Kohonen SOM Research Articles

Neural network analysis of telemetry data of on-board equipment of space crafts

Goal. Research and development of methodology and software tools of machine automated analysis of telemetry data of onboard equipment (OE) of space crafts (SC). Research methods. The developed software tools and methodology are based on the machine learning, neural networks and image processing methods and algorithms. Results. The paper presents solutions for the actual tasks of machine analysis of telemetry data of OE SC with the purpose of detecting the states of its functioning and analyzing the reliability and operability. Software tools and methodology of neural network clustering-classification analysis of OE SC telemetry data based on the application of the neural networks such as the Kohonen SOM and image processing methods have been developed. The software tools were implemented in desktop and web versions and has a flexible modular service-oriented architecture. Conclusion. The presented software tools and the methodology of neural network analysis of OE SC telemetry data were tested on real telemetry data of the Belarusian spacecraft and the SC group AIST and showed the results with a confidence probability value of at least 0.9. The proposed tools of neural network analysis of the OE SC telemetry data make it possible to develop recommendations for improving the indicators of OE SC reliability during design and operation, detecting the OE SC states, making the correct control and operational decisions of the ground control complex. Key words: neural network, Kohonen SOM, image processing, modular service-oriented architecture, machine neural network telemetry data analysis, onboard equipment of space crafts.

Online video summarization with the Kohonen SOM in real time

Online video summarization with the Kohonen SOM in real time

Analysis of Song Popularity in Business Digital Music Streaming for Increasing Quality Using Kohonen SOM Algorithm

Consumption of digital music services has grown dramatically in recent years. There is an increase in music streaming consumption from 2015 to 2016, which is 76.4%. One of the most popular music streaming services, Spotify, has experienced an increase in customers from year to year. This increase enables businessmen / music producers to increase their business profits by analyzing music / songs to find out the audio attributes that make the song enjoyable for many people. Processing and analysis data are using Kohonen SOM Algorithm. The function is to find out which audio attribute groups are most liked by Spotify users where a good music is a music that can be used as a therapy. The result is LR = 0.1, PLR = 0.9, and epoch = 70 - 500, it can be concluded that cluster 2 is the cluster that has the most number of streams with 27 songs where the smallest DBI value is obtained at epoch = 200. Thus, with the statistic analysis, the obtained information is; it is expected that businessman / music producers can increase their business profits by improving their music quality that focus on songs with modes = 0 (Minor) and loudness features

Digital Signs Security System using AES-Blowfish-RSA Hybrid Cryptography Approach

Increasing application of digital signatures in legitimate authentication of administrative documents in both public and private environments is one of the points of concern, especially the issue of security and integrity of ownership of signatures. Digital signature is a mathematical scheme, which a unit to identify and prove the authenticity of the owner of the message or document. The study aims to analyze security patterns and identification of digital signatures on documents using the RSA-AES-Blowfish hybrid cryptographic method approach for securing digital signatures, while the Kohonen SOM method is applied to identify ownership recognition of signature images. The analysis framework used in this study is each signature will be stored in the form of a digital image file that has been encrypted using hybrid method of AES-Blowfish with the SHA 256 hash function. Process of forming private keys and public keys in the signature image using the RSA algorithm. Authentic verification of the use of digital signatures on the document has 2 (two) stages, the first stage is signature will be valid used on the document if the result of hashing the selected signature image is the same based on the private key and public key entered by the user, while the second stage identification is done using the Kohonen SOM method to validate the similarity of the chosen signature with the ownership of the signature.

