China
Africa
Explore chapters and articles related to this topic
On the combination of simulated annealing and semi-supervised clustering for intrusion detection
Published in Amir Hussain, Mirjana Ivanovic, Electronics, Communications and Networks IV, 2015
Furthermore, in order to evaluate the semisupevised learning performance, a supervised intrusion detection algorithm (LVQIR) based on Learning Vector Quantization (LVQ) is designed. LVQ is a supervised learning neural network based on the structure of the competitive learning network (Hu et al. 2013). The LVQ applied for intrusion detection is designed as three-layer structure. The input node number of the input layer is set as 41 corresponding to the network attribute properties. And the output layer has two nodes which represent the normal data and the intrusion data. The node number of the middle layer is set at 80 by trial and error. The LVQ1 is adopted as the learning algorithm. When LVQ is used for intrusion detection, the training network based on the training set is firstly established. Then the test data are input into the network. According to the network output, the normal or abnormal data can be finally judged.
Artificial Neural Networks in Urologic Oncology
Published in Raouf N.G. Naguib, Gajanan V. Sherbet, Artificial Neural Networks in Cancer Diagnosis, Prognosis, and Patient Management, 2001
In a related study, the same authors used a learning vector quantiser (LVQ)-type neural network on the same cellular and patient datasets [31]. An LVQ neural network is optimally designed for classification or pattern recognition tasks. This network is unsupervised and self-organises to cluster similar input patterns together to provide a learned output classification of that data. Using this network, they achieved results similar to their back-propagation network with 90.6% accuracy at predicting cellular classification, and 97.5% accuracy at predicting diagnosis on these patients. These studies demonstrate the neural network’s usefulness in complex image interpretation by assisting the pathologist in the cumbersome task of interpreting nuclear cellular architecture.
Machine Learning (ML) Algorithms for Enabling the Cognitive Internet of Things (CIoT)
Published in Pethuru Raj, Anupama C. Raman, Harihara Subramanian, Cognitive Internet of Things, 2022
Pethuru Raj, Anupama C. Raman, Harihara Subramanian
A downside of KNN worth mentioning is that we need to use the entire training data set whenever a prediction is required. The LVQ is an artificial neural network (ANN) algorithm that allows us to choose how many training instances are needed and learn what those instances should be doing. LVQ is best suited for classification problems. It supports both binary (two-class) and multi-class classification problems based on prototype supervised learning. LVQ has two layers: the input layer and the other is the output layer. The best practice is that if you discover that KNN gives good results on your data set, then try using LVQ to reduce the memory usage.
Predictive modelling of injury severity in bicycle–motor vehicle collisions utilizing learning vector quantization: a case study of Britain’s cycling capital
Published in International Journal of Crashworthiness, 2021
Meisam Siamidoudaran, Mehdi Siamidodaran, Ersun Iscioglu
In this study LVQ classification algorithm was used for cycle rider injury severity prediction in bicycle – MVCs as well as for recognition of significant predictors in the accidents. LVQ is special case of a feed forward ANN based on supervised learning which can be modelled as a competitive neural network. The model was invented by Teuvo Kohonen [17]. LVQ model allows one to select the number of training instances to take into account and pinpoints precisely how such instances need to appear. Throughout the learning process, the significance in terms of the number of instances is continuously improved. LVQ provides winner-take-all Hebbian learning-based method to improve the classification accuracy in pattern recognition problems. In fact, LVQ is a precursor to self-organizing maps (SOM), with the difference being in SOM against LVQ suits superior to clustering, but then again LVQ is a supervised learning technique and that can be utilised when labelled input data such as the current study is devised.