arXiv subjects

Zhen Li

Publications and source records attributed to Zhen Li.

At least 307 records · Page 17Linked to original sources

arXiv2023

An Effective Motion-Centric Paradigm for 3D Single Object Tracking in Point Clouds

3D single object tracking in LiDAR point clouds (LiDAR SOT) plays a crucial role in autonomous driving. Current approaches all follow the Siamese paradigm based on appearance matching. However, LiDAR point clouds are usually textureless and incomplete, which hinders effective appearance matching. Besides, previous methods greatly overlook the critical motion clues among targets. In this work, beyond 3D Siamese tracking, we introduce a motion-centric paradigm to handle LiDAR SOT from a new perspective. Following this paradigm, we propose a matching-free two-stage tracker M^2-Track. At the 1st-stage, M^2-Track localizes the target within successive frames via motion transformation. Then it refines the target box through motion-assisted shape completion at the 2nd-stage. Due to the motion-centric nature, our method shows its impressive generalizability with limited training labels and provides good differentiability for end-to-end cycle training. This inspires us to explore semi-supervised LiDAR SOT by incorporating a pseudo-label-based motion augmentation and a self-supervised loss term. Under the fully-supervised setting, extensive experiments confirm that M^2-Track significantly outperforms previous state-of-the-arts on three large-scale datasets while running at 57FPS (~3%, ~11% and ~22% precision gains on KITTI, NuScenes, and Waymo Open Dataset respectively). While under the semi-supervised setting, our method performs on par with or even surpasses its fully-supervised counterpart using fewer than half of the labels from KITTI. Further analysis verifies each component's effectiveness and shows the motion-centric paradigm's promising potential for auto-labeling and unsupervised domain adaptation.

cs.CV↗

arXiv2023

Robin: A Novel Method to Produce Robust Interpreters for Deep Learning-Based Code Classifiers

Deep learning has been widely used in source code classification tasks, such as code classification according to their functionalities, code authorship attribution, and vulnerability detection. Unfortunately, the black-box nature of deep learning makes it hard to interpret and understand why a classifier (i.e., classification model) makes a particular prediction on a given example. This lack of interpretability (or explainability) might have hindered their adoption by practitioners because it is not clear when they should or should not trust a classifier's prediction. The lack of interpretability has motivated a number of studies in recent years. However, existing methods are neither robust nor able to cope with out-of-distribution examples. In this paper, we propose a novel method to produce \underline{Rob}ust \underline{in}terpreters for a given deep learning-based code classifier; the method is dubbed Robin. The key idea behind Robin is a novel hybrid structure combining an interpreter and two approximators, while leveraging the ideas of adversarial training and data augmentation. Experimental results show that on average the interpreter produced by Robin achieves a 6.11\% higher fidelity (evaluated on the classifier), 67.22\% higher fidelity (evaluated on the approximator), and 15.87x higher robustness than that of the three existing interpreters we evaluated. Moreover, the interpreter is 47.31\% less affected by out-of-distribution examples than that of LEMNA.

cs.SE↗

arXiv2023

Thermodynamic efficiency of atmospheric motion governed by Lorenz system

The Lorenz system was derived on the basis of a model of convective atmospheric motions and may serve as a paradigmatic model for considering a complex climate system. In this study, we formulated the thermodynamic efficiency of convective atmospheric motions governed by the Lorenz system by treating it as a non-equilibrium thermodynamic system. Based on the fluid conservation equations under the Oberbeck-Boussinesq approximation,the work necessary to maintain atmospheric motion and heat fluxes at the boundaries were calculated. Using these calculations, the thermodynamic efficiency was formulated for stationary and chaotic dynamics. The numerical results show that, for both stationary and chaotic dynamics, the efficiency tends to increase as the atmospheric motion is driven out of thermodynamic equilibrium when the Rayleigh number increases. However, it is shown that the efficiency is upper bounded by the maximum efficiency, which is expressed in terms of the parameters characterizing the fluid and the convective system. The analysis of the entropy generation rate was also performed for elucidating the difference between the thermodynamic efficiency of conventional heat engines and the present atmospheric heat engine. It is also found that there exists an abrupt drop in efficiency at the critical Hopf bifurcation point, where the dynamics change from stationary to chaotic. These properties are similar to those found previously in Malkus-Lorenz waterwheel system.

nlin.CD↗

arXiv2023

ArSDM: Colonoscopy Images Synthesis with Adaptive Refinement Semantic Diffusion Models