Analysis of Facial Image Extraction on Facial Recognition using Kohonen SOM for UNPRI SIAKAD Online User Authentication

Academic Information System (Sistem Informasi Akademik aka SIAKAD) Online of Universitas Prima Indonesia (UNPRI) is one of the applications used to facilitate the administration process of lectures which includes the filling process of study plan cards (Kartu Rencana Studi aka KRS), study result cards (Kartu Hasil Studi aka KHS), class schedules, submission of research titles, seminars, and other processes. SIAKAD UNPRI can be accessed by students, lecturers, and academics where every user has a password that has been encrypted to maintain the security of information from people who are not responsible, password security using the encryption method needs to be changed regularly, but there are still many students, lecturers and academic community who are reluctant to change passwords. To improve the security verification stage for SIAKAD users, we propose a face recognition feature approach. Face recognition is a feature that allows the identification of someone from a digital image or video. The way the facial recognition method works is by comparing face data from the camera or images with images that were previously stored in a database. In this study, the Kohonen SOM method is proposed for face identification based on the feature extraction approach of discrete cosine transform (DCT), linear discriminant analysis (LDA) and principal component analysis (PCA) to improve the security of UNPRI SIAKAD users. The analytical framework is done by requiring students to do face taking, where each student will save 5 (five) faces extracted with facial features using the DCT, LDA and PCA model approach, feature extraction results are used as input to the Kohonen SOM network for training and testing facial recognition, then analysis of the effect of DCT, LDA and PCA feature extraction on the Kohonen network on facial recognition accuracy.

Microencapsulation of fish oil – determination of optimal wall material and encapsulation methodology

For the first time, we present a meta-analysis of experimental and literature data to determine which microencapsulation methodology, and which wall material are best suited to protect fish oil. Our analysis covered a period of several decades of research (1984–2018). The analysis was conducted on 196 literature data-points, and 16 data-points determined experimentally for this publication. PLS regression was used to determine the influence of the predictors, with a variety of rational (k-means, Kohonen SOM, and Kennard-Stone) and random training/test division methods. Analysis of the applicability domain, determined by calculating the convex hull of the points, clearly showed that rational division methods are better than random division. It was found, that spray-granulation is best suited to protect fish oil. On the other hand, the commonly cited method of spray-drying performs quite badly. In addition, we found that the best wall materials are protein + lipid + carbohydrate and protein + lipid. The commonly used protein + carbohydrate again fared poorly. Meta-analysis such as these are crucial for the overview of methods and materials employed in the design of functional foods.

Stress Measures in SOM Learning

Various stress measures can be used in generalized version of Sammon’s mapping. Kohonen SOM with iterative or batch learning is a standard tool for data self-organization, too. Our method applies stress functions to pattern relationships in SOM and converts batch learning to discrete optimization task. Due to NP–completeness of SOM learning, optimization heuristics have to be used. Simulated annealing making use of Lévy flights is the recommended heuristics for this task.

Clustering digital forensic string search output

This research comparatively evaluates four competing clustering algorithms for thematically clustering digital forensic text string search output. It does so in a more realistic context, respecting data size and heterogeneity, than has been researched in the past. In this study, we used physical-level text string search output, consisting of over two million search hits found in nearly 50,000 allocated files and unallocated blocks. Holding the data set constant, we comparatively evaluated k-Means, Kohonen SOM, Latent Dirichlet Allocation (LDA) followed by k-Means, and LDA followed by SOM. This enables true cross-algorithm evaluation, whereas past studies evaluated singular algorithms using unique, non-reproducible datasets. Our research shows an LDA + k-Means using a linear, centroid-based user navigation procedure produces optimal results. The winning approach increased information retrieval effectiveness, from the baseline random walk absolute precision rate of 0.04, to an average precision rate of 0.67. We also explored a variety of algorithms for user navigation of search hit results, finding that the performance of k-means clustering can be greatly improved with a non-linear, non-centroid-based cluster and document navigation procedure, which has potential implications for digital forensic tools and use thereof, particularly given the popularity and speed of k-means clustering.

Optimizations of the Gravitationally Organized Related Mapping ANN through Genetic Algorithms

GORMANN is a self-organizing neural network utilizing Newton's law of universal gravitation. The results GORMANN are similar to a cross between a Kohonen SOM and morphological skeleton: the input is discretized and only the key features of the input are preserved. Like the SOM, GORMANN requires a number of input parameters (e.g. the learning rate) which greatly impact the quality of the resulting network. There have been several successful applications of genetic algorithms to optimize SOM parameters. Based on the success of these efforts, we applied a genetic algorithm to GORMANN to achieve the same end. Due to the theoretical nature and immaturity of the GORMANN algorithm, especially when compared to the Kohonen SOM, a group of unique challenges must be overcome. As GORMANN is a young neural network architecture, it has no strong theoretical background. In addition, the uniqueness of GORMANN requires new methods of measuring and comparing performance. In this paper we introduce a genetic algorithm-based method for optimizing the input parameters of GORMANN. A group of two-dimensional input patterns is used to illustrate the effectiveness of our method.

