Results 121 to 130 of about 580,652 (288)

Neural Networks and the Natural Gradient

, 2010
Neural network training algorithms have always suffered from the problem of local minima. The advent of natural gradient algorithms promised to overcome this shortcoming by finding better local minima. However, they require additional training parameters
Bastian, Michael R.
core   +1 more source

Pyramidal Structures on Yttria‐Stabilized Zirconia after High Temperature Exposure at 1500°C: New Features on an Old Material

Advanced Engineering Materials, EarlyView.
New features on yttria‐stabilized zirconia after exposure at 1500°C: Newly discovered pyramidal structures on an old material. After exposure at 1550°C on the cross section of YSZ new features, namely pyramidal structures are discovered. These structures grow with time, increase in numbers, appear as singularities, are often arranged in strings, and ...
Doris Sebold, Dylan Jennings, Robert Vaßen   +2 more
wiley   +1 more source

Natural Gradient Deep Q-learning

CoRR, 2018
We present a novel algorithm to train a deep Q-learning agent using natural-gradient techniques. We compare the original deep Q-network (DQN) algorithm to its natural-gradient counterpart, which we refer to as NGDQN, on a collection of classic control domains.
Ethan Knight, Osher Lerner
openaire   +2 more sources

Performance Analysis of Abradable Coating Systems for Aircraft Gas Turbines

Advanced Engineering Materials, EarlyView.
Three CoNiCrAlY/YSZ/MgAl2O4 abradable liner configurations on a nickel‐superalloy are evaluated by thermal‐gradient cycling and incursion tests. Laser ablation of the bondcoat and/or Y2O3‐stabilized ZrO2 (YSZ) intermediate layer increases mechanical interlocking and bonding for thick topcoats.
Hanna Heyl, Daniel Emil Mack, Martin Tandler, Susanne Schrüfer, Robert Vaßen   +4 more
wiley   +1 more source

On the Locality of the Natural Gradient for Deep Learning

CoRR, 2020
We study the natural gradient method for learning in deep Bayesian networks, including neural networks. There are two natural geometries associated with such learning systems consisting of visible and hidden units. One geometry is related to the full system, the other one to the visible sub-system. These two geometries imply different natural gradients.
openaire   +2 more sources

Transients and Asymptotics of Natural Gradient Learning [PDF]

, 1998
We analyse natural gradient learning in a two-layer feed-forward neural network using a statistical mechanics framework which is appropriate for large input dimension. We find significant improvement over standard gradient descent in both the transient and asymptotic phases of learning.
Rattray, Magnus, Saad, David
openaire   +2 more sources

Additive Manufacturing of Continuous Fibre Reinforced Composites: Process, Characterisation, Modelling, and Sustainability

Advanced Engineering Materials, EarlyView.
Additive manufacturing provides precise control over the placement of continuous fibres within polymer matrices, enabling customised mechanical performance in composite components. This article explores processing strategies, mechanical testing, and modelling approaches for additive manufactured continuous fibre‐reinforced composites.
Cherian Thomas, Amir Hosein Sakhaei
wiley   +1 more source

Advanced biomimetic design strategies for porous structures promoting bone integration with additive-manufactured Ti6Al4V scaffolds

Journal of Materials Research and Technology
The natural bone structure exhibits a radial gradient with dense cortical bone externally and porous cancellous bone internally. However, previous studies proposing scaffold designs have predominantly focused on homogeneous porous structures.
Yongyue Li, Qing Han, Hao Chen, Wenbo Yang, Yongjun Xu, Yongqi Zhang, Jiangbo Zhang, Li Liu, Weilong Zhang, Hao Liu, Bingpeng Chen, Jincheng Wang   +11 more
doaj   +1 more source

Recurrent Natural Policy Gradient for POMDPs

CoRR
Solving partially observable Markov decision processes (POMDPs) remains a fundamental challenge in reinforcement learning (RL), primarily due to the curse of dimensionality induced by the non-stationarity of optimal policies. In this work, we study a natural actor-critic (NAC) algorithm that integrates recurrent neural network (RNN) architectures into ...
Semih Cayci, Atilla Eryilmaz
openaire   +3 more sources

Fatigue Crack Initiation and Growth in Nanocrystalline Ni at Multiple Length‐Scales

Advanced Engineering Materials, EarlyView.
Overview of miniaturized in situ SEM fatigue setup and resultant fatigue crack growth data for nanocrystalline Ni. The presented study focuses on the analysis of fatigue crack growth rate (FCGR) in focused ion beam‐notched microcantilevers prepared from nanocrystalline (NC) Ni as a model material.
Igor Moravcik, Alexander Jelinek, Markus Alfreider, Jutta Luksch, Anton Hohenwarter, Florian Schaefer, Christian Motz, Daniel Kiener   +7 more
wiley   +1 more source