Searcharxiv⌕ Search

arXiv subjects

Jun Chen

Publications and source records attributed to Jun Chen.

At least 271 records · Page 15Linked to original sources
arXiv2023

Two-orbital spin-fermion model study of ferromagnetism in honeycomb lattice

The spin-fermion model was previously successful to describe the complex phase diagrams of colossal magnetoresistive manganites and iron-based superconductors. In recent years, two-dimensional magnets have rapidly raised up as a new attractive branch of quantum materials, which are theoretically described based on classical spin models in most studies. Alternatively, here the two-orbital spin-fermion model is established as a uniform scenario to describe the ferromagnetism in a two-dimensional honeycomb lattice. This model connects the magnetic interactions with the electronic structures. Then the continuous tuning of magnetism in these honeycomb lattices can be predicted, based on a general phase diagram. The electron/hole doping, from the empty $e_{g}$ to half-filled $e_{g}$ limit, is studied as a benchmark. Our Monte Carlo result finds that the ferromagnetic $T_{C}$ reaches the maximum at the quarter-filled case. In other regions, the linear relationship between $T_{C}$ and doping concentration provides a theoretical guideline for the experimental modulations of two-dimensional ferromagnetism tuned by ionic liquid or electrical gating.

cond-mat.str-el↗
arXiv2023

Learning Global-aware Kernel for Image Harmonization

Image harmonization aims to solve the visual inconsistency problem in composited images by adaptively adjusting the foreground pixels with the background as references. Existing methods employ local color transformation or region matching between foreground and background, which neglects powerful proximity prior and independently distinguishes fore-/back-ground as a whole part for harmonization. As a result, they still show a limited performance across varied foreground objects and scenes. To address this issue, we propose a novel Global-aware Kernel Network (GKNet) to harmonize local regions with comprehensive consideration of long-distance background references. Specifically, GKNet includes two parts, \ie, harmony kernel prediction and harmony kernel modulation branches. The former includes a Long-distance Reference Extractor (LRE) to obtain long-distance context and Kernel Prediction Blocks (KPB) to predict multi-level harmony kernels by fusing global information with local features. To achieve this goal, a novel Selective Correlation Fusion (SCF) module is proposed to better select relevant long-distance background references for local harmonization. The latter employs the predicted kernels to harmonize foreground regions with both local and global awareness. Abundant experiments demonstrate the superiority of our method for image harmonization over state-of-the-art methods, \eg, achieving 39.53dB PSNR that surpasses the best counterpart by +0.78dB $\uparrow$; decreasing fMSE/MSE by 11.5\%$\downarrow$/6.7\%$\downarrow$ compared with the SoTA method. Code will be available at \href{https://github.com/XintianShen/GKNet}{here}.

cs.CV↗
arXiv2023

Unified Data-Free Compression: Pruning and Quantization without Fine-Tuning

Structured pruning and quantization are promising approaches for reducing the inference time and memory footprint of neural networks. However, most existing methods require the original training dataset to fine-tune the model. This not only brings heavy resource consumption but also is not possible for applications with sensitive or proprietary data due to privacy and security concerns. Therefore, a few data-free methods are proposed to address this problem, but they perform data-free pruning and quantization separately, which does not explore the complementarity of pruning and quantization. In this paper, we propose a novel framework named Unified Data-Free Compression(UDFC), which performs pruning and quantization simultaneously without any data and fine-tuning process. Specifically, UDFC starts with the assumption that the partial information of a damaged(e.g., pruned or quantized) channel can be preserved by a linear combination of other channels, and then derives the reconstruction form from the assumption to restore the information loss due to compression. Finally, we formulate the reconstruction error between the original network and its compressed network, and theoretically deduce the closed-form solution. We evaluate the UDFC on the large-scale image classification task and obtain significant improvements over various network architectures and compression methods. For example, we achieve a 20.54% accuracy improvement on ImageNet dataset compared to SOTA method with 30% pruning ratio and 6-bit quantization on ResNet-34.

