The dynamic competitive learning method

Dynamic competitive learning is an unsupervised learning technique. The only information the model gets is the pattern imposed on the input units. We don’t give any information concerning the goodness of a decision as e.g. an expected answer, which it can use as an error measure to approach the proper behavior (as e.g. with the back-propagation algorithm).

Face recognition has always been a potential research area because of its demand for reliable identification of a human being especially in government and commercial sectors, such as security systems, criminal identification, border control, etc. where a large number of people interact with each other and/or with the system. The last two decades have witnessed many supervised and unsupervised learning techniques proposed by different researchers for the face recognition system. Principal component analysis (PCA), self‐organizing map (SOM), and independent component analysis (ICA) are the most widely used unsupervised learning techniques reported by research community. This article presents an analysis and comparison of these techniques. The article also includes two SOM processing methods global SOM (GSOM) and local SOM (LSOM) for performance evaluation along with PCA and ICA. We have used two different databases for our analysis. The simulation result establishes the supremacy of GSOM in general among all the unsupervised techniques. © 2010 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 20, 261–267, 2010

Many data mining tasks require the unsupervised partitioning of a data set into clusters. However, in many case we do not really know any prior knowledge about the clusters, for example, the density or the shape. This paper addresses two major issues associated with conventional competitive learning, namely, sensitivity to initialization and difficulty in determining the number of clusters. Many methods exist for such clustering, but most of then have assumed hyper-ellipsoidal clusters. Many heuristically proposed competitive learning methods and its variants, are somewhat ad hoc without any theoretical support. Under above considerations, we propose an algorithm named as Entropy guided Splitting Competitive Learning (ESCL) in the information theorem framework. Simulations show that minimization of partition entropy can be used to guide the competitive learning process, so to estimate the number and structure of probable data generators.

Customer relationship management (CRM) is the tool to enhance customer relationship in any business. Due to the exponential growth of data volume, in any field, it is significant to develop new techniques to discover the customer knowledge, automation of the system and moreover customer satisfaction to win customer lifetime value. CRM with machine learning could bring a catalytic change in business. Several supervised and unsupervised machine learning techniques are utilized to improve the customer experience and profitability of business. This paper reviews the available literature on the CRM with machine learning techniques for customer identification, customer attraction, and customer retention and customer development. This study reveals that supervised learning techniques are 48.48% utilized, unsupervised learning techniques are utilized 15.15%, and 9.09% utilized other techniques in CRM. Paradigm is also shifted toward the deep learning from machine learning as 28.28% text has been reported to deep learning. Decision tree-based algorithm and support vector machine algorithms are most utilized algorithm of supervised learning. E-commerce and telecommunication sectors are the most important areas identified with the exponential growth of the users and hence need a suitable machine learning techniques for customer satisfaction and business profitability.

Chapter 17 Geometry of neural networks: Natural gradient for learning

Kenya is experiencing a significant increase in the prevalence of non-communicable diseases (NCDs) such as cardiovascular diseases, hypertension, Type 2 diabetes, and certain cancers (bowel, lung, prostate, and uterine). This case is not unique to Kenya but is common in many Low and Middle-Income Countries (LMICs) in Africa. Many NCDs, are linked to diets high in added sugars, sodium, saturated fat, and low in fiber. There is a notable lack of information regarding the demographic differences among supermarket customers and their purchasing habits of healthy versus unhealthy foods in some parts of Africa. This gap in knowledge hinders the ability to connect grocery purchase patterns to NCDs, including obesity. Supermarkets in LMICs offer valuable demographic insights through grocery data. This research utilizes NOVA classification tool, data mining and unsupervised machine learning techniques to analyze grocery purchase patterns in 10 supermarkets across 5 counties in Kenya between 2022 and 2023. The apriori algorithm was used to create association rules and an analysis was done on the association rules to find out the relationship between demography (location, gender, and age) with purchase patterns. Individual data was collected along with transaction data, since the supermarkets logged transactions done by loyalty card customers. The main aim is to provide guidance to policymakers in public health. We collected 3 934 122 unique transactions and each transaction was associated with a customer who was identified with a unique customer ID. Findings from this research demonstrate that 53% of food purchases from these transactions were mainly industrially processed food items and males above the age of 50 years were the main consumers of these food items. The findings lead to the conclusion that this purchase trend has a chance of rising NCDs in older people. Therefore we recommend that policymakers adopt our recommendations to safeguard public health.

Clustering as an unsupervised learning technique has been widely used in practice. In this paper, a novel clustering algorithm based on region segmentation (CRS) is proposed. It aims to automatically evolve the optimal number of clusters as well as the clusters of the data sets based on the data density. First, a new data density is given based on the reverse near neighbor enhancement which can make the clusters detection more effectively. Then, the multiple sub-region centers can be determined through the data density. Moreover, a merge criterion is proposed to make the relevant regions be merged and obtain the final clustering results. The proposed algorithm does not need to know the number of clusters in advance and no threshold limit. Therefore, it can be used more widely. In the experiments, we compare the performance of our CRS algorithm with DBSCAN, IS-DBSCAN, STClu, DP, and SCDOT algorithms on synthetic, and real-world data sets. Experimental results demonstrated that the NMI, ACC, F1 and ARI obtained by CRS algorithm is always better than that obtained by the other algorithms for the same data sets.