Colonoscopy analysis, particularly automatic polyp segmentation and detection, is essential for assisting clinical diagnosis and treatment. However, as medical image annotation is labour- and resource-intensive, the scarcity of annotated data limits the effectiveness and generalization of existing methods. Although recent research has focused on data generation and augmentation to address this issue, the quality of the generated data remains a challenge, which limits the contribution to the performance of subsequent tasks. Inspired by the superiority of diffusion models in fitting data distributions and generating high-quality data, in this paper, we propose an Adaptive Refinement Semantic Diffusion Model (ArSDM) to generate colonoscopy images that benefit the downstream tasks. Specifically, ArSDM utilizes the ground-truth segmentation mask as a prior condition during training and adjusts the diffusion loss for each input according to the polyp/background size ratio. Furthermore, ArSDM incorporates a pre-trained segmentation model to refine the training process by reducing the difference between the ground-truth mask and the prediction mask. Extensive experiments on segmentation and detection tasks demonstrate the generated data by ArSDM could significantly boost the performance of baseline methods.

cs.CV↗

arXiv2023

ElecTra Code: Full-Band Electronic Transport Properties of Materials

This paper introduces ElecTra, an open-source code which solves the linearized Boltzmann transport equation in the relaxation time approximation for charge carriers in a full-band electronic structure of arbitrary complexity, including their energy, momentum, and band-index dependence. ElecTra stands for 'ELECtronic TRAnsport' and computes the electronic and thermoelectric transport coefficients electrical conductivity, Seebeck coefficient, electronic thermal conductivity, and mobility, for semiconductor materials, for both unipolar and bipolar (small bandgap) materials. The code uses computed full-bands and relevant scattering parameters as inputs and considers single crystal materials in 3D and 2D. The present version of the code (v1) considers: i) elastic scattering with acoustic phonons and inelastic scattering with non-polar optical phonons in the deformation potential approximation, ii) inelastic scattering with polar phonons, iii) scattering with ionized dopants, and iv) alloy scattering. The user is given the option of intra- and inter-valley scattering considerations. The simulation output also includes relevant relaxation times and mean-free-paths. The transport quantities are computed as a function of Fermi level position, doping density, and temperature. ElecTra can interface with any DFT code which saves the electronic structure in the '.bxsf' format. In this paper ElecTra is validated against ideal electronic transport situations of known analytical solutions, existing codes employing the constant relaxation time approximation, as well as experimentally well-assessed materials such as Si, Ge, SiGe, and GaAs.

cond-mat.mtrl-sci↗

arXiv2023

LATR: 3D Lane Detection from Monocular Images with Transformer

3D lane detection from monocular images is a fundamental yet challenging task in autonomous driving. Recent advances primarily rely on structural 3D surrogates (e.g., bird's eye view) built from front-view image features and camera parameters. However, the depth ambiguity in monocular images inevitably causes misalignment between the constructed surrogate feature map and the original image, posing a great challenge for accurate lane detection. To address the above issue, we present a novel LATR model, an end-to-end 3D lane detector that uses 3D-aware front-view features without transformed view representation. Specifically, LATR detects 3D lanes via cross-attention based on query and key-value pairs, constructed using our lane-aware query generator and dynamic 3D ground positional embedding. On the one hand, each query is generated based on 2D lane-aware features and adopts a hybrid embedding to enhance lane information. On the other hand, 3D space information is injected as positional embedding from an iteratively-updated 3D ground plane. LATR outperforms previous state-of-the-art methods on both synthetic Apollo, realistic OpenLane and ONCE-3DLanes by large margins (e.g., 11.4 gain in terms of F1 score on OpenLane). Code will be released at https://github.com/JMoonr/LATR .

cs.CV↗

arXiv2023

M$^2$-3DLaneNet: Exploring Multi-Modal 3D Lane Detection

Estimating accurate lane lines in 3D space remains challenging due to their sparse and slim nature. Previous works mainly focused on using images for 3D lane detection, leading to inherent projection error and loss of geometry information. To address these issues, we explore the potential of leveraging LiDAR for 3D lane detection, either as a standalone method or in combination with existing monocular approaches. In this paper, we propose M$^2$-3DLaneNet to integrate complementary information from multiple sensors. Specifically, M$^2$-3DLaneNet lifts 2D features into 3D space by incorporating geometry information from LiDAR data through depth completion. Subsequently, the lifted 2D features are further enhanced with LiDAR features through cross-modality BEV fusion. Extensive experiments on the large-scale OpenLane dataset demonstrate the effectiveness of M$^2$-3DLaneNet, regardless of the range (75m or 100m).