cs.LG↗
arXiv2023

Quasicrystalline Nanocrystal Superlattice with Partial Matching Rules

Expanding the library of self-assembled superstructures provides insight into the behavior of atomic crystals and supports the development of materials with mesoscale order. Here we build upon recent findings of soft matter quasicrystals and report a quasicrystalline binary nanocrystal superlattice that exhibits correlations in the form of partial matching rules reducing tiling disorder. We determine a three-dimensional structure model through electron tomography and direct imaging of surface topography. The 12-fold rotational symmetry of the quasicrystal is broken in sub-layers, forming a random tiling of rectangles, large triangles, and small triangles with 6-fold symmetry. We analyze the geometry of the experimental tiling and discuss factors relevant for the stabilization of the quasicrystal. Our joint experimental-computational study demonstrates the power of nanocrystal superlattice engineering and further narrows the gap between the richness of crystal structures found with atoms and in soft matter assemblies.

cond-mat.mtrl-sci↗
arXiv2023

A Weakly Supervised Segmentation Network Embedding Cross-scale Attention Guidance and Noise-sensitive Constraint for Detecting Tertiary Lymphoid Structures of Pancreatic Tumors

The presence of tertiary lymphoid structures (TLSs) on pancreatic pathological images is an important prognostic indicator of pancreatic tumors. Therefore, TLSs detection on pancreatic pathological images plays a crucial role in diagnosis and treatment for patients with pancreatic tumors. However, fully supervised detection algorithms based on deep learning usually require a large number of manual annotations, which is time-consuming and labor-intensive. In this paper, we aim to detect the TLSs in a manner of few-shot learning by proposing a weakly supervised segmentation network. We firstly obtain the lymphocyte density maps by combining a pretrained model for nuclei segmentation and a domain adversarial network for lymphocyte nuclei recognition. Then, we establish a cross-scale attention guidance mechanism by jointly learning the coarse-scale features from the original histopathology images and fine-scale features from our designed lymphocyte density attention. A noise-sensitive constraint is introduced by an embedding signed distance function loss in the training procedure to reduce tiny prediction errors. Experimental results on two collected datasets demonstrate that our proposed method significantly outperforms the state-of-the-art segmentation-based algorithms in terms of TLSs detection accuracy. Additionally, we apply our method to study the congruent relationship between the density of TLSs and peripancreatic vascular invasion and obtain some clinically statistical results.

eess.IV↗
arXiv2023

MC-SpEx: Towards Effective Speaker Extraction with Multi-Scale Interfusion and Conditional Speaker Modulation

The previous SpEx+ has yielded outstanding performance in speaker extraction and attracted much attention. However, it still encounters inadequate utilization of multi-scale information and speaker embedding. To this end, this paper proposes a new effective speaker extraction system with multi-scale interfusion and conditional speaker modulation (ConSM), which is called MC-SpEx. First of all, we design the weight-share multi-scale fusers (ScaleFusers) for efficiently leveraging multi-scale information as well as ensuring consistency of the model's feature space. Then, to consider different scale information while generating masks, the multi-scale interactive mask generator (ScaleInterMG) is presented. Moreover, we introduce ConSM module to fully exploit speaker embedding in the speech extractor. Experimental results on the Libri2Mix dataset demonstrate the effectiveness of our improvements and the state-of-the-art performance of our proposed MC-SpEx.

cs.SD↗
arXiv2023

The fate of reentrant localization phenomenon in the one-dimensional dimerized quasiperiodic chain with long-range hopping

Recently, the exciting reentrant localization transition phenomenon was found in a one-dimensional dimerized lattice with staggered quasiperiodic potentials. Usually, long-range hopping is typically important in actual physical systems. In this work, we study the effect of next-nearest neighbor hopping (NNNH) on the reentrant localization phenomenon. Due to the presence of NNNH, the broken chiral symmetry is further enhanced and the localization properties of electron states in the upper and lower bands become quite different. It is found that the reentrant localization can still persist within a range of NNNH both in Hermitian and non-Hermitian cases. Eventually, the reentrant localization disappears as the strength of NNNH increases to some extent, since the increasing NNNH weakens the dimerization of the system and destroys its competition with the quasiperiodic disorder. Our work thus reveals the effect of long-range hopping in the reentrant localization phenomenon and deepens its physical understanding.

