Search results for " Neural network"
showing 10 items of 1232 documents
Attention-based Model for Evaluating the Complexity of Sentences in English Language
2020
The automation of text complexity evaluation (ATCE) is an emerging problem which has been tackled by means of different methodologies. We present an effective deep learning- based solution which leverages both Recurrent Neural and the Attention mechanism. The developed system is capable of classifying sentences written in the English language by analysing their syntactical and lexical complexity. An accurate test phase has been carried out, and the system has been compared with a baseline tool based on the Support Vector Machine. This paper represents an extension of a previous deep learning model, which allows showing the suitability of Neural Networks to evaluate sentence complexity in tw…
Deep neural attention-based model for the evaluation of italian sentences complexity
2020
In this paper, the Automatic Text Complexity Evaluation problem is modeled as a binary classification task tackled by a Neural Network based system. It exploits Recurrent Neural Units and the Attention mechanism to measure the complexity of sentences written in the Italian language. An accurate test phase has been carried out, and the system has been compared with state-of-art tools that tackle the same problem. The computed performances proof the model suitability to evaluate sentence complexity improving the results achieved by other state-of-the-art systems.
Multi-class Text Complexity Evaluation via Deep Neural Networks
2019
Automatic Text Complexity Evaluation (ATE) is a natural language processing task which aims to assess texts difficulty taking into account many facets related to complexity. A large number of papers tackle the problem of ATE by means of machine learning algorithms in order to classify texts into complex or simple classes. In this paper, we try to go beyond the methodologies presented so far by introducing a preliminary system based on a deep neural network model whose objective is to classify sentences into more of two classes. Experiments have been carried out on a manually annotated corpus which has been preprocessed in order to make it suitable for the scope of the paper. The results sho…
What represents a face? A computational approach for the integration of physiological and psychological data.
1997
Empirical studies of face recognition suggest that faces might be stored in memory by means of a few canonical representations. The nature of these canonical representations is, however, unclear. Although psychological data show a three-quarter-view advantage, physiological studies suggest profile and frontal views are stored in memory. A computational approach to reconcile these findings is proposed. The pattern of results obtained when different views, or combinations of views, are used as the internal representation of a two-stage identification network consisting of an autoassociative memory followed by a radial-basis-function network are compared. Results show that (i) a frontal and a…
Interpretability of Recurrent Neural Networks in Remote Sensing
2020
In this work we propose the use of Long Short-Term Memory (LSTM) Recurrent Neural Networks for multivariate time series of satellite data for crop yield estimation. Recurrent nets allow exploiting the temporal dimension efficiently, but interpretability is hampered by the typically overparameterized models. The focus of the study is to understand LSTM models by looking at the hidden units distribution, the impact of increasing network complexity, and the relative importance of the input covariates. We extracted time series of three variables describing the soil-vegetation status in agroe-cosystems -soil moisture, VOD and EVI- from optical and microwave satellites, as well as available in si…
Spiking Neural Networks models targeted for implementation on Reconfigurable Hardware
2017
La tesis presentada se centra en la denominada tercera generación de redes neuronales artificiales, las Redes Neuronales Spiking (SNN) también llamadas ‘de espigas’ o ‘de eventos’. Este campo de investigación se convirtió en un tema popular e importante en la última década debido al progreso de la neurociencia computacional. Las Redes Neuronales Spiking, que tienen no sólo la plasticidad espacial sino también temporal, ofrecen una alternativa prometedora a las redes neuronales artificiales clásicas (ANN) y están más cerca de la operación real de las neuronas biológicas ya que la información se codifica y transmite usando múltiples espigas o eventos en forma de trenes de pulsos. Este campo h…
Focal Cortical Lesions Induce Bidirectional Changes in the Excitability of Fast Spiking and Non Fast Spiking Cortical Interneurons
2014
A physiological brain function requires neuronal networks to operate within a well-defined range of activity. Indeed, alterations in neuronal excitability have been associated with several pathological conditions, ranging from epilepsy to neuropsychiatric disorders. Changes in inhibitory transmission are known to play a key role in the development of hyperexcitability. However it is largely unknown whether specific interneuronal subpopulations contribute differentially to such pathological condition. In the present study we investigated functional alterations of inhibitory interneurons embedded in a hyperexcitable cortical circuit at the border of chronically induced focal lesions in mouse …
Synaptic scaling generically stabilizes circuit connectivity
2011
Neural systems regulate synaptic plasticity avoiding overly strong growth or shrinkage of the connections, thereby keeping the circuit architecture operational. Accordingly, several experimental studies have shown that synaptic weights increase only in direct relation to their current value, resulting in reduced growth for stronger synapses [1]. It is, however, difficult to extract from these studies unequivocal evidence about the underlying biophysical mechanisms that control weight growth. The theoretical neurosciences have addressed this problem by exploring mechanisms for synaptic weight change that contain limiting factors to regulate growth [2]. The effectiveness of these mechanisms i…
Learning automata based energy-efficient AI hardware design for IoT applications
2020
Energy efficiency continues to be the core design challenge for artificial intelligence (AI) hardware designers. In this paper, we propose a new AI hardware architecture targeting Internet of Things applications. The architecture is founded on the principle of learning automata, defined using propositional logic. The logic-based underpinning enables low-energy footprints as well as high learning accuracy during training and inference, which are crucial requirements for efficient AI with long operating life. We present the first insights into this new architecture in the form of a custom-designed integrated circuit for pervasive applications. Fundamental to this circuit is systematic encodin…
Acoustic characterization of Silica aerogel clamped plates for perfect absorption purpose
2017
International audience; Silica aerogel has been widely studied as bulk material for its extremely low density and thermal conductivity. Plates or membranes made of this extremely soft materials exhibits interesting properties for sound absorption. A novel signal processing method for the characterization of an acoustic metamaterial made of silica aerogel clamped plates is presented. The acoustic impedance of a silica aerogel clamped plate is derived from the elastic theory for the flexural waves, while the transfer matrix method is used to model reflection and transmission coefficients of a single plate. Experimental results are obtained by using an acoustic impedance tube. The difference b…