This article describes the implementation of a system that is able to organize vast document collections according to textual similarities. It is based on the Self-Organizing Map (SOM) algorithm. As the feature vectors for the documents we use statistical representations of their vocabularies. The main goal in our work has been to scale up the SOM algorithm to be able to deal with large amounts of high-dimensional data. In a practical experiment we mapped 6,840,568 patent abstracts onto a 1,002,240-node SOM. As the feature vectors we used 500-dimensional vectors of stochastic figures obtained as random projections of weighted word histograms. Keywords Data mining, exploratory data analysis, knowledge discovery, large databases, parallel implementation, random projection, Self-Organizing Map (SOM), textual documents. I. Introduction A. From simple searches to browsing of self-organized data collections Locating documents on the basis of keywords and simple search expressions is a c...

This thesis is organized in three parts. In the first part, the Self-Organizing Map algorithm is introduced. The discussion focuses on the analysis of the Self-Organizing Map algorithm. It is shown that the nonlinear nature of the algorithm makes it difficult to analyze the algorithm except in some trivial cases. In the second part the Self-Organizing Map algorithm is applied to several patterns sequence analysis tasks. The first application is a voice quality analysis system. It is shown that the Self-Organizing Map algorithm can be applied to voice analysis by providing the visualization of certain deviations. The key point in the applicability of Self-Organizing Map algorithm is the topological nature of the mapping; similar voice samples are mapped to nearby locations in the map. The second application is a speech recognition system. Through several experiments it is demonstrated that by collecting some time dependent features and using them in conjunction with the basic Self-Organ...

This work presents experiments to recognize pattern sequences using hidden Markov models (HMMs). The pattern sequences in the experiments are computed from speech signals and the recognition task is to decode the corresponding phoneme sequences. The training of the HMMs of the phonemes using the collected speech samples is a difficult task because of the natural variation in the speech. Two neural computing paradigms, the Self-Organizing Map (SOM) and the Learning Vector Quantization (LVQ) are used in the experiments to improve the recognition performance of the models. A HMM consists of sequential states which are trained to model the feature changes in the signal produced during the modeled process. The output densities applied in this work are mixtures of Gaussian density functions. SOMs are applied to initialize and train the mixtures to give a smooth and faithful presentation of the feature vector space defined by the corresponding training samples. The SOM maps similar feature vect...

In this paper a Recurrent Self-Organizing Map (RSOM) algorithm is proposed for temporal sequence processing. The RSOM algorithm is close in nature to the Kohonen's Self-Organizing Map, except that in the RSOM context of the temporal sequence is involved both in the best matching unit finding and in the adaptation of the weight vectors of the map via an introduced recursive difference equation associated for each unit of the map. The experimental results in the paper demonstrate that the RSOM is able to learn and distinguish temporal sequences, and that the RSOM algorithm can be utilized, for instance, in electroencephalogram (EEG) based epileptic activity detection.

In order to take advantage of the massive parallelism offered by artificial neural networks, hardware implementations are essential. However, most standard neural network models are not very suitable for implementation in hardware and adaptations are needed. In this section an overview is given of the various issues that are encountered when mapping an ideal neural network model onto a compact and reliable neural network hardware implementation, like quantization, handling non-uniformities and non-ideal responses, and restraining computational complexity. Furthermore, a broad range of hardware-friendly learning rules is presented, which allow for simpler and more reliable hardware implementations. The relevance of these neural network adaptations to hardware is illustrated by their application in existing hardware implementations.