cond-mat.dis-nn↗
arXiv2023

Rotating BTZ-like black hole and central charges in Einstein-bumblebee gravity

We obtain an exact rotating BTZ-like black hole solution by solving the corresponding gravitational field equations and the bumblebee motion equations in Einstein-bumblebee gravity theory. Result is presented for the purely radial Lorentz symmetry violating and can only exist with a linear functional potential of the bumblebee field. This black hole has two horizons and an ergosphere which are dependent on the bumblebee coupling constant $\ell$. The concepts of the area and volume of the horizon should be renewed in this LV spacetime due to the nontrivial contribution of coupling between the bumblebee field and the Ricci tensor. Only in this way, the entropy-area relation, first law of thermodynamics and the Smarr formula can still be constructed. We also study the AdS/CFT correspondence of this black hole, find that the entropy product of its inner and outer horizons is universal. So the central charges of the dual CFT on the boundary can be obtained via the thermodynamic method, and they can reappear black hole mass and angular momentum in the bulk.

gr-qc↗
arXiv2023

AV-SepFormer: Cross-Attention SepFormer for Audio-Visual Target Speaker Extraction

Visual information can serve as an effective cue for target speaker extraction (TSE) and is vital to improving extraction performance. In this paper, we propose AV-SepFormer, a SepFormer-based attention dual-scale model that utilizes cross- and self-attention to fuse and model features from audio and visual. AV-SepFormer splits the audio feature into a number of chunks, equivalent to the length of the visual feature. Then self- and cross-attention are employed to model and fuse the multi-modal features. Furthermore, we use a novel 2D positional encoding, that introduces the positional information between and within chunks and provides significant gains over the traditional positional encoding. Our model has two key advantages: the time granularity of audio chunked feature is synchronized to the visual feature, which alleviates the harm caused by the inconsistency of audio and video sampling rate; by combining self- and cross-attention, feature fusion and speech extraction processes are unified within an attention paradigm. The experimental results show that AV-SepFormer significantly outperforms other existing methods.

cs.MM↗
arXiv2023

Impacts of seasonality and parasitism on honey bee population dynamics

The honeybee plays an extremely important role in ecosystem stability and diversity and in the production of bee pollinated crops. Honey bees and other pollinators are under threat from the combined effects of nutritional stress, parasitism, pesticides, and climate change that impact the timing, duration, and variability of seasonal events. To understand how parasitism and seasonality influence honey bee colonies separately and interactively, we developed a non-autonomous nonlinear honeybee-parasite interaction differential equation model that incorporates seasonality into the egg-laying rate of the queen. Our theoretical results show that parasitism negatively impacts the honey bee population either by decreasing colony size or destabilizing population dynamics through supercritical or subcritical Hopf-bifurcations depending on conditions. Our bifurcation analysis and simulations suggest that seasonality alone may have positive or negative impacts on the survival of honey bee colonies. More specifically, our study indicates that (1) the timing of the maximum egg-laying rate seems to determine when seasonality has positive or negative impacts; and (2) when the period of seasonality is large it can lead to the colony collapsing. Our study further suggests that the synergistic influences of parasitism and seasonality can lead to complicated dynamics that may positively and negatively impact the honey bee colony's survival. Our work partially uncovers the intrinsic effects of climate change and parasites, which potentially provide essential insights into how best to maintain or improve a honey bee colony's health.

q-bio.PE↗
arXiv2023

Scheduling Coflows for Minimizing the Makespan in Identical Parallel Networks

With the rapid advancement of technology, parallel computing applications have become increasingly popular and are commonly executed in large data centers. These applications involve two phases: computation and communication, which are executed repeatedly to complete the work. However, due to the ever-increasing demand for computing power, large data centers are struggling to meet the massive communication demands. To address this problem, coflow has been proposed as a networking abstraction that captures communication patterns in data-parallel computing frameworks. This paper focuses on the coflow scheduling problem in identical parallel networks, where the primary objective is to minimize the makespan, which is the maximum completion time of coflows. It is considered one of the most significant $\mathcal{NP}$-hard problems in large data centers. In this paper, we consider two problems: flow-level scheduling and coflow-level scheduling. In the flow-level scheduling problem, distinct flows can be transferred through different network cores, whereas in the coflow-level scheduling problem, all flows must be transferred through the same network core. To address the flow-level scheduling problem, this paper proposes two algorithms: a $(3-\tfrac{2}{m})$-approximation algorithm and a $(\tfrac{8}{3}-\tfrac{2}{3m})$-approximation algorithm, where $m$ represents the number of network cores. For the coflow-level scheduling problem, this paper proposes a $(2m)$-approximation algorithm. Finally, we conduct simulations on our proposed algorithm and Weaver's algorithm, as presented in Huang \textit{et al.} (2020) in the 2020 IEEE International Parallel and Distributed Processing Symposium (IPDPS). We also validate the effectiveness of the proposed algorithms on heterogeneous parallel networks.

