Results 91 to 100 of about 16,081 (276)
Analysis of the Heavy-ball Algorithm using Integral Quadratic Constraints
, 2019 In this paper, we analyze the convergence rate of the Heavy-ball algorithm applied to optimize a class of continuously differentiable functions. The analysis is performed with the Heavy-ball tuned to achieve the best convergence rate on the sub-class of ...
Seiler, Peter +3 more
core +1 more source
Local Convergence of the Heavy-Ball Method and iPiano for Non-convex Optimization [PDF]
Journal of Optimization Theory and Applications, 2018 A local convergence result for abstract descent methods is proved. The sequence of iterates is attracted by a local (or global) minimum, stays in its neighborhood and converges within this neighborhood. This result allows algorithms to exploit local properties of the objective function.
openaire +3 more sources
Advanced Science, EarlyView.We uncover a large variety of putative inhibitory ligand‐gated ion channels (LGICs) in the phylum Cnidaria, the sister group to all bilaterian animals. Phylogenetic analysis suggests a complex evolutionary history of inhibitory LGICs with diverse neurotransmitter ligands.
Abhilasha Ojha +13 more
wiley +1 more source
Advanced Science, EarlyView.This study investigates how the internal structure of fiber‐reinforced ceramic composites affects their resistance to damage. By combining 3D X‐ray imaging with acoustic emission monitoring during mechanical testing, it reveals how silicon distribution influences crack formation.
Yang Chen +7 more
wiley +1 more source
NALA: a Nesterov accelerated look-ahead optimizer for deep learning [PDF]
PeerJ Computer ScienceAdaptive gradient algorithms have been successfully used in deep learning. Previous work reveals that adaptive gradient algorithms mainly borrow the moving average idea of heavy ball acceleration to estimate the first- and second-order moments of the ...
Xuan Zuo +4 more
doaj +2 more sources
Advanced Science, EarlyView.The pholidota‐inspired electronic skin is successfully constructed, originally realized four‐state reversible switching of terahertz‐wave reflection, absorption, transmission, and secondary reflection. In response to external demands, it issues deformation instructions, verifies the deformation state after the equipment executes the action, and thus ...
Shangjing Li +4 more
wiley +1 more source
AlgorithmsThe paper is devoted to the theoretical and numerical analysis of the two-step method, constructed as a modification of Polyak’s heavy ball method with the inclusion of an additional momentum parameter.
Gerasim V. Krivovichev +1 more
doaj +1 more source
Performance analysis of tabular and Fuzzy Q-learning under varying state and action space resolution [PDF]
International Journal of Electronics and TelecommunicationsReinforcement learning (RL) algorithms, such as Q-learning, are widely applied to control tasks involving continuous state spaces that require discretization or function approximation.
Roman Zajdel
doaj +1 more source
On convergence of gradient-dependent integrands [PDF]
, 1984 summary:We study convergence properties of $\lbrace v(\nabla u_k)\rbrace _{k\in \mathbb{N}}$ if $v\in C(\mathbb{R}^{m\times n})$, $|v(s)|\le C(1+|s|^p ...
Kružík, Martin +4 more
core +1 more source
Advanced Science, EarlyView.An orthogonal ionic‐liquid extraction (Orth‐iEA) enables selective isolation of amyloid fibrils. TMGBF4 disrupts hydrogen‐bonded β‐sheet networks to solubilize amyloid aggregates, whereas C12ImCl interacts with hydrophobic regions of non‐amyloid proteins.
Shiying Zheng +10 more
wiley +1 more source