sparse recovery and fourier sampling
play

Sparse Recovery and Fourier Sampling Eric Price MIT Eric Price - PowerPoint PPT Presentation

Jul 05, 2023 •277 likes •2.15k views

Sparse Recovery and Fourier Sampling Eric Price MIT Eric Price (MIT) Sparse Recovery and Fourier Sampling 1 / 37 The Fourier Transform Conversion between time and frequency domains Frequency Domain Time Domain Fourier Transform

  1. Discrete Fourier Transform (DFT) Definition Given x ∈ C n , compute Fourier transform � x : x i = 1 � ω − ij x j ω = e τ i / n � for n j Eric Price (MIT) Sparse Recovery and Fourier Sampling 12 / 37

  2. Discrete Fourier Transform (DFT) Definition Given x ∈ C n , compute Fourier transform � x : x i = 1 � ω − ij x j ω = e τ i / n � for n j (where τ is the circle constant 6 . 283 ... ) Eric Price (MIT) Sparse Recovery and Fourier Sampling 12 / 37

  3. Discrete Fourier Transform (DFT) Definition Given x ∈ C n , compute Fourier transform � x : x i = 1 � ω − ij x j ω = e τ i / n � for n j F ij = ω − ij / n � x = F x for (where τ is the circle constant 6 . 283 ... ) Eric Price (MIT) Sparse Recovery and Fourier Sampling 12 / 37

  4. Discrete Fourier Transform (DFT) Definition Given x ∈ C n , compute Fourier transform � x : x i = 1 � ω − ij x j ω = e τ i / n � for n j F ij = ω − ij / n � x = F x for Inverse transform almost identical: (where τ is the circle constant 6 . 283 ... ) Eric Price (MIT) Sparse Recovery and Fourier Sampling 12 / 37

  5. Discrete Fourier Transform (DFT) Definition Given x ∈ C n , compute Fourier transform � x : x i = 1 � ω − ij x j ω = e τ i / n � for n j F ij = ω − ij / n � x = F x for Inverse transform almost identical: � ω ij � x i = x j j ◮ ω → ω − 1 , scale (where τ is the circle constant 6 . 283 ... ) Eric Price (MIT) Sparse Recovery and Fourier Sampling 12 / 37

  6. Discrete Fourier Transform (DFT) Definition Given x ∈ C n , compute Fourier transform � x : x i = 1 � ω − ij x j ω = e τ i / n � for n j F ij = ω − ij / n � x = F x for Inverse transform almost identical: � ω ij � x i = x j j ◮ ω → ω − 1 , scale Lots of nice properties (where τ is the circle constant 6 . 283 ... ) Eric Price (MIT) Sparse Recovery and Fourier Sampling 12 / 37

  7. Discrete Fourier Transform (DFT) Definition Given x ∈ C n , compute Fourier transform � x : x i = 1 � ω − ij x j ω = e τ i / n � for n j F ij = ω − ij / n � x = F x for Inverse transform almost identical: � ω ij � x i = x j j ◮ ω → ω − 1 , scale Lots of nice properties ◮ Convolution ← → Multiplication (where τ is the circle constant 6 . 283 ... ) Eric Price (MIT) Sparse Recovery and Fourier Sampling 12 / 37

  8. Algorithm Simpler case: � x is exactly k -sparse. Eric Price (MIT) Sparse Recovery and Fourier Sampling 13 / 37

  9. Algorithm Simpler case: � x is exactly k -sparse. Theorem We can compute � x in O ( k log n ) expected time. Eric Price (MIT) Sparse Recovery and Fourier Sampling 13 / 37

  10. Algorithm Simpler case: � x is exactly k -sparse. Theorem We can compute � x in O ( k log n ) expected time. Still kind of hard. Eric Price (MIT) Sparse Recovery and Fourier Sampling 13 / 37

  11. Algorithm Simpler case: � x is exactly k -sparse. Theorem We can compute � x in O ( k log n ) expected time. Still kind of hard. Simplest case: � x is exactly 1-sparse. Eric Price (MIT) Sparse Recovery and Fourier Sampling 13 / 37