cs.DS↗
arXiv2023

Gesper: A Restoration-Enhancement Framework for General Speech Reconstruction

This paper describes a real-time General Speech Reconstruction (Gesper) system submitted to the ICASSP 2023 Speech Signal Improvement (SSI) Challenge. This novel proposed system is a two-stage architecture, in which the speech restoration is performed, and then cascaded by speech enhancement. We propose a complex spectral mapping-based generative adversarial network (CSM-GAN) as the speech restoration module for the first time. For noise suppression and dereverberation, the enhancement module is performed with fullband-wideband parallel processing. On the blind test set of ICASSP 2023 SSI Challenge, the proposed Gesper system, which satisfies the real-time condition, achieves 3.27 P.804 overall mean opinion score (MOS) and 3.35 P.835 overall MOS, ranked 1st in both track 1 and track 2.

cs.SD↗
arXiv2023

Reentrant quantum spin Hall effect in the presence of an exchange field

Driven by various physical origins, the interesting reentrant phenomena in quantum Hall effect (QHE), quantum anomalous Hall effect (QAHE) and non-Hermitian systems have been discussed recently. Here, we present that the reentrant phenomena of quantum spin Hall effect (QSHE) can be achieved by introducing a uniform exchange field into a topological insulator with a four-band model Hamiltonian. Furthermore, it is found that both time-reversal (TR) symmetry broken QSHE and spin-polarized QAHE can simultaneously be generated in the system with proper exchange field strength by inspecting the corresponding edge states. Through analyzing the energy dependent spin-Chern number $ν_s$ and transmission in the presence of random disorders, we verify the topological protection and the robustness of induced QSHE and QAHE. Finally, the theoretical realization of reentrant QSHE in a topolectrical circuit is demonstrated. Our work not only extends the reentrant phenomena into the QSHE but also stimulates interest in their realization in different systems.

cond-mat.mes-hall↗
arXiv2023

LLM as A Robotic Brain: Unifying Egocentric Memory and Control

Embodied AI focuses on the study and development of intelligent systems that possess a physical or virtual embodiment (i.e. robots) and are able to dynamically interact with their environment. Memory and control are the two essential parts of an embodied system and usually require separate frameworks to model each of them. In this paper, we propose a novel and generalizable framework called LLM-Brain: using Large-scale Language Model as a robotic brain to unify egocentric memory and control. The LLM-Brain framework integrates multiple multimodal language models for robotic tasks, utilizing a zero-shot learning approach. All components within LLM-Brain communicate using natural language in closed-loop multi-round dialogues that encompass perception, planning, control, and memory. The core of the system is an embodied LLM to maintain egocentric memory and control the robot. We demonstrate LLM-Brain by examining two downstream tasks: active exploration and embodied question answering. The active exploration tasks require the robot to extensively explore an unknown environment within a limited number of actions. Meanwhile, the embodied question answering tasks necessitate that the robot answers questions based on observations acquired during prior explorations.

cs.AI↗
arXiv2023

A Model for Confined Solar Eruptions Including External Reconnection

The violent disruption of the coronal magnetic field is often observed to be restricted to the low corona, appearing as a confined eruption. The possible causes of the confinement remain elusive. Here, we model the eruption of a magnetic flux rope in a quadrupolar active region, with the parameters set such that magnetic X-lines exist both below and above the rope. This facilitates the onset of magnetic reconnection in either place but with partly opposing effects on the eruption. The lower reconnection initially adds poloidal flux to the rope, increasing the upward hoop force and supporting the rise of the rope. However, when the flux of the magnetic side lobes enters the lower reconnection, the flux rope is found to separate from the reconnection site and the flux accumulation ceases. At the same time, the upper reconnection begins to reduce the poloidal flux of the rope, decreasing its hoop force; eventually this cuts the rope completely. The relative weight of the two reconnection processes is varied in the model, and it is found that their combined effect and the tension force of the overlying field confine the eruption if the flux ratio of the outer to the inner polarities exceeds a threshold, which is about 1.3 for our Cartesian box and chosen parameters. We hence propose that external reconnection between an erupting flux rope and overlying flux can play a vital role in confining eruptions.

