#### DMCA

## Data Compression Techniques for Urban Traffic Data

Citations: | 1 - 1 self |

### Citations

3401 |
Principal Component Analysis
- Jolliffe
- 2005
(Show Context)
Citation Context ... be misleading. If we consider Singular Value Decomposition (SVD) of MG, we get: MG = UGΛGV T G, (6) where columns of UG and VG are left-singular vectors and right-singular vectors of MG respectively =-=[18]-=-, [19] and: MGM T G = UGΛGΛ T GU T G, (7) M T GMG = VGΛ T GΛGV T G. (8) The representations in (7) and (8) are quite similar to (5) and ∑t, baring the scaling factor and mean subtraction. Hence, compr... |

719 | Tensor decompositions and applications
- Kolda, Bader
- 2009
(Show Context)
Citation Context ... p × r, ΘPCA = d × p and CR will be CRPCA = d × p . (9) d × r + p × r C. Data Compression using Tensor Decomposition Tensor based methods have found applications in many fields including Chemometrics =-=[20]-=-, telecommunications [21] and neuroscience [22], [23]. In this section, we propose a tensor based compression method for traffic data. Techniques such as PCA and DCT consider a 2-way representation of... |

460 | A multilinear singular value decomposition
- Lathauwer, Moor, et al.
(Show Context)
Citation Context ...may also correlate strongly [9]. To utilize these daily patterns for compression, we consider a 3-way network profile MG ∈ Rdt×p×w of G. We consider Higher Order Singular Value Decomposition (HO-SVD) =-=[24]-=-–[26] for data compression, using 3-way representation of the network. Speed (km/hr) 70 60 50 40 30 Speed Profile Reconstruction using Σ s Reconstruction using Σ t 20 12:05 AM 6:00 AM 12:00 PM 6:00 PM... |

196 | Fixed point and Bregman iterative methods for matrix rank minimization
- Ma, Goldfarb, et al.
- 2011
(Show Context)
Citation Context ... w × r3 D. Compression Efficiency We use percent root-mean-square distortion (PRD) [23], [27] to compare compression efficiency of proposed methods. PRD is also often referred as Relative error [28], =-=[29]-=-. It is a commonly used performance metric for low-dimensional estimation of matrices and tensors [23], [27]–[29]. PRD can be considered as a measure of energy loss in the reconstructed network profil... |

126 | Blind PARAFAC receivers for DS-CDMA systems
- Sidiropoulos, Giannakis, et al.
- 2000
(Show Context)
Citation Context ...CR will be CRPCA = d × p . (9) d × r + p × r C. Data Compression using Tensor Decomposition Tensor based methods have found applications in many fields including Chemometrics [20], telecommunications =-=[21]-=- and neuroscience [22], [23]. In this section, we propose a tensor based compression method for traffic data. Techniques such as PCA and DCT consider a 2-way representation of the network G. However, ... |

117 | The discrete cosine transform
- Strang
- 1999
(Show Context)
Citation Context ...these definitions, we develop following methods for network data compression. A. Discrete Cosine Transform DCT is the workhorse behind JPEG, which is the industry standard for image compression [16], =-=[17]-=-. The network profile MG, carrying information regarding network G, can also be considered as an image. In this case, speed value z(s j,ti) for link s j at time ti can be thought of as a pixel such th... |

57 |
Fractal Image Compression”,
- Fisher
- 1995
(Show Context)
Citation Context ...ed on these definitions, we develop following methods for network data compression. A. Discrete Cosine Transform DCT is the workhorse behind JPEG, which is the industry standard for image compression =-=[16]-=-, [17]. The network profile MG, carrying information regarding network G, can also be considered as an image. In this case, speed value z(s j,ti) for link s j at time ti can be thought of as a pixel s... |

57 |
Introduction to Linear Algebra. WellesleyCambridge Press, 3rd edition edition,
- Strang
- 1998
(Show Context)
Citation Context ...sleading. If we consider Singular Value Decomposition (SVD) of MG, we get: MG = UGΛGV T G, (6) where columns of UG and VG are left-singular vectors and right-singular vectors of MG respectively [18], =-=[19]-=- and: MGM T G = UGΛGΛ T GU T G, (7) M T GMG = VGΛ T GΛGV T G. (8) The representations in (7) and (8) are quite similar to (5) and ∑t, baring the scaling factor and mean subtraction. Hence, compression... |