Abstract—Context-aware computing describes the situation where a wearable/mobile computer is aware of its user’s state and surroundings and modifies its behavior based on this information. We designed, implemented, and evaluated a wearable system which can learn context-dependent personal preferences by identifying individual user states and observing how the user interacts with the system in these states. This learning occurs online and does not require external supervision. The system relies on techniques from machine learning and statistical analysis. A case study integrates the approach in a context-aware mobile phone. The results indicate that the method is able to create a meaningful user context model while only requiring data from comfortable wearable sensor devices. Index Terms—Location-dependent and sensitive, wearable computers, mobile computing, machine learning, wearable AI, statistical models. 1

Localization is an essential service for many wireless sensor network applications. While several localization schemes rely on anchor nodes and range measurements to achieve fine-grained positioning, we propose a range-free, anchorfree solution that works using connectivity information only. The approach, suitable for deployments with strict cost constraints, is based on the neural network paradigm of Self-Organizing Maps (SOM). We present a lightweight SOMbased algorithm to compute virtual coordinates that are effective for location-aided routing. This algorithm can also exploit the location information, if available, of few anchor nodes to compute absolute positions. Results of extensive simulations show improvements over the popular Multi-Dimensional Scaling (MDS) scheme, especially for networks with low connectivity, which are intrinsically harder to localize, and in presence of irregular radio pattern or anisotropic deployment. We analytically demonstrate that the proposed scheme has low computation and communication overheads; hence, making it suitable for resource-constrained networks.

We argue in favour of artificial neural networks for exploratory data analysis, clustering and mapping. We propose the Kohonen self-organizing map (SOM) for clustering and mapping according to a multi-maps extension. It is consequently called Multi-SOM. Firstly the Kohonen SOM algorithm is presented. Then the following improvements are detailed: the way of naming the clusters, the map division into logical areas, and the map generalization mechanism. The multi-map display founded on the inter-maps communication mechanism is exposed, and the notion of the viewpoint is introduced. The interest of Multi-SOM is presented for visualization, exploration or browsing, and moreover for scientific and technical information analysis. A case study in patent analysis on transgenic plants illustrates the use of the Multi-SOM. We also show that the inter-map communication mechanism provides support for watching the plants on which patented genetic technology works. It is the first map. The other four related maps provide information about the plant parts that are concerned, the target pathology, the transgenic techniques used for making these plants resistant, and finally the firms involved in genetic engineering and patenting. A method of analysis is also proposed in the use of this computer-based multi-maps environment. Finally, we discuss some critical remarks about the proposed approach at its current state. And we conclude about the advantages that it provides for a knowledge-oriented watching analysis on science and technology. In relation with this remark we introduce in conclusion the notion of knowledge indicators. 1.

Introduction Detection of nonlinearly distorted and multipath-propagated signals is an essential problem in telecommunications. Multipath propagation introduces intersymbol interference (ISI) into the signal. ISI is conventionally compensated using a Decision Feedback Equalizer (DFE) [Benedetto et al., 1988]. The DFE is unable to handle nonlinear distortions, which appear like an irregular configuration of the signal constellation. To compensate for these irregularities, a couple neural-network-based equalizers have been introduced in [Kohonen et al., 1991]. In these equalizers the DFE has been used as a preprocessing unit, and the Self-Organizing Map (SOM) [Kohonen, 1989] as an adaptive detector, respectively. 2 SOM-based equalizers In this application, each node of a SOM corresponds to one state of the signal constellation. When used as an adaptive detector, the SOM updates its decision levels to follow up the nonlinearities. Here we compare equalizers in w

This paper explains how some properties of the Self-Organizing Maps (SOMs) can be exploited in the density models used in continuous density hidden Markov models (HMMs). The three main ideas are the suitable initialization of the centroids for the Gaussian mixtures, the smoothing of the HMM parameters and the use of topology for fast density approximations. The methods are tested here in the automatic speech recognition framework, where the task is to decode the phonetic transcription of spoken words by speaker dependent, but vocabulary independent phoneme models. The results show that the average number of final recognition errors will be over 15 % smaller, if the traditional K-means based initialization is substituted by SOM. The method described for fast SOM density approximation improves the total recognition time by over 40 % for the current online system compared to the default which uses independent complete searches for the best matching units. 1 About the application The auto...