From the Publisher:
This is the first comprehensive treatment of feed-forward neural networks from the perspective of statistical pattern recognition. After introducing the basic concepts, the book examines techniques for modelling probability density functions and the properties and merits of the multi-layer perceptron and radial basis function network models. Also covered are various forms of error functions, principal algorithms for error function minimalization, learning and generalization in neural networks, and Bayesian techniques and their applications. Designed as a text, with over 100 exercises, this fully up-to-date work will benefit anyone involved in the fields of neural computation and pattern recognition.

/pdf/neural-networks-for-pattern-recognition-34puigiau1.pdf

Neural networks for pattern recognition

A memory-based network that provides estimates of continuous variables and converges to the underlying (linear or nonlinear) regression surface is described. The general regression neural network (GRNN) is a one-pass learning algorithm with a highly parallel structure. It is shown that, even with sparse data in a multidimensional measurement space, the algorithm provides smooth transitions from one observed value to another. The algorithmic form can be used for any regression problem in which an assumption of linearity is not justified. >

A general regression neural network

A wavelet network concept, which is based on wavelet transform theory, is proposed as an alternative to feedforward neural networks for approximating arbitrary nonlinear functions. The basic idea is to replace the neurons by 'wavelons', i.e., computing units obtained by cascading an affine transform and a multidimensional wavelet. Then these affine transforms and the synaptic weights must be identified from possibly noise corrupted input/output data. An algorithm of backpropagation type is proposed for wavelet network training, and experimental results are reported. >

Wavelet networks

A standard approach to describe an image for classification and retrieval purposes is to extract a set of local patch descriptors, encode them into a high dimensional vector and pool them into an image-level signature The most common patch encoding strategy consists in quantizing the local descriptors into a finite set of prototypical elements This leads to the popular Bag-of-Visual words representation In this work, we propose to use the Fisher Kernel framework as an alternative patch encoding strategy: we describe patches by their deviation from an "universal" generative Gaussian mixture model This representation, which we call Fisher vector has many advantages: it is efficient to compute, it leads to excellent results even with efficient linear classifiers, and it can be compressed with a minimal loss of accuracy using product quantization We report experimental results on five standard datasets--PASCAL VOC 2007, Caltech 256, SUN 397, ILSVRC 2010 and ImageNet10K--with up to 9M images and 10K classes, showing that the FV framework is a state-of-the-art patch encoding technique

/pdf/image-classification-with-the-fisher-vector-theory-and-4a0lxo3dta.pdf

Image Classification with the Fisher Vector: Theory and Practice

Magnetic skyrmions are small swirling topological defects in the magnetization texture. Their stabilization and dynamics depend strongly on their topological properties. In most cases, they are induced by chiral interactions between atomic spins in non-centrosymmetric magnetic compounds or in thin films with broken inversion symmetry. Skyrmions can be extremely small, with diameters in the nanometre range, and behave as particles that can be moved, created and annihilated, which makes them suitable for ‘abacus’-type applications in information storage and logic technologies. Until recently, skyrmions had been observed only at low temperature and, in most cases, under large applied magnetic fields. An intense research effort has led to the identification of thin-film and multilayer structures in which skyrmions are now stable at room temperature and can be manipulated by electrical currents. The development of skyrmion-based topological spintronics holds promise for applications in the mid-term furure, even though many challenges, such as the achievement of writing, processing and reading functionalities at room temperature and in all-electrical manipulation schemes, still lie ahead. Magnetic skyrmions are topologically protected spin whirls that hold promise for applications because they can be controllably moved, created and annihilated. In this Review, the underlying physics of the stabilization of skyrmions at room temperature and their prospective use for spintronic applications are discussed.

Magnetic skyrmions: advances in physics and potential applications

A modified version of the PNN (probabilistic neural network) learning phase which allows a considerable simplification of network structure by including a vector quantization of learning data is proposed. It can be useful if large training sets are available. The procedure has been successfully tested in two synthetic data experiments. The proposed network has been shown to improve the classification performance of the LVQ (learning vector quantization) procedure. >

Learning vector quantization for the probabilistic neural network

Neural networks found significant applications in microwave CAD. In this paper, after providing a brief description of neural networks employed so far in this context, we illustrate some of their most significant applications and typical issues arising in practical implementation. We also summarize current research tendencies and introduce use of self-organizing maps enhancing model accuracy and applicability. We conclude considering some future developments and exciting perspectives opened from use of neural networks in microwave CAD. ©1999 John Wiley & Sons, Inc. Int J RF and Microwave CAE 9: 158–174, 1999.

A review of artificial neural networks applications in microwave computer‐aided design (invited article)

Magnetic skyrmions are topologically protected states that are very promising for the design of the next generation of ultra-low-power electronic devices. In this letter, we propose a magnetic tunnel junction based spin-transfer torque diode with a magnetic skyrmion as ground state and a perpendicular polarizer patterned as nano-contact for a local injection of the current. The key result is the possibility to achieve sensitivities (i.e., detection voltage over input microwave power) larger than 2000 V/W for optimized contact diameters. We also pointed out that large enough voltage controlled magnetocrystalline anisotropy could significantly improve the sensitivity. Our results can be very useful for the identification of a class of spin-torque diodes with a non-uniform ground state and to understand the fundamental physics of the skyrmion dynamical properties.

Skyrmion based microwave detectors and harvesting

Coded waveforms for optimised air-coupled ultrasonic nondestructive evaluation

This book deals with a number of fundamental issues related to the practical implementation of ultrasonic NDT techniques in an industrial environment. The book discusses advanced academic research results and their application to industrial procedures. The text covers the choice and generation of the signals energizing the system to probe position optimization, from quality assessment evaluation to tomographic inversion. With a focus to deepen a number of fundamental aspects involved in the specific objective of designing and developing an ultrasonic imaging system for nondestructive testing, aimed to automatically classify the entire production of an industrial production line, targeted to the field of precision mechanics. The contents of this book is the result of the common effort of six University Research Groups that focused their research activities for two years on this specific objective, working in direct conjunction with primary industrial firms, in a research project funded by the Italian government as a Strategic Research Project.

Pietro Burrascano

Papers

Learning vector quantization for the probabilistic neural network

A review of artificial neural networks applications in microwave computer‐aided design (invited article)

Skyrmion based microwave detectors and harvesting

Coded waveforms for optimised air-coupled ultrasonic nondestructive evaluation

Ultrasonic Nondestructive Evaluation Systems: Industrial Application Issues