DocumentCode :
719287
Title :
Universal embeddings for kernel machine classification
Author :
Boufounos, Petros T. ; Mansour, Hassan
Author_Institution :
Mitsubishi Electr. Res. Labs., Cambridge, MA, USA
fYear :
2015
fDate :
25-29 May 2015
Firstpage :
307
Lastpage :
311
Abstract :
Visual inference over a transmission channel is increasingly becoming an important problem in a variety of applications. In such applications, low latency and bit-rate consumption are often critical performance metrics, making data compression necessary. In this paper, we examine feature compression for support vector machine (SVM)-based inference using quantized randomized embeddings. We demonstrate that embedding the features is equivalent to using the SVM kernel trick with a mapping to a lower dimensional space. Furthermore, we show that universal embeddings-a recently proposed quantized embedding design-approximate a radial basis function (RBF) kernel, commonly used for kernel-based inference. Our experimental results demonstrate that quantized embeddings achieve 50% rate reduction, while maintaining the same inference performance. Moreover, universal embeddings achieve a further reduction in bit-rate over conventional quantized embedding methods, validating the theoretical predictions.
Keywords :
inference mechanisms; pattern classification; radial basis function networks; support vector machines; SVM kernel trick; feature compression; kernel machine classification; kernel-based inference; quantized randomized embeddings; radial basis function kernel; support vector machine based inference; universal embeddings; Accuracy; Approximation methods; Feature extraction; Image coding; Kernel; Quantization (signal); Support vector machines;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Sampling Theory and Applications (SampTA), 2015 International Conference on
Conference_Location :
Washington, DC
Type :
conf
DOI :
10.1109/SAMPTA.2015.7148902
Filename :
7148902
Link To Document :
بازگشت