Distributed memory and localized memory in neural networks

Segmentation of images into textural homogeneous regions is a fundamental problem in an image understanding system. Most region-oriented segmentation approaches suffer from the problem of different thresholds selecting for different images. In this paper an adaptive image segmentation based on vector quantization is presented. It automatically segments images without preset thresholds. The approach contains a feature extraction module and a two-layer hierarchical clustering module, a vector quantizer (VQ) implemented by a competitive learning neural network in the first layer. A near-optimal competitive learning algorithm (NOLA) is employed to train the vector quantizer. NOLA combines the advantages of both Kohonen self- organizing feature map (KSFM) and K-means clustering algorithm. After the VQ is trained, the weights of the network and the number of input vectors clustered by each neuron form a 3- D topological feature map with separable hills aggregated by similar vectors. This overcomes the inability to visualize the geometric properties of data in a high-dimensional space for most other clustering algorithms. The second clustering algorithm operates in the feature map instead of the input set itself. Since the number of units in the feature map is much less than the number of feature vectors in the feature set, it is easy to check all peaks and find the `correct' number of clusters, also a key problem in current clustering techniques. In the experiments, we compare our algorithm with K-means clustering method on a variety of images. The results show that our algorithm achieves better performance.

The neural network approach by the backpropagation method (BPM) to land-use has been discussed in recent years. Using such a method, the accuracy depends on a training data set which has been selected manually. It takes much time to select a suitable training data set. In this paper, as a preprocessing of classifications the authors use the Kohonen feature map (KFM) and the competitive learning (CL) to get the better training data set. As a first step, the KFM that takes the Landsat TM data as input is adopted to form a rough classification of the wide area based on the observed data. In the next step, the CL whose inputs are the weights of the KFM node data is carried out to determine the category of each node of the KFM. The first weight set of the CL is a set on weights at four corners of the KFM. The combination of the two neural network techniques enables the authors to determine the rough land-use of an object region automatically. After that, the classification results by the KFM and the CL are further classified into more fine items by using the BPM. Finally, the classification results have been compared with other methods.

Maximum likelihood classifier that is often used in classification of satellite images assumes the distribution of each class to Gaussian. Such linear classifier can classify correctly when the case that classification probability of each class is exclusive. Remotely sensed data, however, belong to several classes and have non-linear separable condition. To improve the classification accuracy of non-linear separable data, the application of the single-step multi-layer back propagation neural networks have been studied by many researchers. In this paper, multi-step multi-layer neural networks, so called cooperative learning neural networks, are proposed to classify the non-linear separable satellite data.The cooperative learning neural network consists of extraction networks for each class and an unification network which unifies the extracted values. The unification network is also used for unification of different environments such as time-series data or neighboring regions. The result of the classification of LANDSAT TM data of Nagoya city using the cooperative learning neural network is introduced. Classified image is compared with the detailed digital land cover information (TDT-112) and the images classified using single-step multi-layer neural network, maximum likelihood classifier and fuzzy set reasoning. As the result of the comparison, the cooperative learning neural network classify the remote sensing data more exactly than the other methods.

In this paper, a novel neural-fuzzy controller based on fuzzy differential competitive learning is proposed. Since one of the most important parts is the generation of the fuzzy rules in the design of the fuzzy control system, a fast learning algorithm, fuzzy differential competitive learning (FDCL), for the generation of the rules is applied in the fuzzy control system. The FDCL algorithm adopts a principle of learn according to how well it wins. Unlike the previous competitive learning algorithm such as crisp competitive learning algorithms where only one neuron will win and learn at each competition step every neuron in the neural network based on FDCL algorithm will along with its different distance to the input pattern and learns the pattern accordingly. Compared with the ordinary competitive learning algorithm the proposed FDCL algorithm has the various distinguishing features. The FDCL algorithm is implanted in the neural network based fuzzy system and the network adopted is fuzzy associative memory system (FAMS) which simulates the knowledge representation and inference process by using fuzzy notation and by association in neural networks. In FAMS the fuzzy rules will be generated by clustering the input-output training data through the FDCL paradigm. By using the FDCL algorithm the neural network can highly refine knowledge and represent the expert experience.

In general, in competitive learning the requirement for the initial number of prototypes is a difficult task, as we do not usually know the number of clusters in the input data a priori. The behavior and performance of the competitive algorithms are very sensitive to the initial locations and number of the prototypes. In this paper after investigating several important competitive learning paradigms, we present compensation techniques for overcoming the problems in competitive learning. Our experimental results show that competition with compensation can improve the performance of the learning algorithm.

Decomposition of interacting features using a Kohonen self-organizing feature map neural network

Aimed at the phenomenon that there is few theories and applications about multi-layer ADALINE neural network (NN), in this paper, mathematic expressions of the multi-input/ single-output (MISO) and multi-input/ multi-output (MIMO) ADALINE NN with multi-layer had been deduced according to the theory of linear neuron transfer function. At result, expressions are also linear functions respect with input variables with more complex weights and thresholds. In order to verify the equivalence of multi-layer ADALINE NNand single-layer ADALINE NN, two examples had been simulated with MATLAB software for function approximation, respectively. The simulation results showed that the multi-layer ADALINE NN can approximate linear functions but not as well as the single-layer ADALINE NN does. In conclusion, the existence of multi-layer adaline NN is not necessary, and its function can be simply realized with the single-layer ADALINE NN completely.