Magazine et Edition

Volume 12 (2022): Edition 3 (July 2022)

Volume 12 (2022): Edition 2 (April 2022)

Volume 12 (2022): Edition 1 (January 2022)

Volume 11 (2021): Edition 4 (October 2021)

Volume 11 (2021): Edition 3 (July 2021)

Volume 11 (2021): Edition 2 (April 2021)

Volume 11 (2021): Edition 1 (January 2021)

Volume 10 (2020): Edition 4 (October 2020)

Volume 10 (2020): Edition 3 (July 2020)

Volume 10 (2020): Edition 2 (April 2020)

Volume 10 (2020): Edition 1 (January 2020)

Volume 9 (2019): Edition 4 (October 2019)

Volume 9 (2019): Edition 3 (July 2019)

Volume 9 (2019): Edition 2 (April 2019)

Volume 9 (2019): Edition 1 (January 2019)

Volume 8 (2018): Edition 4 (October 2018)

Volume 8 (2018): Edition 3 (July 2018)

Volume 8 (2018): Edition 2 (April 2018)

Volume 8 (2018): Edition 1 (January 2018)

Volume 7 (2017): Edition 4 (October 2017)

Volume 7 (2017): Edition 3 (July 2017)

Volume 7 (2017): Edition 2 (April 2017)

Volume 7 (2017): Edition 1 (January 2017)

Volume 6 (2016): Edition 4 (October 2016)

Volume 6 (2016): Edition 3 (July 2016)

Volume 6 (2016): Edition 2 (April 2016)

Volume 6 (2016): Edition 1 (January 2016)

Volume 5 (2015): Edition 4 (October 2015)

Volume 5 (2015): Edition 3 (July 2015)

Volume 5 (2015): Edition 2 (April 2015)

Volume 5 (2015): Edition 1 (January 2015)

Volume 4 (2014): Edition 4 (October 2014)

Volume 4 (2014): Edition 3 (July 2014)

Volume 4 (2014): Edition 2 (April 2014)

Volume 4 (2014): Edition 1 (January 2014)

Volume 3 (2013): Edition 4 (October 2013)

Volume 3 (2013): Edition 3 (July 2013)

Volume 3 (2013): Edition 2 (April 2013)

Volume 3 (2013): Edition 1 (January 2013)

Détails du magazine
Format
Magazine
eISSN
2449-6499
Première publication
30 Dec 2014
Période de publication
4 fois par an
Langues
Anglais

Chercher

Volume 8 (2018): Edition 3 (July 2018)

Détails du magazine
Format
Magazine
eISSN
2449-6499
Première publication
30 Dec 2014
Période de publication
4 fois par an
Langues
Anglais

Chercher

5 Articles
Accès libre

One-Match-Ahead Forecasting in Two-Team Sports with Stacked Bayesian Regressions

Publié en ligne: 09 Feb 2018
Pages: 159 - 171

Résumé

Abstract

There is a growing interest in applying machine learning algorithms to real-world examples by explicitly deriving models based on probabilistic reasoning. Sports analytics, being favoured mostly by the statistics community and less discussed in the machine learning community, becomes our focus in this paper. Specifically, we model two-team sports for the sake of one-match-ahead forecasting. We present a pioneering modeling approach based on stacked Bayesian regressions, in a way that winning probability can be calculated analytically. Benefiting from regression flexibility and high standard of performance, Sparse Spectrum Gaussian Process Regression (SSGPR) – an improved algorithm for the standard Gaussian Process Regression (GPR), was used to solve Bayesian regression tasks, resulting in a novel predictive model called TLGProb. For evaluation, TLGProb was applied to a popular sports event – National Basketball Association (NBA). Finally, 85.28% of the matches in NBA 2014/2015 regular season were correctly predicted by TLGProb, surpassing the existing predictive models for NBA.

Mots clés

  • Sports analytics
  • one-match-ahead forecasting
  • winning probability
  • Gaussian process regression
Accès libre

Soft Computing Tools for Virtual Drug Discovery

Publié en ligne: 09 Feb 2018
Pages: 173 - 189

Résumé

Abstract

In this paper, we describe how several soft computing tools can be used to assist in high throughput screening of potential drug candidates. Individual small molecules (ligands) are assessed for their potential to bind to specific proteins (receptors). Committees of multilayer networks are used to classify protein-ligand complexes as good binders or bad binders, based on selected chemical descriptors. The novel aspects of this paper include the use of statistical analyses on the weights of single layer networks to select the appropriate descriptors, the use of Monte Carlo cross-validation to provide confidence measures of network performance (and also to identify problems in the data), the addition of new chemical descriptors to improve network accuracy, and the use of Self Organizing Maps to analyze the performance of the trained network and identify anomalies. We demonstrate the procedures on a large practical data set, and use them to discover a promising characteristic of the data. We also perform virtual screenings with the trained networks on a number of benchmark sets and analyze the results.