astro-ph.SR↗
arXiv2023

Exploring Open-Vocabulary Semantic Segmentation without Human Labels

Semantic segmentation is a crucial task in computer vision that involves segmenting images into semantically meaningful regions at the pixel level. However, existing approaches often rely on expensive human annotations as supervision for model training, limiting their scalability to large, unlabeled datasets. To address this challenge, we present ZeroSeg, a novel method that leverages the existing pretrained vision-language (VL) model (e.g. CLIP) to train open-vocabulary zero-shot semantic segmentation models. Although acquired extensive knowledge of visual concepts, it is non-trivial to exploit knowledge from these VL models to the task of semantic segmentation, as they are usually trained at an image level. ZeroSeg overcomes this by distilling the visual concepts learned by VL models into a set of segment tokens, each summarizing a localized region of the target image. We evaluate ZeroSeg on multiple popular segmentation benchmarks, including PASCAL VOC 2012, PASCAL Context, and COCO, in a zero-shot manner (i.e., no training or adaption on target segmentation datasets). Our approach achieves state-of-the-art performance when compared to other zero-shot segmentation methods under the same training data, while also performing competitively compared to strongly supervised methods. Finally, we also demonstrated the effectiveness of ZeroSeg on open-vocabulary segmentation, through both human studies and qualitative visualizations.

cs.CV↗
arXiv2023

MammalNet: A Large-scale Video Benchmark for Mammal Recognition and Behavior Understanding

Monitoring animal behavior can facilitate conservation efforts by providing key insights into wildlife health, population status, and ecosystem function. Automatic recognition of animals and their behaviors is critical for capitalizing on the large unlabeled datasets generated by modern video devices and for accelerating monitoring efforts at scale. However, the development of automated recognition systems is currently hindered by a lack of appropriately labeled datasets. Existing video datasets 1) do not classify animals according to established biological taxonomies; 2) are too small to facilitate large-scale behavioral studies and are often limited to a single species; and 3) do not feature temporally localized annotations and therefore do not facilitate localization of targeted behaviors within longer video sequences. Thus, we propose MammalNet, a new large-scale animal behavior dataset with taxonomy-guided annotations of mammals and their common behaviors. MammalNet contains over 18K videos totaling 539 hours, which is ~10 times larger than the largest existing animal behavior dataset. It covers 17 orders, 69 families, and 173 mammal categories for animal categorization and captures 12 high-level animal behaviors that received focus in previous animal behavior studies. We establish three benchmarks on MammalNet: standard animal and behavior recognition, compositional low-shot animal and behavior recognition, and behavior detection. Our dataset and code have been made available at: https://mammal-net.github.io.

cs.CV↗
arXiv2023

Photogalvanic effect induced charge and spin photocurrent in group-V monolayer systems

Photogalvanic effect (PGE) occurs in materials with non-centrosymmetric structures when irradiated by linearly or circularly polarized light. Here, using non-equilibrium Green's function combined with density functional theory (NEGF-DFT), we investigated the linear photogalvanic effect (LPGE) in monolayers of group-V elements (As, Sb, and Bi) by first-principles calculations. First, by designing a two-probe structure based on the group-V elements, we found a giant anisotropy photoresponse of As between the armchair and zigzag directions. Then, we analyzed Sb and Bi's charge and spin photocurrent characteristics when considering the spin-orbit coupling (SOC) effect. It is found that when the polarization direction of linearly polarized light is parallel or perpendicular to the transport direction ($θ$ = $0^ \circ$ or $90^ \circ$), the spin up and spin down photoresponse in the armchair direction has the same magnitude and direction, leading to the generation of net charge current. However, in the zigzag direction, the spin up and spin down photoresponse have the same magnitude with opposite directions, leading to the generation of pure spin current. Furthermore, it is understood by analyzing the bulk spin photovoltaic (BSPV) coefficient from the symmetry point of view. Finally, we found that the net charge current generated in the armchair direction and the pure spin current generated in the zigzag direction can be further tuned with the increase of the material's buckling height $|h|$. Our results highlight that these group-V monolayers are promising candidates for novel functional materials, which will provide a broad prospect for the realization of ultrathin ferroelectric devices in optoelectronics due to their spontaneous polarization characteristics and high Curie temperature.

cond-mat.mtrl-sci↗