Search results for "algorithm."
showing 10 items of 4617 documents
Active Learning for Monitoring Network Optimization
2012
Kernel-based active learning strategies were studied for the optimization of environmental monitoring networks. This chapter introduces the basic machine learning algorithms originated in the statistical learning theory of Vapnik (1998). Active learning is closer to an optimization done using sequential Gaussian simulations. The chapter presents the general ideas of statistical learning from data. It derives the basics of kernel-based support vector algorithms. The active learning framework is presented and machine learning extensions for active learning are described in the chapter. Kernel-based active learning strategies are tested on real case studies. The chapter explores the use of a c…
Remote sensing image segmentation by active queries
2012
Active learning deals with developing methods that select examples that may express data characteristics in a compact way. For remote sensing image segmentation, the selected samples are the most informative pixels in the image so that classifiers trained with reduced active datasets become faster and more robust. Strategies for intelligent sampling have been proposed with model-based heuristics aiming at the search of the most informative pixels to optimize model's performance. Unlike standard methods that concentrate on model optimization, here we propose a method inspired in the cluster assumption that holds in most of the remote sensing data. Starting from a complete hierarchical descri…
Is the nonREM–REM sleep cycle reset by forced awakenings from REM sleep?
2002
In selective REM sleep deprivation (SRSD), the occurrence of stage REM is repeatedly interrupted by short awakenings. Typically, the interventions aggregate in clusters resembling the REM episodes in undisturbed sleep. This salient phenomenon can easily be explained if the nonREM–REM sleep process is continued during the periods of forced wakefulness. However, earlier studies have alternatively suggested that awakenings from sleep might rather discontinue and reset the ultradian process. Theoretically, the two explanations predict a different distribution of REM episode duration. We evaluated 117 SRSD treatment nights recorded from 14 depressive inpatients receiving low dosages of Trimipram…
A fuzzy ranking strategy for portfolio selection applied to the Spanish stock market
2007
In this paper we present a fuzzy ranking procedure for the portfolio selection problem. The uncertainty on the returns of each portfolio is approximated by means of a trapezoidal fuzzy number. The expected return and risk of the portfolio are then characteristics of that fuzzy number. A rank index that accounts for both expected return and risk is defined, allowing the decision-maker to compare different portfolios. The paper ends with an application of that fuzzy ranking strategy to the Spanish stock market.
FPGA Implementation of an Adaptive Filter Robust to Impulsive Noise: Two Approaches
2011
Adaptive filters are used in a wide range of applications such as echo cancellation, noise cancellation, system identification, and prediction. Its hardware implementation becomes essential in many cases where real-time execution is needed. However, impulsive noise affects the proper operation of the filter and the adaptation process. This noise is one of the most damaging types of signal distortion, not always considered when implementing algorithms, particularly in specific hardware platforms. Field-programmable gate arrays (FPGAs) are used widely for real-time applications where timing requirements are strict. Nowadays, two main design processes can be followed for embedded system design…
Acceleration of image filtering algorithms for 3D visualization of murine lungs using dataflow engines
2015
Image filtering is one of the most common and important tasks in image processing applications. In this paper, image processing using a mean filtering algorithm combined with thresholding and binarization algorithms for the 3D visualization and analysis of murine lungs is explained. These algorithms are then mapped on the Maxler's MAX2336B Dataflow Engine (DFE) to significantly increase calculation speed. Several different DFE configurations were tested and each yielded different performance characteristics. Optimal algorithm calculation speed was up to 30 fold baseline calculation speed.
Predicting the Short-Term Exchange Rate Between United State Dollar and Czech Koruna Using Hilbert-Huang Transform and Fuzzy Logic
2017
In this paper, the combination of the Hilbert-Huang Transform, fuzzy logic and an embedding theorem is described to predict the short-term exchange rate from United States dollar to Czech Koruna. By Using the Hilbert-Huang Transform as an adaptive filter, the proposed method decreases the embedding dimension space from five (original samples) to four (de-noising samples). This dimension space provides the number of inputs to the fuzzy rule base system, which causes the number of rules, the time for training and the inference process to decrease. Experimental results indicated that this method achieves higher accuracy prediction than the direct use of original data.
Adaptive Kernel Learning for Signal Processing
2018
Adaptive filtering is a central topic in digital signal processing (DSP). By applying linear adaptive filtering principles in the kernel feature space, powerful nonlinear adaptive filtering algorithms can be obtained. This chapter introduces the wide topic of adaptive signal processing, and explores the emerging field of kernel adaptive filtering (KAF). In many signal processing applications, the problem of signal estimation is addressed. Probabilistic models have proven to be very useful in this context. The chapter discusses two families of kernel adaptive filters, namely kernel least mean squares (KLMS) and kernel recursive least‐squares (KRLS) algorithms. In order to design a practical …
Adaptive scheduling of acceleration and gyroscope for motion artifact cancelation in photoplethysmography
2022
Background and objective: Recently, various algorithms have been introduced using wrist-worn photo-plethysmography (PPG) to provide high accuracy of instantaneous heart rate (HR) estimation, including during high-intensity exercise. Most studies focus on using acceleration and/or gyroscope signals for the motion artifact (MA) reference, which attenuates or cancels out noise from the MA-corrupted PPG signals. We aim to open and pave the path to find an appropriate MA reference selection for MA cancelation in PPG.Methods: We investigated how the acceleration and gyroscope reference signals correlate with the MAs of the distorted PPG signals and derived both mathematically and experimentally a…
Determining the Parameters of a Sugeno Fuzzy Controller Using a Parallel Genetic Algorithm
2013
Developed in the mid 1970s, the technique based on genetic algorithms proved its usefulness in finding optimal or near optimal solutions to problems for which accurate solving strategies are either non-existent or require excessively long running time. We implemented a genetic algorithm to determine the parameters of a Sugeno fuzzy controller for the Truck Backer-Upper problem (This problem is considered an acknowledged benchmark in nonlinear system identification.). Less known at first than Mamdami fuzzy controllers, Sugeno fuzzy controllers became popular once they were included into the ANFIS neuro-fuzzy Matlab library. By their nature, Sugeno controllers can be regarded as interpolation…