Comparative Analysis of Kohonen-SOM and K-Means data mining algorithms based on Academic Activities

With the increasing demand of IT and subsequent growth in this sector, the high- dimensional data came into existence. Data Mining plays an important role in analyzing and extracting the useful information. The key information which is extracted from a huge pool of data is useful for decision makers. Clustering, one of the techniques of data mining is the mostly used methods of analyzing the data. In this paper, the approach of Kohonen SOM and K-Means and HAC are discussed. After that these three methods are used for analyzing the academic data set of the faculty members of particular university. Finally a comparative analysis of these algorithms are done against some parameters like number of clusters, error rate and accessing rate, etc. Â This work will present new and improved results from large-scale datasets.

Musical Instrument Recognition using Zero Crossing Rate and Short-time Energy

Traditionally, musical instrument recognition is mainly based on frequency domain analysis (sinusoidal analysis, cepstral coefficients) and shape analysis to extract a set of various features. Instruments are usually classified using k-NN classifiers, HMM, Kohonen SOM and Neural Networks. Recognition of musical instruments in multi-instrumental, polyphonic music is a difficult challenge which is yet far from being solved. Successful instrument recognition techniques in solos (monophonic or polyphonic recordings of single instruments) can help to deal with this task. We introduce an instrument recognition process in solo recordings of a set of instruments (flute, guitar and harmonium), which yields a high recognition rate. A large solo database is used in order to encompass the different sound possibilities of each instrument and evaluate the generalization abilities of the classification process. The basic characteristics are computed in 1sec interval and result shows that the estimation of zero crossing rate and short time energy reflects more effectively the difference in musical instruments.

Reusable components for partitioning clustering algorithms

Clustering algorithms are well-established and widely used for solving data-mining tasks. Every clustering algorithm is composed of several solutions for specific sub-problems in the clustering process. These solutions are linked together in a clustering algorithm, and they define the process and the structure of the algorithm. Frequently, many of these solutions occur in more than one clustering algorithm. Mostly, new clustering algorithms include frequently occurring solutions to typical sub-problems from clustering, as well as from other machine-learning algorithms. The problem is that these solutions are usually integrated in their algorithms, and that original algorithms are not designed to share solutions to sub-problems outside the original algorithm easily. We propose a way of designing cluster algorithms and to improve existing ones, based on reusable components. Reusable components are well-documented, frequently occurring solutions to specific sub-problems in a specific area. Thus we identify reusable components, first, as solutions to characteristic sub-problems in partitioning cluster algorithms, and, further, identify a generic structure for the design of partitioning cluster algorithms. We analyze some partitioning algorithms (K-means, X-means, MPCK-means, and Kohonen SOM), and identify reusable components in them. We give examples of how new cluster algorithms can be designed based on them.

Assignment of EC Numbers to Enzymatic Reactions with MOLMAP Reaction Descriptors and Random Forests

The MOLMAP descriptor relies on a Kohonen SOM that defines types of covalent bonds on the basis of their physicochemical and topological properties. The MOLMAP descriptor of a molecule represents the types of bonds available in that molecule. The MOLMAP descriptor of a reaction is defined as the difference between the MOLMAPs of the products and the reactants and numerically encodes the pattern of changes in bonds during a chemical reaction. In this study, a genome-scale data set of enzymatic reactions available in the KEGG database was encoded by the MOLMAP descriptors and was explored for the assignment of the official EC number from the reaction equation with Random Forests as the machine learning algorithm. EC numbers were correctly assigned in 95%, 90%, and 85% (for independent test sets) at the class, subclass, and subsubclass EC number level, respectively, with training sets including one reaction from each available full EC number. Increasing differences between training and test sets were explored, leading to decreased percentages of correct assignments. The classification of reactions only from the main reactants and products was obtained at the class, subclass, and subsubclass level with accuracies of 78%, 74%, and 63%, respectively.