cs.CV↗

arXiv2023

YONA: You Only Need One Adjacent Reference-frame for Accurate and Fast Video Polyp Detection

Accurate polyp detection is essential for assisting clinical rectal cancer diagnoses. Colonoscopy videos contain richer information than still images, making them a valuable resource for deep learning methods. Great efforts have been made to conduct video polyp detection through multi-frame temporal/spatial aggregation. However, unlike common fixed-camera video, the camera-moving scene in colonoscopy videos can cause rapid video jitters, leading to unstable training for existing video detection models. Additionally, the concealed nature of some polyps and the complex background environment further hinder the performance of existing video detectors. In this paper, we propose the \textbf{YONA} (\textbf{Y}ou \textbf{O}nly \textbf{N}eed one \textbf{A}djacent Reference-frame) method, an efficient end-to-end training framework for video polyp detection. YONA fully exploits the information of one previous adjacent frame and conducts polyp detection on the current frame without multi-frame collaborations. Specifically, for the foreground, YONA adaptively aligns the current frame's channel activation patterns with its adjacent reference frames according to their foreground similarity. For the background, YONA conducts background dynamic alignment guided by inter-frame difference to eliminate the invalid features produced by drastic spatial jitters. Moreover, YONA applies cross-frame contrastive learning during training, leveraging the ground truth bounding box to improve the model's perception of polyp and background. Quantitative and qualitative experiments on three public challenging benchmarks demonstrate that our proposed YONA outperforms previous state-of-the-art competitors by a large margin in both accuracy and speed.

cs.CV↗

arXiv2023

Differential Convolutional Fuzzy Time Series Forecasting

Fuzzy time series forecasting (FTSF) is a typical forecasting method with wide application. Traditional FTSF is regarded as an expert system which leads to loss of the ability to recognize undefined features. The mentioned is the main reason for poor forecasting with FTSF. To solve the problem, the proposed model Differential Fuzzy Convolutional Neural Network (DFCNN) utilizes a convolution neural network to re-implement FTSF with learnable ability. DFCNN is capable of recognizing potential information and improving forecasting accuracy. Thanks to the learnable ability of the neural network, the length of fuzzy rules established in FTSF is expended to an arbitrary length that the expert is not able to handle by the expert system. At the same time, FTSF usually cannot achieve satisfactory performance of non-stationary time series due to the trend of non-stationary time series. The trend of non-stationary time series causes the fuzzy set established by FTSF to be invalid and causes the forecasting to fail. DFCNN utilizes the Difference algorithm to weaken the non-stationary of time series so that DFCNN can forecast the non-stationary time series with a low error that FTSF cannot forecast in satisfactory performance. After the mass of experiments, DFCNN has an excellent prediction effect, which is ahead of the existing FTSF and common time series forecasting algorithms. Finally, DFCNN provides further ideas for improving FTSF and holds continued research value.

cs.LG↗

arXiv2023

Electronic transport computation in thermoelectric materials: From ab initio scattering rates to nanostructures

Over the last two decades a plethora of new thermoelectric materials, their alloys, and their nanostructures were synthesized. The ZT figure of merit, which quantifies the thermoelectric efficiency of these materials increased from values of unity to values consistently beyond two across material families. At the same time, the ability to identify and optimize such materials, has stressed the need for advanced numerical tools for computing electronic transport in materials with arbitrary bandstructure complexity, multiple scattering mechanisms, and a large degree of nanostructuring. Many computational methods have been developed, the majority of which utilize the Boltzmann transport equation (BTE) formalism, spanning from fully ab initio to empirical treatment, with varying degree of computational expense and accuracy. In this paper we describe a suitable computational process that we have recently developed specifically for thermoelectric materials. The method consists of three independent software packages that we have developed and: 1) begins from ab initio calculation of the electron-phonon scattering rates, 2) to then be used within a Boltzmann transport simulator, and 3) calculated quantities from BTE are then passed on to a Monte Carlo simulator to examine electronic transport in highly nanostructured material configurations. The method we describe is computationally significantly advantageous compared to current fully ab initio and existing Monte Carlo methods, but with a similar degree of accuracy, thus making it truly enabling in understanding and assessing thermoelectric transport in complex band, nanostructured materials.

