Deep neural network models have achieved remarkable results in sentiment classification. Traditional feature-based methods perform slightly worse than deep learning methods in terms of classification ...accuracy, but they have their own advantages in interpretability and time complexity. To the best of our knowledge, few works study the ensemble of deep learning methods and traditional feature-based methods. Inspired by the methodology of three-way decisions, we proposed a three-way enhanced convolutional neural network model named 3W-CNN. 3W-CNN can be seen as an ensemble method which uses the enhance model to optimize convolutional neural networks (CNN). The enhance model is selected according to the classification accuracy and the difference in classification results compared to CNN. Support vector machine with naive bayes features (NB-SVM) is selected as the enhance model after comparing with several baseline models. However, the performance of NB-SVM is worse than CNN on most of benchmark datasets. To address this issue, we construct a component named confidence divider and design a confidence function to distinguish the classification quality of CNN. NB-SVM is further utilized to reclassify the predictions with weak confidence. The experimental results validated the effectiveness of 3W-CNN and showed three-way decisions could further improve the accuracy of sentiment classification.
We study the asymptotic behavior of the Oleinik’s solution to the steady Prandtl equation when the outer flow
U
(
x
)
=
1
. Serrin proved that the Oleinik’s solution converges to the famous Blasius ...solution
u
¯
in
L
y
∞
sense as
x
→
+
∞
. The explicit decay estimates of
u
-
u
¯
and its derivatives were proved by Iyer (ARMA 237,2020) when the initial data is a small localized perturbation of the Blasius profile. In this paper, we prove the explicit decay estimate of
u
-
u
¯
for general initial data with exponential decay. We also prove the decay estimates of its derivatives when the initial data has an additional concave assumption. Our proof is based on the maximum principle techniques. The key ingredient is to find a series of barrier functions.
If I provide you a face image of mine (without telling you the actual age when I took the picture) and a large amount of face images that I crawled (containing labeled faces of different ages but not ...necessarily paired), can you show me what I would look like when I am 80 or what I was like when I was 5? The answer is probably a No. Most existing face aging works attempt to learn the transformation between age groups and thus would require the paired samples as well as the labeled query image. In this paper, we look at the problem from a generative modeling perspective such that no paired samples is required. In addition, given an unlabeled image, the generative model can directly produce the image with desired age attribute. We propose a conditional adversarial autoencoder (CAAE) that learns a face manifold, traversing on which smooth age progression and regression can be realized simultaneously. In CAAE, the face is first mapped to a latent vector through a convolutional encoder, and then the vector is projected to the face manifold conditional on age through a deconvolutional generator. The latent vector preserves personalized face features (i.e., personality) and the age condition controls progression vs. regression. Two adversarial networks are imposed on the encoder and generator, respectively, forcing to generate more photo-realistic faces. Experimental results demonstrate the appealing performance and flexibility of the proposed framework by comparing with the state-of-the-art and ground truth.
In this paper, we prove the linear inviscid damping and vorticity depletion phenomena for the linearized Euler equations around the Kolmogorov flow. These results confirm Bouchet and Morita's ...predictions based on numerical analysis. By using the wave operator method introduced by Li, Wei and Zhang, we solve Beck and Wayne's conjecture on the enhanced dissipation rate for the 2-D linearized Navier-Stokes equations around the bar state called Kolmogorov flow. The same dissipation rate is proved for the Navier-Stokes equations if the initial velocity is included in a basin of attraction of the Kolmogorov flow with the size of ν23+, here ν is the viscosity coefficient.
In this paper, we study the linear inviscid damping for the linearized
β
-plane equation around shear flows. We develop a new method to give the explicit decay rate of the velocity for a class of ...monotone shear flows. This method is based on the space-time estimate and the vector field method in the sprit of the wave equation. For general shear flows including the Sinus flow, we also prove the linear damping by establishing the limiting absorption principle, which is based on the compactness method introduced by Wei et al. (Ann PDE 5:3, 2019). The main difficulty is that the Rayleigh–Kuo equation has more singular points due to the Coriolis effects so that the compactness argument becomes more involved and delicate.
In this paper, we study the linearized Navier–Stokes system around monotone shear flows in a finite channel with non-slip boundary condition. We prove that if the flow is linearly stable for the ...Euler equations, then it is also linearly stable for the Navier–Stokes equations at high Reynolds number. More importantly, we establish the inviscid damping and enhanced dissipation estimates for the linearized Navier–Stokes system, which may be crucial for nonlinear stability. One of the key ingredients is the resolvent estimates of the linearized operator. For this, we develop the compactness method and establish some sharper estimates for the boundary layer corrector.
In this paper, we establish the Anderson localization, strong dynamical localization and the
(
1
2
-
)
-Hölder continuity of the integrated density of states (IDS) for some multi-dimensional discrete ...quasi-periodic (QP) Schrödinger operators with asymmetric
C
2
-cosine type potentials. We extend both the iteration scheme of Cao-Shi-Zhang (Commun Math Phys 404(1):495–561, 2023) and the interlacing method of Forman and VandenBoom (Localization and Cantor spectrum for quasiperiodic discrete Schrödinger operators with asymmetric, smooth, cosine-like sampling functions.
arXiv:2107.05461
, 2021) to handle
asymmetric
Rellich functions with
collapsed gaps
.
In this paper, we study the transition threshold problem for the 2-D Navier–Stokes equations around the Couette flow (
y
, 0) at high Reynolds number
Re
in a finite channel. We develop a systematic ...method to establish the resolvent estimates of the linearized operator and space-time estimates of the linearized Navier–Stokes equations. In particular, three kinds of important effects—enhanced dissipation, inviscid damping and a boundary layer–are integrated into the space-time estimates in a sharp form. As an application, we prove that if the initial velocity
v
0
satisfies
‖
v
0
-
(
y
,
0
)
‖
H
2
≦
c
R
e
-
1
2
for some small
c
independent of
Re
, then the solution of the 2-D Navier–Stokes equations remains within
O
(
R
e
-
1
2
)
of the Couette flow for any time.
In this paper, we study the multidimensional lattice Schrödinger operators with
C
2
-cosine like quasi-periodic (QP) potential. We establish quantitative Green’s function estimates, the arithmetic ...version of Anderson (and dynamical) localization, and the finite volume version of
(
1
2
-
)
-Hölder continuity of the integrated density of states for such QP Schrödinger operators. Our proof is based on an extension of the fundamental multi-scale analysis type method of Fröhlich–Spencer–Wittwer (Commun Math Phys 132(1), 5–25, 1990) to the higher lattice dimensions. We resolve the level crossing issue on eigenvalues parameterizations in the case of both higher lattice dimension and
C
2
regular potential.
Abstract
Parasite–host systems are pervasive in nature but are extremely difficult to convincingly identify in the fossil record. Here we report quantitative evidence of parasitism in the form of a ...unique, enduring life association between tube-dwelling organisms encrusted to densely clustered shells of a monospecific organophosphatic brachiopod assemblage from the lower Cambrian (Stage 4) of South China. Brachiopods with encrusting tubes have decreased biomass (indicating reduced fitness) compared to individuals without tubes. The encrusting tubes orient tightly in vectors matching the laminar feeding currents of the host, suggesting kleptoparasitism. With no convincing parasite–host interactions known from the Ediacaran, this widespread sessile association reveals intimate parasite–host animal systems arose in early Cambrian benthic communities and their emergence may have played a key role in driving the evolutionary and ecological innovations associated with the Cambrian radiation.
PDF
