts, then modify the recognition weight to
make the illusions as close as the concepts.
Questions on DBNs:
training vector vs. training set(Patch Training)
How to perform unsupervised classification?
Performances from DBNs
A: 2

D coded representation of hand

written database MNIST by PCA
B: 2

D coded representation of MNIST by DBNs
Results produced by Hinton etc.
Performances from DBNs
A
B
A: 2

D coded representation of documents retrieval data by LSA
B: 2

D coded representation of the same data by DBNs
Results produced by Hinton etc.
Convolutional DBNs
Limitations of DBNs
:
unable to process high dimensional data(DBNs transform 2D images into vectors
and then input them into the networks, thus certain spatial information is lost)
even if using vectors as the input instead, DBNs are unable to be scaled up
properly for real image sizes. They are only suitable for small images
directly extending the DBNs to fit the high dimensional data suffers from
inefficient computation(millions of weights to estimate)
Advantages of CDBNs
:
feature detectors are shared through all locations in an image, therefore they
form the
convolution kernels
and reduce computation
max

pooling
: shrink the representation to be
translation

invariant
and reduce
computation
Architecture of CDBNs
Energy term and Probability are
defined similarly to RBM:
All units are 2D
binary
images,
within one unit of detection layer,
the weights/convolutional kernels
are shared, leading to the
convolution operation :
CDBNs
Training of CDBNs is done by optimizing the networks’ energy via
sparsity
regularization
(imposed by max

pooling):
This yields a similar updating strategy for the weights and biases as the
Contrastive Divergence.
The sparse constraints also give rise to a simple inference of the network:
where
Performance of CDBNs
Results produced by Andrew Y. Ng etc.
Hierarchical representations of Caltech

101 object classification database by
CDBNs. Top: first layer CDBN output. Bottom: second layer CDBN output.
References
Review:
Learning deep architectures for AI, Y.
Bengio
2009
Foundations:
A fast learning algorithm for deep belief nets, Hinton 2006
Reducing the dimensionality of data with neural networks, Hinton 2006
A practical guide to training restricted Boltzmann machines, Hinton 2010
On contrastive divergence learning, Hinton, 2005
On the convergence property of contrastive divergence,
Tieleman
, 2010
Training products of experts by minimizing contrastive divergence, Hinton, 2002
Learning multiple layers of representation, Hinton, 2007
Applications:
Sparse deep belief net model for visual area V2, H. Lee 2008
Convolutional deep belief network for scalable unsupervised learning of hierarchical
representations, H. Lee 2009
Unsupervised learning of invariant feature hierarchies with applications to object
recognition, Y.
LeCun
2007
Comments 0
Log in to post a comment