Multiscale Geometric Analysis - Rice ECE

Sudocodes
Fast measurement and
reconstruction of sparse
signals
Shriram Sarvotham
Dror Baron
Richard Baraniuk
ECE Department
Rice University
dsp.rice.edu/cs
Motivation: coding of sparse data
• Distributed delivery of data with sparse representation
– Content delivery networks
– Peer to peer networks
– Distributed file storage systems
• E.g. thresholded DCT/wavelet coefficients used in JPEG/JPEG2000
Motivation: coding of sparse data
• Distributed coding of sparse data
– Can we exploit sparsity?
– Efficient?
– Low complexity?
Sparse signal processing
• Signal
has
non-zero coefficients
• Efficient ways to measure and recover ?
• Traditional DSP approach:
– Acquisition: first obtain
measurements
– Then exploit sparsity is in the processing stage
Sparse signal processing
• Signal
has
non-zero coefficients
• Efficient ways to measure and recover ?
• Traditional DSP approach:
– Acquisition: first obtain
measurements
– Then exploit sparsity is in the processing stage
• New compressive sampling (CS) approach:
– Acquisition: obtain just
measurements
– Sparsity is exploited during signal acquisition
[Candes et al; Donoho]
Compressive sampling
• Signal
is
-sparse
• Measure signal via few linear projections
• Enough to encode the signal
measurements
sparse
signal
nonzero
entries
Compressive sampling
• Signal
is
-sparse
• Measure signal via few linear projections
• Random Gaussian measurements
measurements
will work!
sparse
signal
nonzero
entries
CS Miracle: L1 reconstruction
• Find the explanation
with smallest L1 norm
[Candes et al; Donoho]
• If
then
perfect reconstruction w/ high probability
measurements
sparse
signal
nonzero
entries
CS Miracle: L1 reconstruction
• Performance
– Polynomial
complexity reconstruction
– Efficient encoding
measurements
sparse
signal
nonzero
entries
CS Miracle: L1 reconstruction
• But…
is still impractical for many applications
• Reconstruction times:
 N=1,000
 N=10,000
 N=100,000
measurements
t=10 seconds
t=3 hours
t=~months
sparse
signal
nonzero
entries
Why is reconstruction expensive?
measurements
sparse
signal
nonzero
entries
Why is reconstruction expensive?
Culprit: dense, unstructured
measurements
sparse
signal
nonzero
entries
Fast CS reconstruction
• Sudocode matrix (sparse)
• Only 0/1 in
• Each row of
contains
measurements
randomly placed 1’s
sparse
signal
nonzero
entries
Sudocodes
• Sudocode performance
– Efficient encoding
– Sub-linear
complexity reconstruction
• Encouraging numerical results
N=100,000
measurements
K=1,000

t=5.47 seconds
M=5,132
sparse
signal
nonzero
entries
Sudocode reconstruction
Process each
in succession
measurements
Each
can recover some
‘’s
sparse
signal
nonzero
entries
Case 1: Zero measurement
Case 1: Zero measurement
• Resolves all coefficients in the support
• Can resolve up to
coefficients
Case 1: Zero measurement
• Resolves all coefficients in the support
• Can resolve up to
coefficients
• Reduces size of problem
Case 2: #(support set)=1
Case 2: #(support set)=1
• Trivially resolves
Case 2: #(support set)=1
• Trivially resolves
Case 3: Matching measurements
Case 3: Matching measurements
• Matches originate from same support
• Disjoint support
Common support
 coefficients = 0
 contain nonzeros
Common
support
Case 3: Matching measurements
• Matches originate from same support
• Disjoint support
Common support
 coefficients = 0
 contain nonzeros
Trigger of revelations
• Recovery of
can trigger more revelations
Trigger of revelations
• Recovery of
can trigger more revelations
• An avalanche of coefficient revelations
Trigger of revelations
• Recovery of
can trigger more revelations
• An avalanche of coefficient revelations
Sudocode reconstruction
Like sudoku puzzles
measurements
sparse
signal
nonzero
entries
Practical considerations
• Bottleneck: search for matches
– With Binary Search Tree, matches ~
• Re-explain measurements: more data structures
Search for
matches
Design of Sudo measurement matrix
• Choice of
Small
:
Most measurements reveal
Many measurements needed
Large
:
Most measurements uninformative
Many measurements needed
Design of Sudo measurement matrix
• Choice of
Small
:
Most measurements reveal
Many measurements needed
Large
:
Most measurements uninformative
Many measurements needed
Design of Sudo measurement matrix
• Choice of
Small
:
Most measurements reveal
Many measurements needed
• Intuition:
Large
:
Most measurements uninformative
Many measurements needed
so that
Design of Sudo measurement matrix
• Choice of
Small
:
Most measurements reveal
Many measurements needed
• Intuition:
Large
:
Most measurements uninformative
Many measurements needed
so that
Related work
• [Cormode, Muthukrishnan]
– CS scheme based on group testing
–
– Complexity
• [Gilbert et. al.] Chaining Pursuit
– CS scheme based on group testing and iterating
–
– Complexity
– Works best for super-sparse signals
Performance comparison
L1
reconstruction
Chaining
Pursuit
Sudocodes
N=10,000
K=10
M=99
T3 hours
M=5,915
t=0.16 sec
M=461
t=0.14 sec
N=10,000
K=100
M=664
T3 hours
M=90,013
t=2.43 sec
M=803
t=0.37 sec
N=100,000
K=10
M=1,329
Tmonths
M=17,398
t=1.13 sec
M=931
t=1.09 sec
N=10,000
K=1000
M=3,321
T3 hours
M>106
t>30 sec
M=5,132
t=5.47 sec
sparse
signal
measurements
nonzero
entries
Utility in CDNs
• Measurements come from different sources
• Needs enough measurements
Ongoing work
• Statistical dependencies between non-zero
coefficients
• Irregular degree distributions
• Adaptive linear projections
• Noisy measurements
Conclusions
• Sudocodes for CS
– highly efficient
– low complexity
• Key idea: use sparse
• Applications to content distribution
THE END
Compressed sensing webpage:
dsp.rice.edu/cs
Number of measurements
Theorem: With
coefficients
Proof sketch:
, phase 1 requires
to exactly reconstruct
Two phase decoding
is not measured
•
•
•
Phase 1: decode
coefficients
Phase 2: decode remaining
coefficients
Why?
•
Phase 2 saves a factor of
– When most coefficients are decoded,
measurements
Phase 2 measurements and
decoding
•
is non-sparse of dimension
Phase 2 measurements and
decoding
•
is non-sparse of dimension
• Resolve remaining coefficients
by inverting the
sub-matrix of
Phase 2 measurements and
decoding
•
is non-sparse of dimension
• Resolve remaining coefficients
by inverting the
sub-matrix of
• Phase 2 complexity =
• Key: choose
• Phase 2 complexity is
Compressive Sampling
• Signal
is
-sparse in basis/dictionary
– WLOG assume sparsity in space domain
• Measure signal via few linear projections
measurements
sparse
signal
nonzero
entries
• Random sparse measurements
will work!
Signal model
• Signal is strictly sparse
• Every nonzero
~ continuous distribution
 each nonzero coefficient is unique almost surely
measurements
sparse
signal
nonzero
entries