50 |
Scalable tensor factorizations for incomplete data
- Acar, Dunlavy, et al.
- 2011
(Show Context)
Citation Context ...× r2 + w × r3 D. Compression Efficiency We use percent root-mean-square distortion (PRD) [23], [27] to compare compression efficiency of proposed methods. PRD is also often referred as Relative error =-=[28]-=-, [29]. It is a commonly used performance metric for low-dimensional estimation of matrices and tensors [23], [27]–[29]. PRD can be considered as a measure of energy loss in the reconstructed network ... |

39 |
Multiway analysis of epilepsy tensors
- Acar, Aykut-Bingol, et al.
- 2007
(Show Context)
Citation Context ... p . (9) d × r + p × r C. Data Compression using Tensor Decomposition Tensor based methods have found applications in many fields including Chemometrics [20], telecommunications [21] and neuroscience =-=[22]-=-, [23]. In this section, we propose a tensor based compression method for traffic data. Techniques such as PCA and DCT consider a 2-way representation of the network G. However, observations taken at ... |

37 | Dimensionality reduction in higher-order signal processing and rank-(r1,r2,...,rn) reduction in multilinear algebra
- Lathauwer, Vandewalle
(Show Context)
Citation Context ...5 0.1 0.05 0 1 2 3 4 5 6 7 8 9 10 Number of Principal Components Fig. 7: Proportion of variance in Σs explained by principal components. Using HO-SVD, we can decompose the network profile MG as [24], =-=[25]-=-: MG = CG ×1 Y (1) ×2 Y (2) ×3 Y (3) , (10) where CG ∈ R dt×p×w is called core tensor and matrix {Y (n) } 3 n=1 contains n-mode singular vectors of MG [25]. F = A ×n B is called the n-mode product bet... |

35 | The weighted diagnostic distortion (WDD) measure for ECG signal compression,
- Zigel, Cohen, et al.
- 2000
(Show Context)
Citation Context ...2 × r3 + dt × r1 + p × r2 + w × r3 and CR will be dt × p × w CRHOSVD = . (12) r1 × r2 × r3 + dt × r1 + p × r2 + w × r3 D. Compression Efficiency We use percent root-mean-square distortion (PRD) [23], =-=[27]-=- to compare compression efficiency of proposed methods. PRD is also often referred as Relative error [28], [29]. It is a commonly used performance metric for low-dimensional estimation of matrices and... |

34 |
Road traffic prediction with spatiotemporal correlations
- Min, Wynter, et al.
- 2007
(Show Context)
Citation Context ...e road networks, these problems can potentially compound into serious bottlenecks [12]. Traffic parameters such as speed, flow and travel time usually exhibit strong spatial and temporal correlations =-=[5]-=-. Moreover, there exist certain temporal and spatial trends in traffic data [9], [13]. These relationships have previously been utilized for applications such as traffic prediction [5], sensing [4], d... |

29 | Bayen A.: ‘A traffic model for velocity data assimilation - Work, Blandin, et al. - 2010 |

22 |
Compressive sensing approach to urban traffic sensing
- Li, Zhu, et al.
- 2011
(Show Context)
Citation Context ...ns [5]. Moreover, there exist certain temporal and spatial trends in traffic data [9], [13]. These relationships have previously been utilized for applications such as traffic prediction [5], sensing =-=[4]-=-, data imputation [14] and even data compression [9], [15]. The previous studies related to traffic data compression have mainly focused on a few intersections [9], [13], [15]. Practical implementatio... |

21 |
Data-driven intelligent transportation systems: A survey,”
- Zhang, Wang, et al.
- 2011
(Show Context)
Citation Context ...compression. I. INTRODUCTION Data Driven Intelligent Transport Systems (D2ITS) have increasingly found greater role in applications such as traffic management, sensing, route guidance, and prediction =-=[1]-=-–[8]. This has been made possible due to availability of large amount of data, collected by sensors such as GPS probes and loop detectors. D2ITS require traffic data with high spatial and temporal res... |

15 |
Online-SVR for short-term traffic flow prediction under typical and atypical traffic conditions,”
- Castro-Neto, Jeong, et al.
- 2009
(Show Context)
Citation Context ...uts strain on resources of data management systems [9]. The huge amount of information also tends to hinder the scalability of many data driven applications dealing with route guidance and prediction =-=[10]-=-, [11]. For D2ITS dealing with large road networks, these problems can potentially compound into serious bottlenecks [12]. Traffic parameters such as speed, flow and travel time usually exhibit strong... |

12 | Stochastic Motion Planning and Applications to Traffic - Lim, Balakrishnan, et al. - 2008 |

10 | Vehicle Route Guidance Systems: Classification and Comparison - Schmitt, Jula |