  12. Algorithm Simpler case: � x is exactly k -sparse. Theorem We can compute � x in O ( k log n ) expected time. Still kind of hard. Simplest case: � x is exactly 1-sparse. Lemma We can compute a 1 -sparse � x in O ( 1 ) time. Eric Price (MIT) Sparse Recovery and Fourier Sampling 13 / 37

  13. Algorithm for k = 1 Lemma a We can compute a 1 -sparse � x in O ( 1 ) time. � x : � a if i = t � x i = t 0 otherwise Eric Price (MIT) Sparse Recovery and Fourier Sampling 14 / 37

  14. Algorithm for k = 1 Lemma a We can compute a 1 -sparse � x in O ( 1 ) time. � x : � a if i = t � x i = t 0 otherwise Then x = ( a , a ω t , a ω 2 t , a ω 3 t , . . . , a ω ( n − 1 ) t ) . Eric Price (MIT) Sparse Recovery and Fourier Sampling 14 / 37

  15. Algorithm for k = 1 Lemma a We can compute a 1 -sparse � x in O ( 1 ) time. � x : � a if i = t � x i = t 0 otherwise Then x = ( a , a ω t , a ω 2 t , a ω 3 t , . . . , a ω ( n − 1 ) t ) . x 0 = a Eric Price (MIT) Sparse Recovery and Fourier Sampling 14 / 37

  16. Algorithm for k = 1 Lemma a We can compute a 1 -sparse � x in O ( 1 ) time. � x : � a if i = t � x i = t 0 otherwise Then x = ( a , a ω t , a ω 2 t , a ω 3 t , . . . , a ω ( n − 1 ) t ) . x 1 = a ω t x 0 = a Eric Price (MIT) Sparse Recovery and Fourier Sampling 14 / 37

  17. Algorithm for k = 1 Lemma a We can compute a 1 -sparse � x in O ( 1 ) time. � x : � a if i = t � x i = t 0 otherwise Then x = ( a , a ω t , a ω 2 t , a ω 3 t , . . . , a ω ( n − 1 ) t ) . x 1 = a ω t x 0 = a x 1 / x 0 = ω t = ⇒ t . Eric Price (MIT) Sparse Recovery and Fourier Sampling 14 / 37

  18. Algorithm for k = 1 Lemma a We can compute a 1 -sparse � x in O ( 1 ) time. � x : � a if i = t � x i = t 0 otherwise Then x = ( a , a ω t , a ω 2 t , a ω 3 t , . . . , a ω ( n − 1 ) t ) . x 1 = a ω t x 0 = a x 1 / x 0 = ω t = ⇒ t . � Eric Price (MIT) Sparse Recovery and Fourier Sampling 14 / 37

  19. Algorithm for k = 1 Lemma a We can compute a 1 -sparse � x in O ( 1 ) time. � x : � a if i = t � x i = t 0 otherwise Then x = ( a , a ω t , a ω 2 t , a ω 3 t , . . . , a ω ( n − 1 ) t ) . x 1 = a ω t x 0 = a x 1 / x 0 = ω t = ⇒ t . � (Related to OFDM, Prony’s method, matrix pencil.) Eric Price (MIT) Sparse Recovery and Fourier Sampling 14 / 37

  20. Algorithm for general k Reduce general k to k = 1. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  21. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  22. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  23. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  24. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  25. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  26. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  27. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  28. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  29. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  30. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. Random permutation 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  31. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. Random permutation 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  32. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. Random permutation 1-sparse recovery 1-sparse recovery x x ′ � Permute Filters O ( k ) 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  33. Algorithm for general k Reduce general k to k = 1. “Filters”: partition frequencies into O ( k ) buckets. ◮ Sample from time domain of each bucket with O ( log n ) overhead. ◮ Recovered by k = 1 algorithm Most frequencies alone in bucket. Random permutation 1-sparse recovery 1-sparse recovery x x ′ � Permute Filters O ( k ) 1-sparse recovery Recovers most of � x : 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Eric Price (MIT) Sparse Recovery and Fourier Sampling 15 / 37

