Luisa Helena Bartocci Liboni

Artificial Neural Network Architectures and Training Processes

The architecture of an artificial neural network defines how its several neurons are arranged, or placed, in relation to each other. These arrangements are structured essentially by directing the synaptic connections of the neurons.

Multilayer Perceptron Networks

Multilayer Perceptron (MLP) network features, at least, one intermediate (hidden) neural layer, which is placed between the input layer and the respective output layer.

Forecast of Stock Market Trends Using Recurrent Networks

Conventional methods for predicting the behavior of financial papers are based on specialist’s analysis and decision-making, and automatic methods are unavailable for most situations.

Computational Tools for Data Processing in Smart Cities

Smart Grids provide many benefits for society. Reliability, observability across the energy distribution system and the exchange of information between devices are just some of the features that make Smart Grids so attractive. One of the main products of a Smart Grid is to data. The amount of data available nowadays increases fast and carries several kinds of information. Smart metres allow engineers to perform multiple measurements and analyse such data. For example, information about consumption, power quality and digital protection, among others, can be extracted. However, the main challenge in extracting information from data arises from the data quality. In fact, many sectors of the society can benefit from such data. Hence, this information needs to be properly stored and readily available. In this chapter, we will address the main concepts involving Technology Information, Data Mining, Big Data and clustering for deploying information on Smart Grids.

Coffee Global Quality Estimation Using Multilayer Perceptron

This application uses artificial neural networks for qualifying coffee batches or coffee brands from a set of sensors based on conductive polymers, which were developed by EMBRAPA (Brazilian Agricultural Research Corporation).

Recognition of Disturbances Related to Electric Power Quality Using MLP Networks

The technical and scientific evaluation of power quality (PQ) had a great advance after the use of intelligent systems .

Pattern Identification of Adulterants in Coffee Powder Using Kohonen Self-organizing Map

The purpose of this application is to identify adulterant patterns found in samples of roasted ground coffee using the Kohonen network.

The Perceptron Network

The Perceptron , created by Rosenblatt , is the simplest configuration of an artificial neural network ever created, whose purpose was to implement a computational model based on the retina, aiming an element for electronic perception. One application of the Perceptron was to identify geometric patterns.

ART (Adaptive Resonance Theory) Networks

The Adaptive resonance theory (ART), initially proposed by Grossberg (1976a, b), was developed from the observation of some biological phenomena, regarding vision, speech, cortical development, and cognitive-emotional interactions.

The ADALINE Network and Delta Rule

The ADALINE (Adaptive Linear Element) was created by Widrow and Hoff in 1960. Its main application was in switching circuits of telephone networks, which was one of the first industrial applications that effectively involved artificial neural networks (Widrow and Hoff 1960).