Mots clés

  • drug discovery
  • virtual screening
  • multilayer network
  • SOM
Accès libre

An Environment for Collective Perception based on Fuzzy and Semantic Approaches

Publié en ligne: 09 Feb 2018
Pages: 191 - 210

Résumé

Abstract

This work proposes a software environment implementing a methodology for acquiring and exploiting the collective perception (CP) of Points of Interests (POIs) in a Smart City, which is meant to support decision makers in urban planning and management. This environment relies upon semantic knowledge discovery techniques and fuzzy computational approaches, including natural language processing, sentiment analysis, POI signatures and Fuzzy Cognitive Maps, turning them into a cohesive architectural blend in order to effectively gather the realistic perception of a user community towards given areas and attractions of a Smart City. The environment has been put to the test via a thorough experimentation against a massive user base of an online community with respect to a large metropolitan city (the City of Naples). Such an experimentation yielded consistent results, useful for providing decision makers with a clear awareness of the positive as well as critical aspects of urban areas, and thus helping them shape the measures to be taken for an improved city management and development.

Mots clés

  • smart cities
  • fuzzy logic
  • text mining
  • sentiment analysis
Accès libre

Effect of Strategy Adaptation on Differential Evolution in Presence and Absence of Parameter Adaptation: An Investigation

Publié en ligne: 09 Feb 2018
Pages: 211 - 235

Résumé

Abstract

Differential Evolution (DE) is a simple, yet highly competitive real parameter optimizer in the family of evolutionary algorithms. A significant contribution of its robust performance is attributed to its control parameters, and mutation strategy employed, proper settings of which, generally lead to good solutions. Finding the best parameters for a given problem through the trial and error method is time consuming, and sometimes impractical. This calls for the development of adaptive parameter control mechanisms. In this work, we investigate the impact and efficacy of adapting mutation strategies with or without adapting the control parameters, and report the plausibility of this scheme. Backed with empirical evidence from this and previous works, we first build a case for strategy adaptation in the presence as well as in the absence of parameter adaptation. Afterwards, we propose a new mutation strategy, and an adaptive variant SA-SHADE which is based on a recently proposed self-adaptive memory based variant of Differential evolution, SHADE. We report the performance of SA-SHADE on 28 benchmark functions of varying complexity, and compare it with the classic DE algorithm (DE/Rand/1/bin), and other state-of-the-art adaptive DE variants including CoDE, EPSDE, JADE, and SHADE itself. Our results show that adaptation of mutation strategy improves the performance of DE in both presence, and absence of control parameter adaptation, and should thus be employed frequently.

Mots clés

  • Evolutionary algorithms
  • Differential evolution
  • mutation strategy
  • adaptive control
Accès libre

Complex-Valued Associative Memories with Projection and Iterative Learning Rules

Publié en ligne: 09 Feb 2018
Pages: 237 - 249

Résumé

Abstract

In this paper, we investigate the stability of patterns embedded as the associative memory distributed on the complex-valued Hopfield neural network, in which the neuron states are encoded by the phase values on a unit circle of complex plane. As learning schemes for embedding patterns onto the network, projection rule and iterative learning rule are formally expanded to the complex-valued case. The retrieval of patterns embedded by iterative learning rule is demonstrated and the stability for embedded patterns is quantitatively investigated.

Mots clés

  • complex-valued neural networks
  • associative memory
  • projection
5 Articles
Accès libre

One-Match-Ahead Forecasting in Two-Team Sports with Stacked Bayesian Regressions

Publié en ligne: 09 Feb 2018
Pages: 159 - 171

Résumé

Abstract

There is a growing interest in applying machine learning algorithms to real-world examples by explicitly deriving models based on probabilistic reasoning. Sports analytics, being favoured mostly by the statistics community and less discussed in the machine learning community, becomes our focus in this paper. Specifically, we model two-team sports for the sake of one-match-ahead forecasting. We present a pioneering modeling approach based on stacked Bayesian regressions, in a way that winning probability can be calculated analytically. Benefiting from regression flexibility and high standard of performance, Sparse Spectrum Gaussian Process Regression (SSGPR) – an improved algorithm for the standard Gaussian Process Regression (GPR), was used to solve Bayesian regression tasks, resulting in a novel predictive model called TLGProb. For evaluation, TLGProb was applied to a popular sports event – National Basketball Association (NBA). Finally, 85.28% of the matches in NBA 2014/2015 regular season were correctly predicted by TLGProb, surpassing the existing predictive models for NBA.

Mots clés

  • Sports analytics
  • one-match-ahead forecasting
  • winning probability
  • Gaussian process regression
Accès libre

Soft Computing Tools for Virtual Drug Discovery

Publié en ligne: 09 Feb 2018
Pages: 173 - 189

Résumé

Abstract

In this paper, we describe how several soft computing tools can be used to assist in high throughput screening of potential drug candidates. Individual small molecules (ligands) are assessed for their potential to bind to specific proteins (receptors). Committees of multilayer networks are used to classify protein-ligand complexes as good binders or bad binders, based on selected chemical descriptors. The novel aspects of this paper include the use of statistical analyses on the weights of single layer networks to select the appropriate descriptors, the use of Monte Carlo cross-validation to provide confidence measures of network performance (and also to identify problems in the data), the addition of new chemical descriptors to improve network accuracy, and the use of Self Organizing Maps to analyze the performance of the trained network and identify anomalies. We demonstrate the procedures on a large practical data set, and use them to discover a promising characteristic of the data. We also perform virtual screenings with the trained networks on a number of benchmark sets and analyze the results.

Mots clés

  • drug discovery
  • virtual screening
  • multilayer network
  • SOM
Accès libre

An Environment for Collective Perception based on Fuzzy and Semantic Approaches

Publié en ligne: 09 Feb 2018
Pages: 191 - 210

Résumé

Abstract

This work proposes a software environment implementing a methodology for acquiring and exploiting the collective perception (CP) of Points of Interests (POIs) in a Smart City, which is meant to support decision makers in urban planning and management. This environment relies upon semantic knowledge discovery techniques and fuzzy computational approaches, including natural language processing, sentiment analysis, POI signatures and Fuzzy Cognitive Maps, turning them into a cohesive architectural blend in order to effectively gather the realistic perception of a user community towards given areas and attractions of a Smart City. The environment has been put to the test via a thorough experimentation against a massive user base of an online community with respect to a large metropolitan city (the City of Naples). Such an experimentation yielded consistent results, useful for providing decision makers with a clear awareness of the positive as well as critical aspects of urban areas, and thus helping them shape the measures to be taken for an improved city management and development.

Mots clés

  • smart cities
  • fuzzy logic
  • text mining
  • sentiment analysis
Accès libre

Effect of Strategy Adaptation on Differential Evolution in Presence and Absence of Parameter Adaptation: An Investigation

Publié en ligne: 09 Feb 2018
Pages: 211 - 235

Résumé

Abstract

Differential Evolution (DE) is a simple, yet highly competitive real parameter optimizer in the family of evolutionary algorithms. A significant contribution of its robust performance is attributed to its control parameters, and mutation strategy employed, proper settings of which, generally lead to good solutions. Finding the best parameters for a given problem through the trial and error method is time consuming, and sometimes impractical. This calls for the development of adaptive parameter control mechanisms. In this work, we investigate the impact and efficacy of adapting mutation strategies with or without adapting the control parameters, and report the plausibility of this scheme. Backed with empirical evidence from this and previous works, we first build a case for strategy adaptation in the presence as well as in the absence of parameter adaptation. Afterwards, we propose a new mutation strategy, and an adaptive variant SA-SHADE which is based on a recently proposed self-adaptive memory based variant of Differential evolution, SHADE. We report the performance of SA-SHADE on 28 benchmark functions of varying complexity, and compare it with the classic DE algorithm (DE/Rand/1/bin), and other state-of-the-art adaptive DE variants including CoDE, EPSDE, JADE, and SHADE itself. Our results show that adaptation of mutation strategy improves the performance of DE in both presence, and absence of control parameter adaptation, and should thus be employed frequently.

Mots clés

  • Evolutionary algorithms
  • Differential evolution
  • mutation strategy
  • adaptive control
Accès libre

Complex-Valued Associative Memories with Projection and Iterative Learning Rules

Publié en ligne: 09 Feb 2018
Pages: 237 - 249

Résumé

Abstract

In this paper, we investigate the stability of patterns embedded as the associative memory distributed on the complex-valued Hopfield neural network, in which the neuron states are encoded by the phase values on a unit circle of complex plane. As learning schemes for embedding patterns onto the network, projection rule and iterative learning rule are formally expanded to the complex-valued case. The retrieval of patterns embedded by iterative learning rule is demonstrated and the stability for embedded patterns is quantitatively investigated.

Mots clés

  • complex-valued neural networks
  • associative memory
  • projection

Planifiez votre conférence à distance avec Sciendo