  34. Overall outline � x Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  35. Overall outline x ′ x − � � Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  36. Overall outline x ′ x − � � Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Repeat, k → k / 2 → k / 4 → · · · Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  37. Overall outline x ′ x − � � Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Repeat, k → k / 2 → k / 4 → · · · Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  38. Overall outline x ′ x − � � Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Repeat, k → k / 2 → k / 4 → · · · Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  39. Overall outline x ′ x − � � Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Repeat, k → k / 2 → k / 4 → · · · Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  40. Overall outline x ′ x − � � Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Repeat, k → k / 2 → k / 4 → · · · Theorem We can compute � x in O ( k log n ) expected time. Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  41. Overall outline x ′ x − � � Partial k -sparse recovery 1-sparse recovery 1-sparse recovery x x ′ � O ( k ) Permute Filters 1-sparse recovery 1-sparse recovery Lemma (Partial sparse recovery) x ′ such that In O ( k log n ) expected time, we can compute an estimate � x ′ is k / 2 -sparse. x − � � Repeat, k → k / 2 → k / 4 → · · · Theorem We can compute � x in O ( k log n ) expected time. Eric Price (MIT) Sparse Recovery and Fourier Sampling 16 / 37

  42. How can you isolate frequencies? Time Frequency n -dimensional DFT: O ( n log n ) x → � x × ∗ = = Eric Price (MIT) Sparse Recovery and Fourier Sampling 17 / 37

  43. How can you isolate frequencies? Time Frequency n -dimensional DFT: O ( n log n ) x → � x × ∗ = = Eric Price (MIT) Sparse Recovery and Fourier Sampling 17 / 37

  44. How can you isolate frequencies? Time Frequency n -dimensional DFT: O ( n log n ) x → � x × ∗ n -dimensional DFT of first k terms: O ( n log n ) x · rect → � x ∗ sinc. = = Eric Price (MIT) Sparse Recovery and Fourier Sampling 17 / 37

  45. How can you isolate frequencies? Time Frequency n -dimensional DFT: O ( n log n ) x → � x × ∗ n -dimensional DFT of first k terms: O ( n log n ) x · rect → � x ∗ sinc. = = Eric Price (MIT) Sparse Recovery and Fourier Sampling 17 / 37

  46. How can you isolate frequencies? Time Frequency n -dimensional DFT: O ( n log n ) x → � x × ∗ n -dimensional DFT of first k terms: O ( n log n ) x · rect → � x ∗ sinc. = = k -dimensional DFT of first k terms: O ( B log B ) alias ( x · rect ) → subsample ( � x ∗ sinc ) . Eric Price (MIT) Sparse Recovery and Fourier Sampling 17 / 37

  47. How can you isolate frequencies? Time Frequency n -dimensional DFT: O ( n log n ) x → � x × ∗ n -dimensional DFT of first k terms: O ( n log n ) x · rect → � x ∗ sinc. = = k -dimensional DFT of first k terms: O ( B log B ) alias ( x · rect ) → subsample ( � x ∗ sinc ) . Eric Price (MIT) Sparse Recovery and Fourier Sampling 17 / 37

  48. The issue We want to isolate frequencies. Frequency Eric Price (MIT) Sparse Recovery and Fourier Sampling 18 / 37

  49. The issue We want to isolate frequencies. Frequency The sinc filter “leaks”. Contamination from other buckets. Eric Price (MIT) Sparse Recovery and Fourier Sampling 18 / 37

  50. The issue We want to isolate frequencies. Frequency The sinc filter “leaks”. Contamination from other buckets. We introduce a better filter: (Gaussian / prolate spheroidal sequence) convolved with rectangle. Eric Price (MIT) Sparse Recovery and Fourier Sampling 18 / 37