Heart sound classification using wavelet transform and incremental self-organizing map

Determination of heart condition by heart auscultation is a difficult task and requires special training of medical staff. Computerized techniques suggest objective and more accurate results in a fast and easy manner. Hence, in this study it is aimed to perform computer-aided heart sound analysis to give support to medical doctors in decision making. In this study, a novel method is presented for the classification of heart sounds (HSs). Discrete wavelet transform is applied to windowed one cycle of HS. Wavelet transform is used both for the segmentation of S1–S2 sounds and determination of the features. Based on the third, fourth and the fifth decomposition-level detail coefficients, the timings of S1–S2 sounds are determined by an adaptive peak-detector. For the feature extraction, powers of detail coefficients in all five sub-bands are utilized. In the classification stage, Kohonen's SOM network and an incremental self-organizing map (ISOM) are examined comparatively. In order to increase the performance of heart sound classification, an incremental neural network is proposed in this study. It is observed that ISOM successfully classifies the HSs even in noisy environment.

A novel hardware-oriented Kohonen SOM image compression algorithm and its FPGA implementation

Kohonen self-organizing map (K-SOM) has proved to be suitable for lossy compression of digital images. The major drawback of the software implementation of this technique is its very computational intensive task. Fortunately, the structure is fairly easy to convert into hardware processing units executing in parallel. The resulting hardware system, however, consumes much of a microchip’s internal resources, i.e. slice registers and look-up table units. This results in utilising more than a single microchip to realize the structure in pure hardware implementation. Previously proposed K-SOM realizations were mainly targetted on implementing on an application specific integrated circuit (ASIC) with low restriction on resource utilization. In this paper, we propose an alternative architecture of K-SOM suitable for moderate density FPGAs with acceptable image quality and frame rate. In addition, its hardware architecture and synthesis results are presented. The proposed K-SOM algorithm compromises between the image quality, the frame rate throughput, the FPGA’s resource utilization and, additionally, the topological relationship among neural cells within the network. The architecture has been proved to be successfully synthesized on a single moderate resource FPGA with acceptable image quality and frame rate.

A study on two‐stage self‐organizing map and its application to clustering problems

AbstractThis paper presents a two‐stage self‐organizing map algorithm that we call two‐stage SOM which combines Kohonen's basic SOM (BSOM) and Aoki's SOM with threshold operation (THSOM). In the first stage of two‐stage SOM, we use BSOM algorithm in order to acquire topological structure of input data, and then we apply THSOM algorithm so that inactivated code vectors move to appropriate region reflecting the distribution of the input data. Furthermore, we show that two‐stage SOM can be applied to clustering problems. Some experimental results reveal that two‐stage SOM is effective for clustering problems in comparison with conventional methods. © 2007 Wiley Periodicals, Inc. Electr Eng Jpn, 159(1): 46–53, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.20268

Learning in human neural networks on microelectrode arrays

This paper describes experiments involving the growth of human neural networks of stem cells on a MEA (microelectrode array) support. The microelectrode arrays (MEAs) are constituted by a glass support in which a set of tungsten electrodes are inserted. The artificial neural network (ANN) paradigm was used by stimulating the neurons in parallel with digital patterns distributed on eight channels, then by analyzing a parallel multichannel output. In particular, the microelectrodes were connected following two different architectures, one inspired by the Kohonen's SOM, the other by the Hopfield network. The output signals have been analyzed in order to evaluate the possibility of organized reactions by the natural neurons.f The results show that the network of human neurons reacts selectively to the subministered digital signals, i.e., it produces similar output signals referred to identical or similar patterns, and clearly differentiates the outputs coming from different stimulations. Analyses performed with a special artificial neural network called ITSOM show the possibility to codify the neural responses to different patterns, thus to interpret the signals coming from the network of biological neurons, assigning a code to each output. It is straightforward to verify that identical codes are generated by the neural reactions to similar patterns. Further experiments are to be designed that improve the hybrid neural networks’ capabilities and to test the possibility of utilizing the organized answers of the neurons in several ways.

