76 Hits in 2.5 sec

Kernelized Rényi distance for speaker recognition

Balaji Vasan Srinivasan, Ramani Duraiswami, Dmitry N. Zotkin
2010 2010 IEEE International Conference on Acoustics, Speech and Signal Processing  
The proposed entropic distance, the Kernelized Rényi distance (KRD), is formulated in a non-parametric way and the resulting measure is efficiently evaluated in a parallelized fashion on a graphical processor  ...  We propose a new information theoretic approach for computation of the matching score using the Rényi entropy.  ...  We will refer to this measure as the Kernelized Rényi Distance (KRD) and will use it for calculating matching scores.  ... 
doi:10.1109/icassp.2010.5495587 dblp:conf/icassp/SrinivasanDZ10 fatcat:xkikemyfkvh4xd55na34f6vgd4

Efficient subset selection via the kernelized Rényi distance

Balaji Vasan Srinivasan, Ramani Duraiswami
2009 2009 IEEE 12th International Conference on Computer Vision  
Information theoretic measures have been used for sampling the data, retaining its original information content. We propose an efficient Rényi entropy based subset selection algorithm.  ...  In the second application, our subset selection approach is used to replace vector quantization in a standard object recognition algorithm, and improvements are shown.  ...  We derive a distance measure termed the kernelized Rényi distance (KRD) based on the Renyi entropy with α = 2.  ... 
doi:10.1109/iccv.2009.5459395 dblp:conf/iccv/SrinivasanD09 fatcat:65tumgf35nasnfwjvyvqoq5ne4

On the generalization of Shannon entropy for speech recognition

Nicolas Obin, Marco Liuni
2012 2012 IEEE Spoken Language Technology Workshop (SLT)  
This confirms the role of noisiness for speech recognition, and will further be extended to the classification of voice quality for the design of an automatic voice casting system in video games.  ...  The improvement is around 10% in relative error reduction, and is particularly significant for the recognition of noisy speech -i.e., whispery/breathy speech.  ...  In particular, the objective of a voice casting system differs qualitatively from standard speaker recognition applications : in standard speech recognition applications (speaker identification/verification  ... 
doi:10.1109/slt.2012.6424204 dblp:conf/slt/ObinL12 fatcat:dxkgactlkfedfnutgxnwn6l3hu

On the Correlation between Reservoir Metrics and Performance for Time Series Classification under the Influence of Synaptic Plasticity

Joseph Chrol-Cannon, Yaochu Jin, William W. Lytton
2014 PLoS ONE  
We provide a comprehensive empirical study of four metrics; class separation, kernel quality, Lyapunov's exponent and spectral radius.  ...  We find that the two metrics that correlate most strongly with the classification performance are Lyapunov's exponent and kernel quality.  ...  Speaker recognition: A speaker recognition task is a classification problem dealing with mapping time-series audio input data to target speaker labels.  ... 
doi:10.1371/journal.pone.0101792 pmid:25010415 pmcid:PMC4092026 fatcat:73iqx4tr2ndxzhimbjp6x4zv4q

Robust and complex approach of pathological speech signal analysis

Jiri Mekyska, Eva Janousova, Pedro Gomez-Vilda, Zdenek Smekal, Irena Rektorova, Ilona Eliasova, Milena Kostalova, Martina Mrackova, Jesus B. Alonso-Hernandez, Marcos Faundez-Zanuy, Karmele López-de-Ipiña
2015 Neurocomputing  
92 speech features where some of them are already widely used in this field of science and some of them have not been tried yet (they come from different areas of speech signal processing like speech recognition  ...  The other two features BCMD (BiCepstral Module Distance) and BCPD (BiCepstral Phase Distance) are based on distance measures.  ...  kernel, κ (i, j, r) = 1 1 + x[i]−x[j] 2 r , (46) for x[i] − x[j] < r, zero otherwise and triangular kernel, κ (i, j, r) = 1 − |x[i] − x[j]| r , (47) for |x[i] − x[j]| < r, zero otherwise.  ... 
doi:10.1016/j.neucom.2015.02.085 fatcat:nhd5qbpfcjaernkqg42tgxq5pa