  51. Algorithm for exactly sparse signals Original signal x Goal ˆ x Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  52. Algorithm for exactly sparse signals Computed F · x Filtered signal c F ∗ c x Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  53. Algorithm for exactly sparse signals F · x aliased to k terms Filtered signal c F ∗ c x Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  54. Algorithm for exactly sparse signals F · x aliased to k terms Computed samples of c F ∗ c x Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  55. Algorithm for exactly sparse signals F · x aliased to k terms Computed samples of c F ∗ c x Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  56. Algorithm for exactly sparse signals F · x aliased to k terms Knowledge about ˆ x Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  57. Algorithm for exactly sparse signals F · x aliased to k terms Knowledge about ˆ x Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  58. Algorithm for exactly sparse signals F · x aliased to k terms Knowledge about ˆ x Lemma If t is isolated in its bucket and in the “super-pass” region, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Eric Price (MIT) Sparse Recovery and Fourier Sampling 19 / 37

  59. Algorithm for exactly sparse signals Lemma For most t, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Eric Price (MIT) Sparse Recovery and Fourier Sampling 20 / 37

  60. Algorithm for exactly sparse signals Lemma For most t, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Time-shift x by one and repeat: b ′ = � x t ω t . Divide to get b ′ / b = ω t Eric Price (MIT) Sparse Recovery and Fourier Sampling 20 / 37

  61. Algorithm for exactly sparse signals Lemma For most t, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Time-shift x by one and repeat: b ′ = � x t ω t . Divide to get b ′ / b = ω t = ⇒ can compute t . Eric Price (MIT) Sparse Recovery and Fourier Sampling 20 / 37

  62. Algorithm for exactly sparse signals Lemma For most t, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Time-shift x by one and repeat: b ′ = � x t ω t . Divide to get b ′ / b = ω t = ⇒ can compute t . ◮ Just like our 1-sparse recovery algorithm, x 1 / x 0 = ω t . Eric Price (MIT) Sparse Recovery and Fourier Sampling 20 / 37

  63. Algorithm for exactly sparse signals Lemma For most t, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Time-shift x by one and repeat: b ′ = � x t ω t . Divide to get b ′ / b = ω t = ⇒ can compute t . ◮ Just like our 1-sparse recovery algorithm, x 1 / x 0 = ω t . x ′ such that � x ′ is k / 2-sparse. Gives partial sparse recovery: � x − � 1-sparse recovery 1-sparse recovery x ′ x � Permute Filters O ( k ) 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 20 / 37

  64. Algorithm for exactly sparse signals Lemma For most t, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Time-shift x by one and repeat: b ′ = � x t ω t . Divide to get b ′ / b = ω t = ⇒ can compute t . ◮ Just like our 1-sparse recovery algorithm, x 1 / x 0 = ω t . x ′ such that � x ′ is k / 2-sparse. Gives partial sparse recovery: � x − � 1-sparse recovery 1-sparse recovery x ′ x � Permute Filters O ( k ) 1-sparse recovery 1-sparse recovery Repeat k → k / 2 → k / 4 → · · · Eric Price (MIT) Sparse Recovery and Fourier Sampling 20 / 37

  65. Algorithm for exactly sparse signals Lemma For most t, the value b we compute for its bucket satisfies b = � x t . Computing the b for all O ( k ) buckets takes O ( k log n ) time. Time-shift x by one and repeat: b ′ = � x t ω t . Divide to get b ′ / b = ω t = ⇒ can compute t . ◮ Just like our 1-sparse recovery algorithm, x 1 / x 0 = ω t . x ′ such that � x ′ is k / 2-sparse. Gives partial sparse recovery: � x − � 1-sparse recovery 1-sparse recovery x ′ x � Permute Filters O ( k ) 1-sparse recovery 1-sparse recovery Repeat k → k / 2 → k / 4 → · · · O ( k log n ) time sparse Fourier transform. � Eric Price (MIT) Sparse Recovery and Fourier Sampling 20 / 37

  66. Algorithm for approximately sparse signals Eric Price (MIT) Sparse Recovery and Fourier Sampling 21 / 37

  67. Algorithm for approximately sparse signals What changes with noise? Eric Price (MIT) Sparse Recovery and Fourier Sampling 21 / 37

  68. Algorithm for approximately sparse signals What changes with noise? Identical architecture: Partial sparse recovery 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Eric Price (MIT) Sparse Recovery and Fourier Sampling 21 / 37