베이지안 네트워크와 신경망을 이용한 구매패턴 분석

소비자의 구매패턴을 분석하기 위해서는 문화적 요인, 사회적 요인, 개인적 요인, 심리적 요인 등을 고려해야 한다. 소비자의 구매에 이르는 심리나 내부 상태까지 깊이 고려하여 분석하게 되면 소비자의 구매행동을 정확하게 예측하거나 구매요인을 추측할 수가 있기 때문에 기업에서는 소비자의 기호에 맞는 상품개발에 효과적으로 이용할 수가 있다. 이러한 요인들은 직접적인 정보처리로 분석하기 어렵기 때문에 불확실한 정보를 취급하는 기술이 필요하다. 베이지안 네트워크는 불완전한 자료나 관측이 불가능한 자료에도 대응이 가능하며 구매자의 구매에 이르는 심리나 내적상태등과 같은 관측이 곤란한 요소도 다룰 수 있다. 따라서 본 논문에서는 상품구매에 따르는 소비자의 구매행동 패턴을 분석하기 위해 판매자의 노하우와 소비자의 구매의식을 조사하여 이 데이터를 바탕으로 베이지안 네트워크를 구성하고 구매패턴을 예측하는 방법을 제안하였다. 베이지안 네트워크를 이용하여 불필요한 속성을 가진 데이터를 제거하고 코호넨의 SOM을 이용하여 소비자의 구매패턴을 분석하였다. To analyze the consumer's purchase pattern, we must consider a factor which is a cultural, social, individual, psychological and so on. If we consider the internal state by the consumer's purchase, Both the consumer's purchase action and the purchase factor can be predicted, so the corporation can use effectively in suitable goods development in a consumer's preference. These factors need a technology that treat uncertain information, because it is difficult to analyze by directly information processing. Therefore, bayesian network manages elements those the observation of inner state such as consumer's purchase is difficult. In addition, it is interpretable about data that the observation is impossible. In this paper, we examine the seller's know-how and the way of consumer's purchase to analyze consumer's purchase action pattern through goods purchase. Also, we compose the bayesian network based on the examined data, and propose the method that predicts purchase patterns. Finally, we remove the data including unnecessary attribute using the bayesian network, and analyze the consumer's Purchase pattern using Kohonen's SOM method.

A Study on Two-stage Self-Organizing Map and Its Application to Clustering Problems

This paper presents a two-stage self-organizing map algorithm what we call Two-stage SOM which combines Kohonen's basic SOM (BSOM) and Aoki's SOM with threshold operation (THSOM). In the first stage of Two-stage SOM, we use BSOM algorithm in order to acquire topological structure of input data, and then we apply THSOM algorithm so that inactivated code-vectors move to appropriate region reflecting the distribution of the input data. Furthermore, we show that Two-stage SOM can be applied to clustering problems. Some experimental results reveal that Two-stage SOM is effective for clustering problems in comparison with conventional methods.

Decoding of auditory cortex signals with a LAMSTAR neural network

Objectives: Each neuron has a specific set of stimuli, which it preferentially responds to (the receptive field of the neuron). For implantable cortical prosthetic devices specific points of the cortex (or groups of neurons) have to be stimulated to create perceptions of sensory stimulus with specific attributes (such as frequency, temporal characteristics, etc). Such applications would need real time decoding of signals. Previously mathematical techniques, such as computing the receptive field (using electrophysiology data) and artificial neural networks (Kohonen network or SOM and back propagation network) have been used to decode neural signals.Methods: A Large Adaptive Memory Storage and Retrieval (LAMSTAR) neural-network-based decoder was designed to decode responses recorded from neurons in the auditory cortex. It was designed to identify the frequency of the tonal stimuli that elicited a particular discharge rate pattern recorded on two channels of a tungsten wire electrode array.Results: The network functioned efficiently as a decoder with 100% accuracy for the small sample of stimulus-response data used.Discussion: The results show that the network is effective in studying the functional organization of the auditory cortex and other sensory systems. Depending on the input sub-word, information about the kind of stimuli that activates particular parts of the sensory cortex can be studied.