Quantized nonnegative matrix factorization

de Fréin, Ruairí (2014) Quantized nonnegative matrix factorization. In: IEEE Digital Signal Processing (DSP), 2014 19th International Conference on, 20 - 23 August 2014, Hong Kong.

[thumbnail of rdefreinDSP14.pdf]

Download (362kB) | Preview
Official URL: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?...


Even though Nonnegative Matrix Factorization (NMF) in its original form performs rank reduction and signal compaction implicitly, it does not explicitly consider storage or transmission constraints. We propose a Frobenius-norm Quantized Nonnegative Matrix Factorization algorithm that is 1) almost as precise as traditional NMF for decomposition ranks of interest (with in 1–4dB), 2) admits to practical encoding techniques by learning a factorization which is simpler than NMF's (by a factor of 20–70) and 3) exhibits a complexity which is comparable with state-of-the-art NMF methods. These properties are achieved by considering the quantization residual via an outer quantization optimization step, in an extended NMF iteration, namely QNMF. This approach comes in two forms: QNMF with 1) quasi-fixed and 2) adaptive quantization levels. Quantized NMF considers element-wise quantization constraints in the learning algorithm to eliminate defects due to post factorization quantization. We demonstrate significant reduction in the cardinality of the factor signal values set for comparable Signal-to-Noise-Ratios in a matrix decomposition task.

Item Type: Conference or Workshop Item (Paper)
Additional Information: @INPROCEEDINGS{deFrein14Quantized, author={de Frein, Ruairi}, booktitle={Digital Signal Processing (DSP), 2014 19th International Conference on}, title={Quantized nonnegative matrix factorization}, year={2014}, month={Aug}, pages={377-382}, keywords={Approximation methods;Convergence;Digital signal processing;Matrix decomposition;Optimization;Quantization (signal);Signal processing algorithms;low rank;nmf;quantization}, doi={10.1109/ICDSP.2014.6900690},}
Uncontrolled Keywords: Approximation methods; Convergence; Digital signal processing; Matrix decomposition; Optimization; Quantization (signal); Signal processing algorithms; low rank; NMF; quantization
Departments or Groups: Walton Institute for Information and Communications Systems Science
Walton Institute for Information and Communications Systems Science > Centre for Converged IP Communications Services (3CS)
Divisions: School of Science > Department of Computing, Maths and Physics
Depositing User: Ruairi De Frein
Date Deposited: 23 Aug 2016 11:27
Last Modified: 23 Aug 2016 11:27
URI: https://repository.wit.ie/id/eprint/3160

Actions (login required)

View Item View Item