  69. Algorithm for approximately sparse signals What changes with noise? Identical architecture: Partial sparse recovery 1-sparse recovery 1-sparse recovery x x ′ O ( k ) � Permute Filters 1-sparse recovery 1-sparse recovery Just requires robust 1-sparse recovery. Eric Price (MIT) Sparse Recovery and Fourier Sampling 21 / 37

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Tutorial: Sparse Recovery Using Sparse Matrices Piotr Indyk MIT Problem Formulation

Tutorial: Sparse Recovery Using Sparse Matrices Piotr Indyk MIT Problem Formulation

Tutorial: Sparse Recovery Using Sparse Matrices Piotr Indyk MIT Problem Formulation (approximation theory, learning Fourier coeffs, linear sketching, finite rate of innovation, compressed sensing...) Setup: Data/signal in n-dimensional

418 views • 25 slides
Compressive sensing principles and iterative sparse recovery for inverse and ill-posed problems

Compressive sensing principles and iterative sparse recovery for inverse and ill-posed problems

Compressive sensing principles and iterative sparse recovery for inverse and ill-posed problems Evelyn Herrholz Gerd Teschke November 9, 2010 Abstract In this paper we shall be concerned with compressive sampling strategies and sparse

417 views • 30 slides
New Constructions of RIP Matrices with Fast Multiplication and Fewer Rows aka, sparse recovery

New Constructions of RIP Matrices with Fast Multiplication and Fewer Rows aka, sparse recovery

New Constructions of RIP Matrices with Fast Multiplication and Fewer Rows aka, sparse recovery from Fourier -like measurements with applications to fast Johnson-Lindenstrauss transforms , etc. Jelani Nelson, Eric Price, and Mary Wootters February

394 views • 24 slides
Sparse Fourier Transforms Eric Price UT Austin Eric Price Sparse Fourier Transforms 1 / 36

Sparse Fourier Transforms Eric Price UT Austin Eric Price Sparse Fourier Transforms 1 / 36

Sparse Fourier Transforms Eric Price UT Austin Eric Price Sparse Fourier Transforms 1 / 36 The Fourier Transform Conversion between time and frequency domains Frequency Domain Time Domain Fourier Transform Displacement of Air Concert A

1.94k views • 181 slides
Sparse sampling sampling design in design in Sparse population PK/PD studies studies

Sparse sampling sampling design in design in Sparse population PK/PD studies studies

Sparse sampling sampling design in design in Sparse population PK/PD studies studies population PK/PD Sylvie Retout & France Mentr INSERM U738, Universit Paris 7, UFR de Mdecine, Hpital Bichat, Paris. 1 Outline Outline

674 views • 21 slides
Two added structures in sparse recovery: nonnegativity and disjointedness Simon Foucart

Two added structures in sparse recovery: nonnegativity and disjointedness Simon Foucart

Two added structures in sparse recovery: nonnegativity and disjointedness Simon Foucart University of Georgia Semester Program on High-Dimensional Approximation ICERM 7 October 2014 Part I: Nonnegative Sparse Recovery (joint work with

957 views • 69 slides
(Nearly) Sample Optimal Sparse Fourier Transform Piotr Indyk 1 Michael Kapralov 1 Eric Price 2 1

(Nearly) Sample Optimal Sparse Fourier Transform Piotr Indyk 1 Michael Kapralov 1 Eric Price 2 1

(Nearly) Sample Optimal Sparse Fourier Transform Piotr Indyk 1 Michael Kapralov 1 Eric Price 2 1 MIT 2 MIT IBM Almaden UT Austin SODA14 Indyk, Kapralov, Price (MIT, IBM Almaden) (Nearly) Optimal Fourier Sampling SODA14 1 / 28

560 views • 41 slides
= = h ( t ) ( t kT ) that has the Fourier

= = h ( t ) ( t kT ) that has the Fourier

= x ( t ) cos( 0 t ) Sampling of at sampling rate s 1. Multiplying impulse train. We have the following cosine signal with period T . 0 ( ) 2 = = x c ( t ) cos t cos( t ) 0 T 0 2 =

