Topological Clustering via Adaptive Resonance Theory With Information Theoretic Learning

Abstract

This paper proposes a topological clustering algorithm by integrating topological structure and information theoretic learning, i.e., correntropy, into adaptive resonance theory (ART). Specifically, the proposed algorithm utilizes the correntropy induced metric (CIM) for defining a similarity measure, a node insertion criterion, and an edge creation criterion. Other types of the ART-based topological clustering algorithms have been developed, however, these algorithms have various drawbacks such as a large number of parameters, sensitivity to noisy data. Moreover, generated topological networks cannot represent the distribution of data. In contrast, the proposed algorithm realizes a stable computation and reduces the number of parameters compared to existing algorithms. Furthermore, improving the ability to express the data structure more appropriately by the topological network, a mechanism that adaptively controls the node insertion criterion is introduced to the proposed algorithm. The experimental results showed that the proposed algorithm has superior performance with respect to the self-organizing and the classification abilities compared with the state-of-the-art topological clustering algorithms. INDEX TERMS Adaptive resonance theory, correntropy, information theoretic learning, topological clustering.

Time-Series Prediction of Iron and Silicon Content in Aluminium Electrolysis Based on Machine Learning

“…The input and output of neurons in the GMDH network need a reference function. The Volterra function is commonly used to represent the input-output relationship as follows:  (19) where the i a is the weight vector determined by the regression method to reduce the error between y and ŷ .…”

“…Due to the lack of ideal samples, a simple but very practical GMDH polynomial network is used to estimation the state of battery health by Wu [18]. However, the above models are still not stable enough, especially when working in the non-stationary and noisy environment of aluminium electrolysis [19].…”

Adaptive Resonance Theory-based Topological Clustering with a Divisive Hierarchical Structure Capable of Continual Learning

“…Another successful approach is ART-based clustering algorithms such as Fuzzy ART [7], Bayes ART [8], and their variants [9], [10]. In general, ART-based clustering algorithms show superior clustering performance than GNGbased and SOINN-based algorithms [11], [12]. Moreover, because ART-based clustering algorithms can theoretically realize sequential and class-incremental learning without catastrophic forgetting, a number of ART-based clustering algorithms and their improvements have been proposed in both supervised learning [22]- [24] and unsupervised learning [7], [8], [25], [26].…”

Individualized AI Tutor Based on Developmental Learning Networks

“…where β I and β O are the learning rates for the internal division and outer division, respectively. According to (15), the high effectiveness template k w JH expands to the high-effectiveness content k x H . On the other hand, the area of the low effectiveness template k w JL is reduced by (16).…”

“…By the work of many researchers, the ART network has been improved so that it can handle complicated input patterns [9]- [11]. Recently, the ART network was utilized for classification and recommendation [12], [13], hybrid data regression [14], topological clustering [15], implementation of long term memory [16]- [19], and interactive learning [20]. However, ART networks still have limitations in direct application to AI tutors.…”