Spectral Features for Emotional Speaker Recognition

P Sandhya, V Spoorthy, Shashidhar G. Koolagudi, N.V Sobhana
2020 2020 Third International Conference on Advances in Electronics, Computers and Communications (ICAECC)  
Speaker recognition in an emotive environment is a bit challenging task because of influence of emotions in a speech.  ...  The work aims to identify the speaker in an emotional environment using spectral features and classify using any of the classification techniques and to achieve a high speaker recognition rate.  ...  Speaker Recognition Block Diagram TABLE II ACCURACY% OF SPEAKER RECOGNITION SYSTEM USING GMM TABLE III PERFORMANCE OF VARIOUS SPECTRAL FEATURES FOR SPEAKER RECOGNITION UNDER VARIOUS EMOTIONSTABLE IV  ... 
doi:10.1109/icaecc50550.2020.9339502 fatcat:b7jf5hagczd3zpse2eli4kij34

Shannon Entropy Estimation Based On High-Rate Quantization Theory

Mattias Nilsson, W. Bastiaan Kleijn
2004 Zenodo  
source coding and pattern recognition.  ...  It is well known that density estimation is a complex and delicate problem, which involves issues such as bin-width selection for histogram methods, and kernel-type and number of components for methods  ... 
doi:10.5281/zenodo.38504 fatcat:36bywra5uzeatjfv4srvolvhvi


S Nitesh Singh, K Pratyasha Singha, Pratik Agarwal, Dr. Pranab Das
2020 International Journal of Engineering Applied Sciences and Technology  
Recognition of emotion is always a difficult problem, particularly if the recognition of emotion is done by using speech signal.  ...  In the past decade a lot of research has gone into Speech Emotion Recognition (SER).  ...  Pranab Das, project guide for expressing his confidence in us by his continuous support, help, and encouragement.  ... 
doi:10.33564/ijeast.2020.v04i12.083 fatcat:hf4gyd7vpjbgvjndt5m4w6vzee

Toward Sharing Brain Images: Differentially Private TOF-MRA Images With Segmentation Labels Using Generative Adversarial Networks

Tabea Kossen, Manuel A. Hirzel, Vince I. Madai, Franziska Boenisch, Anja Hennemuth, Kristian Hildebrand, Sebastian Pokutta, Kartikey Sharma, Adam Hilbert, Jan Sobesky, Ivana Galinovic, Ahmed A. Khalil (+2 others)
2022 Frontiers in Artificial Intelligence  
Additionally, the Fréchet Inception Distance (FID) was calculated between the generated images and the real images to assess their similarity.  ...  Our best segmentation model, trained on synthetic and private data, achieved a Dice Similarity Coefficient (DSC) of 0.75 for ϵ = 7.4 compared to 0.84 for ϵ = ∞ in a brain vessel segmentation paradigm (  ...  The distance to the test data was similar for different ǫ values. Figure 6B shows the difference between the distances to the training images and test images for different values of ǫ.  ... 
doi:10.3389/frai.2022.813842 fatcat:a33awkwbezcuzcya3x45i4ysne

Algorithmic and statistical challenges in modern largescale data analysis are the focus of MMDS 2008

Michael W. Mahoney, LekHeng Lim, Gunnar E. Carlsson
2008 SIGKDD Explorations  
We provide a report for the ACM SIGKDD community about the 2008 Workshop on Algorithms for Modern Massive Data Sets (MMDS 2008), its origin in MMDS 2006, and future directions for this interdisciplinary  ...  recognition.  ...  For instance, the Euclidean distance between DNA expression profiles in highthroughput microarray experiments may or may not capture a meaningful notion of distance between genes.  ... 
doi:10.1145/1540276.1540294 fatcat:tzzasuhsj5eb7bsqxgdglhldbq

Algorithmic and Statistical Challenges in Modern Large-Scale Data Analysis are the Focus of MMDS 2008 [article]

