| 1. |
- Alonso-Fernandez, Fernando, 1978-, et al.
(författare)
-
Log-Likelihood Score Level Fusion for Improved Cross-Sensor Smartphone Periocular Recognition
- 2017
-
Ingår i: 2017 25th European Signal Processing Conference (EUSIPCO). - Piscataway : IEEE. - 9780992862671 ; , s. 281-285
-
Konferensbidrag (refereegranskat)abstract
- The proliferation of cameras and personal devices results in a wide variability of imaging conditions, producing large intra-class variations and a significant performance drop when images from heterogeneous environments are compared. However, many applications require to deal with data from different sources regularly, thus needing to overcome these interoperability problems. Here, we employ fusion of several comparators to improve periocular performance when images from different smartphones are compared. We use a probabilistic fusion framework based on linear logistic regression, in which fused scores tend to be log-likelihood ratios, obtaining a reduction in cross-sensor EER of up to 40% due to the fusion. Our framework also provides an elegant and simple solution to handle signals from different devices, since same-sensor and crosssensor score distributions are aligned and mapped to a common probabilistic domain. This allows the use of Bayes thresholds for optimal decision making, eliminating the need of sensor-specific thresholds, which is essential in operational conditions because the threshold setting critically determines the accuracy of the authentication process in many applications. © EURASIP 2017
|
|
| 2. |
- Brynolfsson, Johan, et al.
(författare)
-
Classification of one-dimensional non-stationary signals using the Wigner-Ville distribution in convolutional neural networks
- 2017
-
Ingår i: 25th European Signal Processing Conference, EUSIPCO 2017. - 9780992862671 ; , s. 326-330
-
Konferensbidrag (refereegranskat)abstract
- In this paper we argue that the Wigner-Ville distribution (WVD), instead of the spectrogram, should be used as basic input into convolutional neural network (CNN) based classification schemes. The WVD has superior resolution and localization as compared to other time-frequency representations. We present a method where a large-size kernel may be learned from the data, to enhance features important for classification. We back up our claims with theory, as well as application on simulated examples and show superior performance as compared to the commonly used spectrogram.
|
|
| 3. |
- Ho, Chung Duc, et al.
(författare)
-
Multi-Way Massive MIMO with Maximum-Ratio Processing and Imperfect CSI
- 2017
-
Ingår i: 2017 25TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO). - : IEEE. - 9780992862671 ; , s. 1704-1708
-
Konferensbidrag (refereegranskat)abstract
- This paper considers a multi-way massive multiple-input multiple-output amplify-and-forward relaying system, where single-antenna users exchange their information-bearing signals with the assistance of one relay station equipped with unconventionally many antennas. The relay first estimates the channels to all users through the pilot signals transmitted from them. Then, the relay uses maximum-ratio processing (i.e. maximum-ratio combining in the multiple-access phase and maximum-ratio transmission in the broadcast phase) to process the signals. A rigorous closed-form expression for the spectral efficiency is derived. We show that by deploying massive antenna arrays at the relay and simple maximum-ratio processing, we can serve many users in the same time-frequency resource, while maintaining a given quality-of-service for each user.
|
|
| 4. |
- Klasson, Marcus, et al.
(författare)
-
Conjugate-prior-regularized multinomial pLSA for collaborative filtering
- 2017
-
Ingår i: 25th European Signal Processing Conference, EUSIPCO 2017. - 9780992862671 ; 2017-January, s. 2501-2505
-
Konferensbidrag (refereegranskat)abstract
- We consider the over-fitting problem for multinomial probabilistic Latent Semantic Analysis (pLSA) in collaborative filtering, using a regularization approach. For big data applications, the computational complexity is at a premium and we, therefore, consider a maximum a posteriori approach based on conjugate priors that ensure that complexity of each step remains the same as compared to the un-regularized method. In the numerical section, we show that the proposed regularization method and training scheme yields an improvement on commonly used data sets, as compared to previously proposed heuristics.
|
|
| 5. |
- Kronvall, Ted, et al.
(författare)
-
Online group-sparse estimation using the covariance fitting criterion
- 2017
-
Ingår i: 25th European Signal Processing Conference, EUSIPCO 2017. - 9780992862671 ; , s. 2101-2105
-
Konferensbidrag (refereegranskat)abstract
- In this paper, we present a time-recursive implementation of a recent hyperparameter-free group-sparse estimation technique. This is achieved by reformulating the original method, termed group-SPICE, as a square-root group-LASSO with a suitable regularization level, for which a time-recursive implementation is derived. Using a proximal gradient step for lowering the computational cost, the proposed method may effectively cope with data sequences consisting of both stationary and non-stationary signals, such as transients, and/or amplitude modulated signals. Numerical examples illustrates the efficacy of the proposed method for both coherent Gaussian dictionaries and for the multi-pitch estimation problem.
|
|
| 6. |
- Nagano-Madsen, Yasuko, 1952, et al.
(författare)
-
Perception and Production of L2 Mandarin Tones by Swedish Learners
- 2017
-
Ingår i: 2017 25th European Signal Processing Conference (EUSIPCO). - : Institute of Electrical and Electronics Engineers (IEEE). - 9780992862671 ; , s. 578-582
-
Konferensbidrag (refereegranskat)abstract
- This study presents the results of perception and production of L2 Mandarin tones in mono- and di-syllabic words by Swedish learners at the beginner level. Although studies of perception and production on Mandarin tones are many, those by speakers of lexical-pitch accent language such as Swedish are still very limited. The result reveals both discrepancy and agreement between perception and production. Swedish learners perform best in discriminating a level tone (T1) from contour tones (T2, T3, T4) both in perception and production. Discrepancy between perception and production was noted for T3. In perception, the identification of T3 was second best after the level tone (T1), but the production of T3 was found to be difficult.
|
|
| 7. |
- Reinhold, Isabella, et al.