cond-mat.mtrl-sci↗

arXiv2023

WeakPolyp: You Only Look Bounding Box for Polyp Segmentation

Limited by expensive pixel-level labels, polyp segmentation models are plagued by data shortage and suffer from impaired generalization. In contrast, polyp bounding box annotations are much cheaper and more accessible. Thus, to reduce labeling cost, we propose to learn a weakly supervised polyp segmentation model (i.e., WeakPolyp) completely based on bounding box annotations. However, coarse bounding boxes contain too much noise. To avoid interference, we introduce the mask-to-box (M2B) transformation. By supervising the outer box mask of the prediction instead of the prediction itself, M2B greatly mitigates the mismatch between the coarse label and the precise prediction. But, M2B only provides sparse supervision, leading to non-unique predictions. Therefore, we further propose a scale consistency (SC) loss for dense supervision. By explicitly aligning predictions across the same image at different scales, the SC loss largely reduces the variation of predictions. Note that our WeakPolyp is a plug-and-play model, which can be easily ported to other appealing backbones. Besides, the proposed modules are only used during training, bringing no computation cost to inference. Extensive experiments demonstrate the effectiveness of our proposed WeakPolyp, which surprisingly achieves a comparable performance with a fully supervised model, requiring no mask annotations at all.

cs.CV↗

arXiv2023

Characterization of partial wetting by CMAS droplets using multiphase many-body dissipative particle dynamics and data-driven discovery based on PINNs

The molten sand, a mixture of calcia, magnesia, alumina, and silicate, known as CMAS, is characterized by its high viscosity, density, and surface tension. The unique properties of CMAS make it a challenging material to deal with in high-temperature applications, requiring innovative solutions and materials to prevent its buildup and damage to critical equipment. Here, we use multiphase many-body dissipative particle dynamics (mDPD) simulations to study the wetting dynamics of highly viscous molten CMAS droplets. The simulations are performed in three dimensions, with varying initial droplet sizes and equilibrium contact angles. We propose a coarse parametric ordinary differential equation (ODE) that captures the spreading radius behavior of the CMAS droplets. The ODE parameters are then identified based on the Physics-Informed Neural Network (PINN) framework. Subsequently, the closed form dependency of parameter values found by PINN on the initial radii and contact angles are given using symbolic regression. Finally, we employ Bayesian PINNs (B-PINNs) to assess and quantify the uncertainty associated with the discovered parameters. In brief, this study provides insight into spreading dynamics of CMAS droplets by fusing simple parametric ODE modeling and state-of-the-art machine learning techniques.

physics.flu-dyn↗

arXiv2023

SkeletonMAE: Graph-based Masked Autoencoder for Skeleton Sequence Pre-training

Skeleton sequence representation learning has shown great advantages for action recognition due to its promising ability to model human joints and topology. However, the current methods usually require sufficient labeled data for training computationally expensive models, which is labor-intensive and time-consuming. Moreover, these methods ignore how to utilize the fine-grained dependencies among different skeleton joints to pre-train an efficient skeleton sequence learning model that can generalize well across different datasets. In this paper, we propose an efficient skeleton sequence learning framework, named Skeleton Sequence Learning (SSL). To comprehensively capture the human pose and obtain discriminative skeleton sequence representation, we build an asymmetric graph-based encoder-decoder pre-training architecture named SkeletonMAE, which embeds skeleton joint sequence into Graph Convolutional Network (GCN) and reconstructs the masked skeleton joints and edges based on the prior human topology knowledge. Then, the pre-trained SkeletonMAE encoder is integrated with the Spatial-Temporal Representation Learning (STRL) module to build the SSL framework. Extensive experimental results show that our SSL generalizes well across different datasets and outperforms the state-of-the-art self-supervised skeleton-based action recognition methods on FineGym, Diving48, NTU 60 and NTU 120 datasets. Additionally, we obtain comparable performance to some fully supervised methods. The code is avaliable at https://github.com/HongYan1123/SkeletonMAE.