Michael W. Mahoney, Lek-Heng Lim, Gunnar E. Carlsson
2008 arXiv   pre-print
The 2008 Workshop on Algorithms for Modern Massive Data Sets (MMDS 2008), sponsored by the NSF, DARPA, LinkedIn, and Yahoo!, was held at Stanford University, June 25--28.  ...  The goals of MMDS 2008 were (1) to explore novel techniques for modeling and analyzing massive, high-dimensional, and nonlinearly-structured scientific and internet data sets; and (2) to bring together  ...  Acknowledgments The authors are grateful to the numerous individuals (in particular, Mayita Romero, Victor Olmo, and David Gleich) who provided assistance prior to and during MMDS 2008; to Diane Lambert for  ... 
arXiv:0812.3702v1 fatcat:ihribi3vibb7xhpx7xtpo5nxyq

Optimized Kernel Entropy Components

Emma Izquierdo-Verdiguier, Valero Laparra, Robert Jenssen, Luis Gomez-Chova, Gustau Camps-Valls
2017 IEEE Transactions on Neural Networks and Learning Systems  
Results show that OKECA returns projections with more expressive power than KECA, the most successful rule for estimating the kernel parameter is based on maximum likelihood, and OKECA is more robust to  ...  This maximum entropy preservation suggests that OKECA features are more efficient than KECA features for density estimation.  ...  training points, which is a common approach in kernel methods for classification, σ d1 , in the experiments, and the 15% of the median distance between points, which is the classical employed in KECA,  ... 
doi:10.1109/tnnls.2016.2530403 pmid:26930695 fatcat:dqxcmklfrjdbtlp5fsf5t7hkcq

The MEE Principle in Data Classification: A Perceptron-Based Analysis

Luís M. Silva, J. Marques de Sá, Luís A. Alexandre
2010 Neural Computation  
Our study also clarifies the role of the kernel density estimator of the error density in achieving the minimum probability of error in practice.  ...  ) for the Shannon entropy of the error, H S , or as R ≡ H R2 (E) = − ln E f 2 (e)de (1.3) for the quadratic Rényi entropy, H R2 .  ...  Figure 1 shows (for t = 1) the distance functions L MSE (e) = e 2 and L C E (e) = ln 1 2−te .  ... 
doi:10.1162/neco_a_00013 pmid:20569178 fatcat:5y6ihdzp6zcqtpybjbqtxpahqi

Bootstrap Equilibrium and Probabilistic Speaker Representation Learning for Self-supervised Speaker Verification [article]

Sung Hwan Mun, Min Hyun Han, Dongjune Lee, Jihwan Kim, Nam Soo Kim
2021 arXiv   pre-print
In the back-end stage, the probabilistic speaker embeddings are estimated by maximizing the mutual likelihood score between the speech samples belonging to the same speaker, which provide not only speaker  ...  probabilistic speaker embedding training in the back-end.  ...  ., “Augmentation adversarial training for unsupervised speaker recognition,” in Workshop on Self-upervised Learning for Speech and Audio Processing, NeurIPS, 2020. [21] H.  ... 
arXiv:2112.08929v1 fatcat:cm4plnaw2ngtnk23s5pq3cmjhe

Recognition of Activities of Daily Living Based on Environmental Analyses Using Audio Fingerprinting Techniques: A Systematic Review

Ivan Pires, Rui Santos, Nuno Pombo, Nuno Garcia, Francisco Flórez-Revuelta, Susanna Spinsante, Rossitza Goleva, Eftim Zdravevski
2018 Sensors  
An increase in the accuracy of identification of Activities of Daily Living (ADL) is very important for different goals of Enhanced Living Environments and for Ambient Assisted Living (AAL) tasks.  ...  Although this is usually used to identify the location, ADL recognition can be improved with the identification of the sound in that particular environment.  ...  The authors would also like to acknowledge the contribution of the COST Action IC1303-AAPELE-Architectures, Algorithms and Protocols for Enhanced Living Environments.  ... 
doi:10.3390/s18010160 pmid:29315232 pmcid:PMC5795595 fatcat:52ebtpevzjfdtbcehfkcw26xby
« Previous Showing results 1 — 15 out of 76 results