1,721,020 research outputs found
A VLSI ‘rest’ processing element with improved bus connections (an evaluation)
No full textA highly parallel architecture based on ‘’ type processor elements () and designed to make good use of Very Large Scale Integration technology is proposed. The system is realized with 512 : the special structure of the single , each provided with the proper control firmware, allows to specialize, in the bootstrap phase, the functionality, implementing dynamically both and structures. The layout of a C-mos transmission-gate, for a special sub-unit (su), is presented. The single-cell transmission-gate, realized with 3 um technology, has a dimension of 68 um. The propagation time evaluation for the single cell is 3.2 us. As an example of the use of this structure the algorithm for computing the product of a 512∗512 matrix by a vector of 512 elements is also shown: the execution time for this algorithm is about 1 ms
On Designing Semantic Lexicon-Based Architectures for Web Information Retrieval
No full textIn this work, a novel framework for designing Web Information Retrieval systems with particular reference to semantic search engines is presented. The key idea is to add the semantic dimension to the classical Term-Document Matrix thus having a three-dimensional dataset. This enhancement allows for defining a lexico-semantic user interface where the query process is performed at the conceptual level thanks to the use of a Semantic Lexicon. WordNet Semantic Lexicon is used here as golden ontology for handling polysemy and synonymy, hence it is useful for disambiguating user queries at the semantic level. A layered multi-agent system is employed for supporting the design process. Particular emphasis is given to formal system knowledge representation, the interface layer managing user-system interaction and the markup layer performing the semantic tagging process
Manuale teorico pratico su “Analisi chimiche per Fluorescenza RX ed elaborazione al calcolatore dei dati per la correzione degli effetti di matrice”
No full textMultiscale modeling of neuromorphic computing: From materials to device operations
No full textIn this paper, a multiscale modeling platform for neuromorphic computing devices connecting the atomic material properties to the electrical device performances is presented. The main ingredients of the modeling platform are discussed in view of the different technologies (e.g. RRAM, PCM, FTJ) proposed for 3D integrated neuromorphic computing
M-DUST: an Innovative Low-Cost Smart PM Sensor
No full textIn this work, we present M-DUST, a novel low-cost and real-time smart monitoring sensor for Particulate Matter (PM) emission measurement. It is based on the Tyndall scattering process to count particles concentration. A comparison on different methods to evaluate particles concentration has been discussed. A mechanical filter is used to select the particulate matter with the appropriate cut-off aerodynamic diameter. The presented device is an intelligent sensor thanks to its features, such as: ability to make self-diagnosis, self-adaptation and transparency to communication interface. Tests are carried out in the Italian city of Taranto by using non-toxic substances and analysis chamber
Technical image reduction using N.N. and Wavelet
No full textA general-purpose procedure for scaling technical line drawings, suitable for video presentation, is described in this paper. The proposed method is based on the separate processing of scaleable (layout) and non-scaleable (symbol) elements, drawn from standard technical drafting. Symbols, detected by a cluster-based template procedure and a Minimum Distance Classifier, are extracted from drawings and utilized to form a symbols position table. To obtain the clusters of symbols, a Rival Penalized Competitive Learning neural network and a human template labeling procedure have been adopted. The extraction of symbols from drawings produces clear layouts. These layouts are scaled down by wavelet based algorithm and the symbols are then restored or replaced, through the symbols position table, with different graphs or textual representations, according to the scaling factors and the display device. The results of an experimental study on a large database of technical drawing are presented and the accuracy of the system is discussed
Improving Classification Performance Using the Semi-pivoted QR Approximation Algorithm
No full textAim of this paper is to present a method to improve the classification performance of a Fuzzy C-means based classifier. The obtained results show that this method can improve the performance of the classifier both in terms of computational efficiency (by reducing the amount of data to be analyzed) and in terms of classification error rate. The proposed method is based on the Semi-Pivoted QR approximation (SPQR) algorithm. It reduces a numeric dataset (a matrix) to its more important features (where each feature is a column of the matrix). The framework discussed in this article can be used by researchers and practi-tioners to set up high-performance machine learning system
A proposal of a new technique for meteorological data quality evaluation
No full textAim of this paper is to propose the use of a Generative Adversarial Imputation Nets (GAIN) to evaluate the quality of time series of environmental data. These time series are becoming more and more popular because they are used by many services available online such as weather forecasting, weather alert systems, etc. The performance of these systems is strictly correlated to the quality of their input data. For this reason, the quality of meteorological time series is becoming a key point for many services.The GAIN method is used to impute missing data in meteorological time series derived by two different kinds of sensors (terrestrial weather station and virtual sensor). The quality of the rebuilt series can be used as a measure of the quality of the original one
Modelling An Urban Traffic Network By A Consumer/Producer Architecture
No full textIn this paper a proposal for modelling urban traffic network is presented. The proposed traffic model is specified by investigating the possible isomorphism between traffic system and operating system. In this short paper we show that such an analogy is consistent and it could be helpful when dealing with dynamic network loading. In fact, with respect to a given urban traffic network (region of interest), we assume each road (link) to act as resource consumer or as buffer. The relationships with the environment outside the region of interest is modelled by a set of external producers (centroids). The work is just at the beginning and the research on practical performances of such an approach are on going
- …