246 views • 3 slides
Optimized design and analysis of Optimized design and analysis of sparse-sampling fMRI

Optimized design and analysis of Optimized design and analysis of sparse-sampling fMRI

Optimized design and analysis of Optimized design and analysis of sparse-sampling fMRI experiments sparse-sampling fMRI experiments of speech and hearing of speech and hearing Tyler Perrachione John Gabrieli Satrajit Ghosh Overview

401 views • 15 slides
Weak Fourier-Schur Sampling, the Hidden Subgroup Problem & the Quantum Collision Problem

Weak Fourier-Schur Sampling, the Hidden Subgroup Problem & the Quantum Collision Problem

Weak Fourier-Schur Sampling, the Hidden Subgroup Problem & the Quantum Collision Problem Pawel M. Wocjan School of Electrical Engineering and Computer Science University of Central Florida Orlando wocjan@cs.ucf.edu Weak Fourier-Schur

264 views • 25 slides
3. Lecture Fourier Transformation Sampling Some slides taken from Digital Image Processing: An

3. Lecture Fourier Transformation Sampling Some slides taken from Digital Image Processing: An

3. Lecture Fourier Transformation Sampling Some slides taken from Digital Image Processing: An Algorithmic Introduction using Java, Wilhelm Burger and Mark James Burge Separability The 2D DFT can be separated in two 1D DFT's: M 1 N

559 views • 44 slides
Bounds on Sparse Recovery with Additional Structures Abbas Kazemipour University of Maryland.

Bounds on Sparse Recovery with Additional Structures Abbas Kazemipour University of Maryland.

Bounds on Sparse Recovery with Additional Structures Abbas Kazemipour University of Maryland. College Park kaazemi@umd.edu March 23, 2015 Abbas Kazemipour (UMD) Sparse Recovery with Side Info. March 23, 2015 1 / 19 Overview 1 Restricted

599 views • 42 slides
Simple and Practical Algorithm for the Sparse Fourier Transform Haitham Hassanieh Piotr Indyk

Simple and Practical Algorithm for the Sparse Fourier Transform Haitham Hassanieh Piotr Indyk

Simple and Practical Algorithm for the Sparse Fourier Transform Haitham Hassanieh Piotr Indyk Dina Katabi Eric Price MIT 2012-01-19 Hassanieh, Indyk, Katabi, and Price (MIT) Simple and Practical Algorithm for the Sparse Fourier Transform

488 views • 26 slides
1 Sampling and Aliasing Image Processing pipeline Artifacts due to undersampling or poor

1 Sampling and Aliasing Image Processing pipeline Artifacts due to undersampling or poor

Outline Foundations of Computer Graphics Basic ideas of sampling, reconstruction, aliasing (Fall 2012) Signal processing and Fourier analysis CS 184, Lectures 19: Sampling and Reconstruction Implementation of digital filters

487 views • 10 slides
Nearly Optimal Sparse Fourier Transform Haitham Hassanieh Piotr Indyk Dina Katabi Eric Price

Nearly Optimal Sparse Fourier Transform Haitham Hassanieh Piotr Indyk Dina Katabi Eric Price

Nearly Optimal Sparse Fourier Transform Haitham Hassanieh Piotr Indyk Dina Katabi Eric Price MIT 2012-04-27 Hassanieh, Indyk, Katabi, and Price (MIT) Nearly Optimal Sparse Fourier Transform 2012-04-27 1 / 33 Outline Introduction 1

1.43k views • 117 slides
Tensor learning approach to sparse QMC sampling of two-particle Greens function in DMFT

Tensor learning approach to sparse QMC sampling of two-particle Greens function in DMFT

Tensor learning approach to sparse QMC sampling of two-particle Greens function in DMFT Hiroshi SHINAOKA Collaborators N. Chikano, J. Otsuki, M. Ohzeki, K. Yoshimi, K. Haule, M. Wallerberger, J. Li, E. Gull, D. Geffroy, J. Kune Matsubara

