M. D. Plumbley. Adaptive lateral inhibition for non-negative ICA. In Proceedings of the international Conference on Independent Component Analysis and Blind Signal Separation (ICA2001), pages 516--21, 2001.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... a neurobiological context by Lee and Seung [81] calling it non negative matrix factorization) who also developed e#cient algorithms for solving the problem [81, 82] In the context of ICA, non negativity constraints (on A, s, or both) have recently been considered by several authors, see e.g. [97, 106, 114, 117, 118]. The main contributions of Publication 4 were the application of non negativity constraints in the sparse coding framework [110, 111] and the extension of the algorithm proposed in [82] to this case. 9.3 Learning receptive fields In Publication 5, the algorithm proposed in Publication 4 was ....

M. Plumbley, "Adaptive lateral inhibition for non-negative ICA," in Proc. Int. Workshop on Independent Component Analysis and Blind Signal Separation (ICA2001.

.... a neurobiological context by Lee and Seung [81] calling it non negative matrix factorization) who also developed e#cient algorithms for solving the problem [81, 82] In the context of ICA, non negativity constraints (on A, s, or both) have recently been considered by several authors, see e.g. [97, 106, 114, 117, 118]. The main contributions of Publication 4 were the application of non negativity constraints in the sparse coding framework [110, 111] and the extension of the algorithm proposed in [82] to this case. 9.3 Learning receptive fields In Publication 5, the algorithm proposed in Publication 4 was ....

M. Plumbley, "Adaptive lateral inhibition for non-negative ICA," in Proc. Int. Workshop on Independent Component Analysis and Blind Signal Separation (ICA2001.

....in addition to non negativity, and it may be that this is sufficient to constrain the solution. The current author proposed coupling independence with non negativity constraints, using a neural network with rectified outputs and anti Hebbian lateral inhibitory connections between the output units [14]. Taking a slightly different approach, it is possible to show that the dual constraints of non negativity and independence of sources lead to the conclusions that, under certain reasonable assumptions, source separation of pre whitened data can be achieved by finding an orthonormal matrix that ....

M.D. Plumbley, "Adaptive lateral inhibition for non-negative ICA," in Proceedings of the International Conference on Independent Component Analysis and Signal Sepa- ration (ICA2001.

....sources have large concentration of probability around zero, representing a high probability of being o . We would expect that a learning algorithm developed from the principles outlined in this letter would work particularly well for sources with this type of distribution. In a previous paper [6] we experimentally investigated learning algorithms based on inhibitory connections to remove the covariance in y . The current proof is slightly di erent, in that the pre whitening ensures that the covariance in y (rather than y ) is removed, and therefore Theorem 1 is not a convergence ....

M. D. Plumbley, \Adaptive lateral inhibition for non-negative ICA," 2001.

No context found.

M. D. Plumbley. Adaptive lateral inhibition for non-negative ICA. In Proceedings of the international Conference on Independent Component Analysis and Blind Signal Separation (ICA2001), pages 516--21, 2001.

No context found.

M. D. Plumbley. "Adaptive lateral inhibition for nonnegative ICA", Proc. 3rd Int. Conf. on Independent Component Analysis and Blind Signal Separation (ICA 2001.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC