Jiuwen Cao

Kezhi Mao

Amaury Lendasse

Jonathan Wu

This book contains some selected papers from the International Conference on Extreme Learning Machine 2015, which was held in Hangzhou, China, December 15-17, 2015. This conference brought together researchers and engineers to share and exchange R&D experience on both theoretical studies and practical applications of the Extreme Learning Machine (ELM) technique and brain learning. This book covers theories, algorithms ad applications of ELM. It gives readers a glance of the most recent advances of ELM.

Theory, Algorithms and Applications (I)

Proceedings of ELM-2015 Volume 1

Chi Man Vong

Yoan Miche

This book contains some selected papers from the International Conference on Extreme Learning Machine (ELM) 2017, held in Yantai, China, October 4–7, 2017. The book covers theories, algorithms and applications of ELM. Extreme Learning Machines (ELM) aims to enable pervasive learning and pervasive intelligence. As advocated by ELM theories, it is exciting to see the convergence of machine learning and biological learning from the long-term point of view. ELM may be one of the fundamental `learning particles’ filling the gaps between machine learning and biological learning (of which activation functions are even unknown). ELM represents a suite of (machine and biological) learning techniques in which hidden neurons need not be tuned: inherited from their ancestors or randomly generated. ELM learning theories show that effective learning algorithms can be derived based on randomly generated hidden neurons (biological neurons, artificial neurons, wavelets, Fourier series, etc) as long as they are nonlinear piecewise continuous, independent of training data and application environments. Increasingly, evidence from neuroscience suggests that similar principles apply in biological learning systems. ELM theories and algorithms argue that “random hidden neurons” capture an essential aspect of biological learning mechanisms as well as the intuitive sense that the efficiency of biological learning need not rely on computing power of neurons. ELM theories thus hint at possible reasons why the brain is more intelligent and effective than current computers. This conference will provide a forum for academics, researchers and engineers to share and exchange R&D experience on both theoretical studies and practical applications of the ELM technique and brain learning. It gives readers a glance of the most recent advances of ELM.

Proceedings of ELM-2017

Erik Cambria

This book contains some selected papers from the International Conference on Extreme Learning Machine 2016, which was held in Singapore, December 13-15, 2016. This conference will provide a forum for academics, researchers and engineers to share and exchange R&D experience on both theoretical studies and practical applications of the ELM technique and brain learning. Extreme Learning Machines (ELM) aims to break the barriers between the conventional artificial learning techniques and biological learning mechanism. ELM represents a suite of (machine or possibly biological) learning techniques in which hidden neurons need not be tuned. ELM learning theories show that very effective learning algorithms can be derived based on randomly generated hidden neurons (with almost any nonlinear piecewise activation functions), independent of training data and application environments. Increasingly, evidence from neuroscience suggests that similar principles apply in biological learning systems. ELM theories and algorithms argue that “random hidden neurons” capture an essential aspect of biological learning mechanisms as well as the intuitive sense that the efficiency of biological learning need not rely on computing power of neurons. ELM theories thus hint at possible reasons why the brain is more intelligent and effective than current computers. ELM offers significant advantages over conventional neural network learning algorithms such as fast learning speed, ease of implementation, and minimal need for human intervention. ELM also shows potential as a viable alternative technique for large‐scale computing and artificial intelligence. This book covers theories, algorithms ad applications of ELM. It gives readers a glance of the most recent advances of ELM.

Proceedings of ELM-2016

This book contains some selected papers from the International Conference on Extreme Learning Machine 2018, which was held in Singapore, November 21–23, 2018. This conference provided a forum for academics, researchers and engineers to share and exchange R&D experience on both theoretical studies and practical applications of the ELM technique and brain learning. Extreme Learning Machines (ELM) aims to enable pervasive learning and pervasive intelligence. As advocated by ELM theories, it is exciting to see the convergence of machine learning and biological learning from the long-term point of view. ELM may be one of the fundamental “learning particles” filling the gaps between machine learning and biological learning (of which activation functions are even unknown). ELM represents a suite of (machine and biological) learning techniques in which hidden neurons need not be tuned: inherited from their ancestors or randomly generated. ELM learning theories show that effective learning algorithms can be derived based on randomly generated hidden neurons (biological neurons, artificial neurons, wavelets, Fourier series, etc.) as long as they are nonlinear piecewise continuous, independent of training data and application environments. Increasingly, evidence from neuroscience suggests that similar principles apply in biological learning systems. ELM theories and algorithms argue that “random hidden neurons” capture an essential aspect of biological learning mechanisms as well as the intuitive sense that the efficiency of biological learning need not rely on computing power of neurons. ELM theories thus hint at possible reasons why the brain is more intelligent and effective than current computers. The main theme of ELM2018 is Hierarchical ELM, AI for IoT, Synergy of Machine Learning and Biological Learning. This book covers theories, algorithms and applications of ELM. It gives readers a glance at the most recent advances of ELM.

Proceedings of ELM 2018

L'analyse et la prédiction de séries temporelles sont des défis scientifiques importants, qui trouvent leurs applications dans des domaines aussi variés que la finance, la production électrique, l’hydrologie, la climatologie, etc. Comme ils englobent les modèles linéaires, les modèles non linéaires offrent potentiellement des performances supérieures mais ils posent cependant également des problèmes complexes tels que des minima locaux pour la fonction à optimiser, des temps de calcul très longs, une sélection de structure de modèle rendue plus difficile et une détermination de régresseur plus ardue. On définit tout d’abord la meilleure structure de modèle comme celle qui minimise une erreur de généralisation. Les différentes méthodes permettant d’estimer cette erreur sont présentées : Cross-Validation, Leave-One-Out, Bootstrap, etc. Une comparaison expérimentale de ces méthodes sur une série benchmark classique montre la supériorité des méthodes de Bootstrap. Une accélération de ces méthodes ainsi qu’une solution au problème des minima locaux sont apportées. Les différentes méthodes de détermination du meilleur régresseur, c’est-à-dire le vecteur d’entrées utilisé pour la prédiction, sont étudiées. Afin de donner une borne inférieure et une borne supérieure à la taille de ce régresseur, plusieurs interprétations du théorème de Takens sont formulées. Une méthodologie pratique permettant la construction d’un régresseur par projection non linéaire est proposée et illustrée sur des exemples de prédiction de séries financières. Des modèles non linéaires simples basés sur la quantification vectorielle sont développés. Ils proposent une alternative dont les performances et la complexité se situent entre celles des modèles linéaires et des modèles non linéaires classiques : les performances sont meilleures que celles obtenues avec des modèles linéaires mais les temps de calcul demandés sont moins importants que pour les autres modèles non linéaires. Finalement, une méthode d’analyse de données basée sur les cartes auto-organisées et l’algorithme de Ward est présentée. Cette méthode d’analyse débouche sur une méthode de prédiction à long terme dont les résultats sont évalués sur la prédiction, sur une durée de 24h, de la consommation électrique

Amaury Lendasse

Books

Your next favorite book is out there