Tomoki Matsuno | Searcharxiv

arXiv2021

Anomalous Diffusion in a Monolayer of Lightweight Spheres Fluidized in Airflow

This paper presents statistical analyses of random motions in a single layer of fluidized lightweight spherical particles. Foam polystyrene spheres were driven by an upward airflow through the sieve mesh, and their two-dimensional motion was acquired using image analysis. In the bulk region, the particle velocity distributions changed from Gaussian to heavy-tailed distribution as the bulk packing fraction $ϕ_b$ was increased. The mean square displacement of the particles exhibited transition to subdiffusion at much lower $ϕ_b$ than observed in previous studies using similar setup but with heavier particles. A slight superdiffusion and significant growth of the correlation length in the two-body velocity correlation was observed at further large $ϕ_b$. The effect of the wall on the dynamics of the particles was also investigated and the anisotropy of the granular temperature was found to be a useful index to discriminate between the wall region and the bulk. The turbulence statistics in the wake of a particle indicated a strong wall-normal asymmetry of aerodynamic forcing as the ``thermal'' agitation in the wall region.

cond-mat.soft↗

arXiv2018

Reduction of Parameter Redundancy in Biaffine Classifiers with Symmetric and Circulant Weight Matrices

Currently, the biaffine classifier has been attracting attention as a method to introduce an attention mechanism into the modeling of binary relations. For instance, in the field of dependency parsing, the Deep Biaffine Parser by Dozat and Manning has achieved state-of-the-art performance as a graph-based dependency parser on the English Penn Treebank and CoNLL 2017 shared task. On the other hand, it is reported that parameter redundancy in the weight matrix in biaffine classifiers, which has O(n^2) parameters, results in overfitting (n is the number of dimensions). In this paper, we attempted to reduce the parameter redundancy by assuming either symmetry or circularity of weight matrices. In our experiments on the CoNLL 2017 shared task dataset, our model achieved better or comparable accuracy on most of the treebanks with more than 16% parameter reduction.

cs.CL↗