Searcharxiv⌕ Search

arXiv subjects

Jin Xia

Publications and source records attributed to Jin Xia.

3 recordsLinked to original sources

Approximation numbers of Hankel Operators on Weighted Fock Spaces and the Berger--Coburn Phenomenon

Let $φ\in C^2(\mathbb{C}^n)$ be a real-valued weight whose real Hessian is uniformly positive definite and bounded. For $0<p<\infty$, $1\le q<\infty$, and possibly unbounded symbols $f$, we give necessary and sufficient conditions for prescribed decay of the approximation numbers of Hankel operators $H_f$ from weighted Fock spaces $F^p(φ)$ to weighted Lebesgue spaces $L^q(φ)$ in terms of the non-increasing rearrangement of the IDA function of $f$. We also characterize membership in the approximation ideals $\mathfrak A_s(F^p(φ),L^q(φ))$ and their weak counterparts $\mathfrak A_{s,\infty}(F^p(φ),L^q(φ))$ for $0<s<\infty$. The symbol criteria involve Lorentz, Lorentz--Zygmund, and Copson--Lorentz spaces. Our approach also yields the converse to the sufficient condition in [15], thereby completing the characterization of prescribed singular value decay for Hankel operators on a broad class of weighted Bergman spaces. For the pair $H_f$ and $H_{\bar f}$ on weighted Fock spaces, we characterize simultaneous decay of their approximation numbers and simultaneous membership in these ideals in terms of the mean oscillation function of $f$. For bounded symbols, we establish the Berger--Coburn phenomenon for both ideals in sharp exponent ranges. A new feature in the case $p<2<q$ is that the Berger--Coburn equivalence holds for $s_0(p,q)<s\le1$, where the sharp threshold $s_0(p,q)$ is strictly less than $1$.

math.FA↗

Asynchronous Interaction Aggregation for Action Detection

Understanding interaction is an essential part of video action detection. We propose the Asynchronous Interaction Aggregation network (AIA) that leverages different interactions to boost action detection. There are two key designs in it: one is the Interaction Aggregation structure (IA) adopting a uniform paradigm to model and integrate multiple types of interaction; the other is the Asynchronous Memory Update algorithm (AMU) that enables us to achieve better performance by modeling very long-term interaction dynamically without huge computation cost. We provide empirical evidence to show that our network can gain notable accuracy from the integrative interactions and is easy to train end-to-end. Our method reports the new state-of-the-art performance on AVA dataset, with 3.7 mAP gain (12.6% relative improvement) on validation split comparing to our strong baseline. The results on dataset UCF101-24 and EPIC-Kitchens further illustrate the effectiveness of our approach. Source code will be made public at: https://github.com/MVIG-SJTU/AlphAction .

cs.CV↗

Three Branches: Detecting Actions With Richer Features

We present our three branch solutions for International Challenge on Activity Recognition at CVPR2019. This model seeks to fuse richer information of global video clip, short human attention and long-term human activity into a unified model. We have participated in two tasks: Task A, the Kinetics challenge and Task B, spatio-temporal action localization challenge. For Kinetics, we achieve 21.59% error rate. For the AVA challenge, our final model obtains 32.49% mAP on the test sets, which outperforms all submissions to the AVA challenge at CVPR 2018 for more than 10% mAP. As the future work, we will introduce human activity knowledge, which is a new dataset including key information of human activity.

cs.CV↗