(författare)
-
The scaled reassigned spectrogram adapted for detection and localisation of transient signals
- 2017
-
Ingår i: European Signal Processing Conference. - 9780992862671 ; , s. 937-941
-
Konferensbidrag (refereegranskat)abstract
- The reassigned spectrogram can be used to improve the readability of a time-frequency representation of a non-stationary and multi-component signal. However for transient signals the reassignment needs to be adapted in order to achieve good localisation of the signal components. One approach is to scale the reassignment. This paper shows that by adapting the shape of the time window used with the spectrogram and by scaling the reassignment, perfect localisation can be achieved for a transient signal component. It is also shown that without matching the shape of the window, perfect localisation is not achieved. This is used to both identify the time-frequency centres of components in a multi-component signal, and to detect the shapes of the signal components. The scaled reassigned spectrogram with the matching shape window is shown to be able to resolve close components and works well for multi-components signals with noise. An echolocation signal from a beluga whale (Delphinapterus leucas) provides an example of how the method performs on a measured signal.
|
|
| 8. |
- Ribeiro, Eduardo, et al.
(författare)
-
Exploring Deep Learning Image Super-Resolution for Iris Recognition
- 2017
-
Ingår i: 25th European Signal Processing Conference (EUSIPCO 2017). - : Institute of Electrical and Electronics Engineers (IEEE). - 9780992862671 - 9780992862688 - 9781538607510 ; , s. 2176-2180
-
Konferensbidrag (refereegranskat)abstract
- In this work we test the ability of deep learning methods to provide an end-to-end mapping between low and high resolution images applying it to the iris recognition problem. Here, we propose the use of two deep learning single-image super-resolution approaches: Stacked Auto-Encoders (SAE) and Convolutional Neural Networks (CNN) with the most possible lightweight structure to achieve fast speed, preserve local information and reduce artifacts at the same time. We validate the methods with a database of 1.872 near-infrared iris images with quality assessment and recognition experiments showing the superiority of deep learning approaches over the compared algorithms. © EURASIP 2017.
|
|
| 9. |
- Sandsten, Maria, et al.
(författare)
-
Classification of bird song syllables using wigner-ville ambiguity function cross-terms
- 2017
-
Ingår i: 25th European Signal Processing Conference, EUSIPCO 2017. - 9780992862671 ; 2017-January, s. 1739-1743
-
Konferensbidrag (refereegranskat)abstract
- A novel feature extraction method for lowdimensional signal representation is presented. The features are useful for classification of non-stationary multi-component signals with stochastic variation in amplitudes and time-frequency locations. Using a penalty function to suppress the Wigner-Ville ambiguity function auto-terms, the proposed feature set is based on the cross-term doppler- and lag profiles. The investigation considers classification where strong similar components appear in all signals and where the differences between classes are related to weaker components. The approach is evaluated and compared with established methods for simulated data and bird song syllables of the great reed warbler. The results show that the novel feature extraction method gives a better classification than established methods used in bird song analysis.
|
|
| 10. |
- Sundin, Martin, 1983-, et al.
(författare)
-
A Connectedness Constraint for Learning Sparse Graphs
- 2017
-
Ingår i: 2017 25TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO). - : IEEE. - 9780992862671 ; , s. 151-155
-
Konferensbidrag (refereegranskat)abstract
- Graphs are naturally sparse objects that are used to study many problems involving networks, for example, distributed learning and graph signal processing. In some cases, the graph is not given, but must be learned from the problem and available data. Often it is desirable to learn sparse graphs. However, making a graph highly sparse can split the graph into several disconnected components, leading to several separate networks. The main difficulty is that connectedness is often treated as a combinatorial property, making it hard to enforce in e.g. convex optimization problems. In this article, we show how connectedness of undirected graphs can be formulated as an analytical property and can be enforced as a convex constraint. We especially show how the constraint relates to the distributed consensus problem and graph Laplacian learning. Using simulated and real data, we perform experiments to learn sparse and connected graphs from data.
|
|
| 11. |
- Swärd, Johan, et al.
(författare)
-
Designing optimal sampling schemes
- 2017
-
Ingår i: 25th European Signal Processing Conference, EUSIPCO 2017. - 9780992862671 ; 2017-January, s. 912-916
-
Konferensbidrag (refereegranskat)abstract
- In this work, we propose a method for finding an optimal, non-uniform, sampling scheme for a general class of signals in which the signal measurements may be non-linear functions of the parameters to be estimated. Formulated as a convex optimization problem reminiscent of the sensor selection problem, the method determines an optimal sampling scheme given a suitable estimation bound on the parameters of interest. The formulation also allows for putting emphasis on a particular set of parameters of interest by scaling the optimization problem in such a way that the bound to be minimized becomes more sensitive to these parameters. For the case of imprecise a priori knowledge of these parameters, we present a framework for customizing the sampling scheme to take such uncertainty into account. Numerical examples illustrate the efficiency of the proposed scheme.
|
|
| 12. |
- Zaki, Ahmed, et al.
(författare)
-
Distributed Greedy Sparse Learning over Doubly Stochastic Networks
- 2017
-
Ingår i: 2017 25TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO). - : IEEE. - 9780992862671 ; , s. 361-364
-
Konferensbidrag (refereegranskat)abstract
- In this paper, we develop a greedy algorithm for sparse learning over a doubly stochastic network. In the proposed algorithm, nodes of the network perform sparse learning by exchanging their individual intermediate variables. The algorithm is iterative in nature. We provide a restricted isometry property (RIP)-based theoretical guarantee both on the performance of the algorithm and the number of iterations required for convergence. Using simulations, we show that the proposed algorithm provides good performance.
|
|
