next up previous contents
Next: 2.3 Mixture of local Up: 2.2 Vector quantization Previous: 2.2.3 Deterministic annealing


2.2.4 Neural Gas

Martinetz et al. (1993) presented an algorithm called `Neural Gas' that outperforms deterministic annealing (Rose et al., 1990). Neural Gas is also a variant of soft-clustering, and it also uses annealing. In contrast, Neural Gas has a different (heuristic) assignment, and it can be only carried out as an on-line version.

The algorithm starts by randomly choosing m data points as starting points for the m code-book vectors. The annealing consists of a predefined number of tmax steps. During each annealing step t, a pattern $ \bf x_{i}^{}$ is randomly drawn from the training set. Then, the code-book vectors are sorted in the order of their Euclidean distance to $ \bf x_{i}^{}$. Let k($ \bf x_{i}^{}$,$ \bf c_{j}^{}$(t)) be the resulting rank of each code-book vector, with k = 0 for the closest vector, and k = m - 1 for the most distant vector. The soft-assignment is then given by the function h$\scriptstyle \varrho$ij = exp(- k($ \bf x_{i}^{}$,$ \bf c_{j}^{}$(t))/$ \varrho$). The parameter $ \varrho$ is a measure of the neighborhood range. Given this assignment, all code-book vectors $ \bf c_{j}^{}$ are updated according to

$\displaystyle \bf c_{j}^{}$(t + 1) = $\displaystyle \bf c_{j}^{}$(t) + $\displaystyle \varepsilon$(t)h$\scriptstyle \varrho$(t)ij$\displaystyle \left[\vphantom{{\bf x}_i - {\bf c}_j(t)}\right.$$\displaystyle \bf x_{i}^{}$ - $\displaystyle \bf c_{j}^{}$(t)$\displaystyle \left.\vphantom{{\bf x}_i - {\bf c}_j(t)}\right]$ . (2.15)

During the annealing process both parameters $ \varepsilon$ and $ \varrho$ decrease exponentially, $ \varrho$(t) = $ \varrho$(0)[$ \varrho$(tmax)/$ \varrho$(0)](t/tmax), and $ \varepsilon$(t) = $ \varepsilon$(0)[$ \varepsilon$(tmax)/$ \varepsilon$(0)](t/tmax). At the end of the annealing, the algorithm changes to on-line k-means, and thus, also (locally) minimizes (2.12). The exponential decay of $ \varepsilon$ enforces convergence. Neural Gas is stable and does not depend on the initial configuration of {$ \bf c_{i}^{}$} (Martinetz et al., 1993).


next up previous contents
Next: 2.3 Mixture of local Up: 2.2 Vector quantization Previous: 2.2.3 Deterministic annealing
Heiko Hoffmann
2005-03-22