337 views • 17 slides
The ODE labs are numbered 1, 4, 5, 7 and 8. (written by Suzanne Lenhart and John Workman) Lab 1:

The ODE labs are numbered 1, 4, 5, 7 and 8. (written by Suzanne Lenhart and John Workman) Lab 1:

The ODE labs are numbered 1, 4, 5, 7 and 8. (written by Suzanne Lenhart and John Workman) Lab 1: Introductory Example For each problem, there is a user-friendly interface that will guide you through. Each lab consists of two different programs,

336 views • 16 slides
Gaussian Measurements Canyi Lu 1 , Jiashi Feng 2 , Zhouchen Lin 3 , Shuicheng Yan 2 1 Carnegie

Gaussian Measurements Canyi Lu 1 , Jiashi Feng 2 , Zhouchen Lin 3 , Shuicheng Yan 2 1 Carnegie

Exact Low Tubal Rank Tensor Recovery ry from Gaussian Measurements Canyi Lu 1 , Jiashi Feng 2 , Zhouchen Lin 3 , Shuicheng Yan 2 1 Carnegie Mellon University 2 National University of Singapore 3 Peking University Low dimensional structures in

436 views • 20 slides
Signal Recovery from Random Measurements Joel A. Tropp Anna C. Gilbert

Signal Recovery from Random Measurements Joel A. Tropp Anna C. Gilbert

Signal Recovery from Random Measurements Joel A. Tropp Anna C. Gilbert {jtropp|annacg}@umich.edu Department of Mathematics The University of Michigan 1 The Signal Recovery Problem Let s be an m -sparse signal in R d , for example

554 views • 29 slides
Supplemental Slides Second Quarter 2019 Earnings July 31, 2019 Forward-Looking Statements This

Supplemental Slides Second Quarter 2019 Earnings July 31, 2019 Forward-Looking Statements This

Supplemental Slides Second Quarter 2019 Earnings July 31, 2019 Forward-Looking Statements This presentation contains "forward-looking statements" within the meaning of federal securities laws. Investors and prospective investors should

191 views • 17 slides
Federal Partners in Disaster Recovery: DOE, FEMA, and SBA 2019 CDBG-DR Problem Solving Clinic

Federal Partners in Disaster Recovery: DOE, FEMA, and SBA 2019 CDBG-DR Problem Solving Clinic

Federal Partners in Disaster Recovery: DOE, FEMA, and SBA 2019 CDBG-DR Problem Solving Clinic Kansas City Overland Park | J u l y 3 0 A u g u s t 1 , 2 0 1 9 2019 CDBG-DR Problem Solving Clinic 1 Agenda & Speakers Agenda:

458 views • 44 slides
Examining Vegeta,on Recovery Time A5er a Small Scale Disaster

Examining Vegeta,on Recovery Time A5er a Small Scale Disaster

Examining Vegeta,on Recovery Time A5er a Small Scale Disaster using MODIS Data and the OSDC Zac Flamig (University of Oklahoma) Gilbert Warren

73 views • 5 slides
A Natural Language Approach to Automated Cryptanalysis of Two-time Pads Joshua Mason Kathryn

A Natural Language Approach to Automated Cryptanalysis of Two-time Pads Joshua Mason Kathryn

A Natural Language Approach to Automated Cryptanalysis of Two-time Pads Joshua Mason Kathryn Watkins Jason Eisner Adam Stubblefield The Two Time Pad Problem Attack at Dawn doQvYcSWIPyXaC Attack at Dawn doQvYcSWIPyXaC Take the

1.15k views • 99 slides
Building DR Solutions with VMware Site Recovery Manager March 2019 John A. Davis Virtualization

Building DR Solutions with VMware Site Recovery Manager March 2019 John A. Davis Virtualization

Building DR Solutions with VMware Site Recovery Manager March 2019 John A. Davis Virtualization Architect, @johnnyadavis, vLoreBlog.com Problems Addressed Lets focus on these issues today Many organizations have components of a Disaster

675 views • 30 slides

More recommend