10 |
Multichannel EEG compression based on matrix and tensor decompositions
- Dauwels, Srinivasan, et al.
(Show Context)
Citation Context ...9) d × r + p × r C. Data Compression using Tensor Decomposition Tensor based methods have found applications in many fields including Chemometrics [20], telecommunications [21] and neuroscience [22], =-=[23]-=-. In this section, we propose a tensor based compression method for traffic data. Techniques such as PCA and DCT consider a 2-way representation of the network G. However, observations taken at the sa... |

9 |
PPCA-based missing data imputation for traffic flow volume: a systematical approach,”
- Qu, Li, et al.
- 2009
(Show Context)
Citation Context ...re exist certain temporal and spatial trends in traffic data [9], [13]. These relationships have previously been utilized for applications such as traffic prediction [5], sensing [4], data imputation =-=[14]-=- and even data compression [9], [15]. The previous studies related to traffic data compression have mainly focused on a few intersections [9], [13], [15]. Practical implementations require data compre... |

8 | Online map-matching based on hidden markov model for real-time traffic sensing applications - Goh, Dauwels, et al. |

8 |
On the best rank-1 and rank-(r 1, r 2,..., rn) approximation of higher-order tensors
- Lathauwer, Moor, et al.
(Show Context)
Citation Context ...lso correlate strongly [9]. To utilize these daily patterns for compression, we consider a 3-way network profile MG ∈ Rdt×p×w of G. We consider Higher Order Singular Value Decomposition (HO-SVD) [24]–=-=[26]-=- for data compression, using 3-way representation of the network. Speed (km/hr) 70 60 50 40 30 Speed Profile Reconstruction using Σ s Reconstruction using Σ t 20 12:05 AM 6:00 AM 12:00 PM 6:00 PM 12:0... |

4 |
Simultaneously prediction of network traffic flow based on PCA-SVR
- Jin, Zhang, et al.
- 2007
(Show Context)
Citation Context ...rain on resources of data management systems [9]. The huge amount of information also tends to hinder the scalability of many data driven applications dealing with route guidance and prediction [10], =-=[11]-=-. For D2ITS dealing with large road networks, these problems can potentially compound into serious bottlenecks [12]. Traffic parameters such as speed, flow and travel time usually exhibit strong spati... |

3 |
A flow volumes data compression approach for traffic network based on principal component analysis
- Li, Jianming, et al.
- 2007
(Show Context)
Citation Context ... have temporal resolution of 5 minutes. While the sensors provide detailed information regarding the state of the network, the large volume of data puts strain on resources of data management systems =-=[9]-=-. The huge amount of information also tends to hinder the scalability of many data driven applications dealing with route guidance and prediction [10], [11]. For D2ITS dealing with large road networks... |

2 | Unsupervised learning based performance analysis of n-support vector regression for speed prediction of a large road network
- Asif, Dauwels, et al.
- 2012
(Show Context)
Citation Context ...ression. I. INTRODUCTION Data Driven Intelligent Transport Systems (D2ITS) have increasingly found greater role in applications such as traffic management, sensing, route guidance, and prediction [1]–=-=[8]-=-. This has been made possible due to availability of large amount of data, collected by sensors such as GPS probes and loop detectors. D2ITS require traffic data with high spatial and temporal resolut... |

2 |
Reliability in stochastic time-dependent traffic networks with correlated link travel times
- Dong, Li, et al.
(Show Context)
Citation Context ...ty of many data driven applications dealing with route guidance and prediction [10], [11]. For D2ITS dealing with large road networks, these problems can potentially compound into serious bottlenecks =-=[12]-=-. Traffic parameters such as speed, flow and travel time usually exhibit strong spatial and temporal correlations [5]. Moreover, there exist certain temporal and spatial trends in traffic data [9], [1... |

2 |
2012) Exploring application perspectives of principal component analysis in predicting dynamic OD matrices
- Djukic, Lint, et al.
(Show Context)
Citation Context ...2]. Traffic parameters such as speed, flow and travel time usually exhibit strong spatial and temporal correlations [5]. Moreover, there exist certain temporal and spatial trends in traffic data [9], =-=[13]-=-. These relationships have previously been utilized for applications such as traffic prediction [5], sensing [4], data imputation [14] and even data compression [9], [15]. The previous studies related... |

1 |
The traffic data compression and decompression for intelligent traffic systems based on two-dimensional wavelet transformation
- Xiao, Xie, et al.
(Show Context)
Citation Context ...l trends in traffic data [9], [13]. These relationships have previously been utilized for applications such as traffic prediction [5], sensing [4], data imputation [14] and even data compression [9], =-=[15]-=-. The previous studies related to traffic data compression have mainly focused on a few intersections [9], [13], [15]. Practical implementations require data compression algorithms for large and diver... |