Gaussian ARTMAP: A Neural Network for Fast Incremental Learning of Noisy Multidimensional Maps

By James R. Williamson. 1996

A new neural network architecture for incremental supervised learning of analog multidimensional maps is introduced. The architecture, called Gaussian ARTMAP, is a synthesis of a Gaussian classifier and an adaptive resonance theory (ART) neural network, achieved by defining the ART choice function as the discriminant function of a Gaussian classifier with separable distributions, and the ART match function as the same, but with the distributions normalized to a unit height. While Gaussian ARTMAP retains the attractive parallel computing and fast learning properties of fuzzy ARTMAP, it learns a more efficient internal representation of a mapping while being more resistant to noise than fuzzy ARTMAP on a number of benchmark databases. SSeveral simulations are presented which demonstrate that Gaussian ARTMAP consistently obtains a better trade-off of classification rate to number of categories than fuzzy ARTMAP. Results on a vowel classification problem are also presented which demonstrate that Gaussian ARTMAP outperforms many other classifiers. Copyright 1996 Elsevier Science Ltd

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Memory Efficient Experience Replay for Streaming Learning (2018, background): “ARTMAP networks [11], [12], [13], [71] are another approach and they learn faster.”
Lifelong Machine Learning with Deep Streaming Linear Discriminant Analysis (2019, methodology): “There have been shallow neural networks designed for streaming learning, including ARTMAP networks [7, 8, 49]; however, ARTMAP is sensitive to the order in which training data is presented and it is not capable of representation learning.”
Adaptive resonance theory (1997, background): “Because vigilance then varies across learning trials, the memories of a single ARTMAP system typically exhibit a range of degrees of refinement.”
Clustering: A neural network approach (2010, methodology): “Othermembers of the ARTMAP family are the ART-EMAP (Carpenter & Ross, 1995), the ARTMAP-IC (Carpenter & Markuzon, 1998), the Gaussian ARTMAP (Williamson, 1996), the distributed ARTMAP (dARTMAP) (Carpenter, 1997), the default ARTMAP (Carpenter, 2003), and the simplified fuzzy ARTMAP (Kasuba, 1993;…”
Learn++: an incremental learning algorithm for supervised neural networks (2001, background): “Various algorithms have been suggested to overcome such difficulties [15]–[20].”
Survey of clustering algorithms (2005, background): “Determine the next generation by keeping the individuals with the highest ﬁ tness.”
Learn$^{++}$ .NC: Combining Ensemble of Classifiers With Dynamically Weighted Consult-and-Vote for Efficient Incremental Learning of New Classes (2009, background): “Various solutions, modifying ARTMAP [10], [34]–[36], or other approaches such as growing neural gas networks and cell structures [37] have also been proposed.”
Biased ART: A neural architecture that shifts attention toward previously disregarded features following an incorrect prediction (2010, unknown): “Design principles derived from scientific analyses and design constraints imposed by targeted applications have jointly guided the development of many variants of the basic networks, including fuzzy ARTMAP (Carpenter et al., 1992), ARTMAP-IC (Carpenter & Markuzon, 1998), and Gaussian ARTMAP (William”