cs.CV↗

arXiv2023

Improved sd shell effective interactions from Daejeon16

We present new microscopic effective shell-model interactions in the valence sd shell, obtained from the modern Daejeon16 nucleon-nucleon potential using no-core shell-model (NCSM) wave functions of 18F at Nmax=6 (total oscillator quanta of excitation) model space and the Okubo-Lee-Suzuki transformation. First, we explore the convergence properties of our calculations and show that the excitation energies of states in 18F, characterized by the largest valence-like configurations, are reasonably converged and the lowest states are in sensible agreement with experiment. Then, we investigate the monopole properties of that interaction in comparison with the phenomenological universal sd-shell interaction, USDB, and with the previously derived interaction at Nmax=4. Theoretical binding energies and low-energy spectra of the O isotopes, as well as low-energy spectra of a selection of $sd$-shell nuclei, are presented. We conclude that the use of larger-space NCSM wave functions leads to a noticeable improvement in the quality of the derived effective interaction. We propose monopole modifications of the Daejeon16 centroids which further improve the agreement with experiment throughout the sd shell, as demonstrated by a compilation of spectra contained in Supplemental Material.

nucl-th↗

arXiv2023

Converging Many-Body Perturbation Theory for Ab Initio Nuclear-Structure: I. Brillouin-Wigner Perturbation Series for Closed-Shell Nuclei

Convergence aspects of nuclear many-body perturbation theory for ground states of closed-shell nuclei are explored using a Brillouin-Wigner formulation with a new vertex function enabling high-order calculations. A general formalism for Hamiltonian partitioning and a convergence criterion for the perturbation series are proposed. Analytical derivation shows that with optimal partitionings, the convergence criterion for ground states can always be satisfied. This feature attributes to the variational principle and does not depend on the choice of an internucleon interaction or a many-body basis. Numerical calculations of the ground state energies of 4He and 16O with Daejeon16 and a bare N3LO potential in both harmonic-oscillator and Hartree-Fock bases confirm this finding.

nucl-th↗

arXiv2023

Formation Control with Unknown Directions and General Coupling Coefficients

Generally, the normal displacement-based formation control has a sensing mode that requires the agent not only to have certain knowledge of its direction, but also to gather its local information characterized by nonnegative coupling coefficients. However, the direction may be unknown in the sensing processes, and the coupling coefficients may also involve negative ones due to some circumstances. This paper introduces these phenomena into a class of displacement-based formation control problem. Then, a geometric approach have been employed to overcome the difficulty of analysis on the introduced phenomena. The purpose of this approach is to construct some convex polytopes for containing the effects caused by the unknown direction, and to analyze the non-convexity by admitting the negative coupling coefficients in a certain range. Under the actions of these phenomena, the constructed polytopes are shown to be invariant in view of the contractive set method. It means that the convergence of formation shape can be guaranteed. Subsequently, an example is given to examine the applicability of derived result.

math.OC↗

arXiv2023

Toward Explainable and Fine-Grained 3D Grounding through Referring Textual Phrases

Recent progress in 3D scene understanding has explored visual grounding (3DVG) to localize a target object through a language description. However, existing methods only consider the dependency between the entire sentence and the target object, ignoring fine-grained relationships between contexts and non-target ones. In this paper, we extend 3DVG to a more fine-grained and interpretable task, called 3D Phrase Aware Grounding (3DPAG). The 3DPAG task aims to localize the target objects in a 3D scene by explicitly identifying all phrase-related objects and then conducting the reasoning according to contextual phrases. To tackle this problem, we manually labeled about 227K phrase-level annotations using a self-developed platform, from 88K sentences of widely used 3DVG datasets, i.e., Nr3D, Sr3D and ScanRefer. By tapping on our datasets, we can extend previous 3DVG methods to the fine-grained phrase-aware scenario. It is achieved through the proposed novel phrase-object alignment optimization and phrase-specific pre-training, boosting conventional 3DVG performance as well. Extensive results confirm significant improvements, i.e., previous state-of-the-art method achieves 3.9%, 3.5% and 4.6% overall accuracy gains on Nr3D, Sr3D and ScanRefer respectively.

cs.CV↗

arXiv2023

Deep Learning based Forecasting: a case study from the online fashion industry

Demand forecasting in the online fashion industry is particularly amendable to global, data-driven forecasting models because of the industry's set of particular challenges. These include the volume of data, the irregularity, the high amount of turn-over in the catalog and the fixed inventory assumption. While standard deep learning forecasting approaches cater for many of these, the fixed inventory assumption requires a special treatment via controlling the relationship between price and demand closely. In this case study, we describe the data and our modelling approach for this forecasting problem in detail and present empirical results that highlight the effectiveness of our approach.

cs.LG↗