arXiv subjects

Li Zhang

Publications and source records attributed to Li Zhang.

At least 649 records · Page 36Linked to original sources

arXiv2022

ONCE-3DLanes: Building Monocular 3D Lane Detection

We present ONCE-3DLanes, a real-world autonomous driving dataset with lane layout annotation in 3D space. Conventional 2D lane detection from a monocular image yields poor performance of following planning and control tasks in autonomous driving due to the case of uneven road. Predicting the 3D lane layout is thus necessary and enables effective and safe driving. However, existing 3D lane detection datasets are either unpublished or synthesized from a simulated environment, severely hampering the development of this field. In this paper, we take steps towards addressing these issues. By exploiting the explicit relationship between point clouds and image pixels, a dataset annotation pipeline is designed to automatically generate high-quality 3D lane locations from 2D lane annotations in 211K road scenes. In addition, we present an extrinsic-free, anchor-free method, called SALAD, regressing the 3D coordinates of lanes in image view without converting the feature map into the bird's-eye view (BEV). To facilitate future research on 3D lane detection, we benchmark the dataset and provide a novel evaluation metric, performing extensive experiments of both existing approaches and our proposed method. The aim of our work is to revive the interest of 3D lane detection in a real-world scenario. We believe our work can lead to the expected and unexpected innovations in both academia and industry.

cs.CV↗

arXiv2022

Is "My Favorite New Movie" My Favorite Movie? Probing the Understanding of Recursive Noun Phrases

Recursive noun phrases (NPs) have interesting semantic properties. For example, "my favorite new movie" is not necessarily my favorite movie, whereas "my new favorite movie" is. This is common sense to humans, yet it is unknown whether language models have such knowledge. We introduce the Recursive Noun Phrase Challenge (RNPC), a dataset of three textual inference tasks involving textual entailment and event plausibility comparison, precisely targeting the understanding of recursive NPs. When evaluated on RNPC, state-of-the-art Transformer models only perform around chance. Still, we show that such knowledge is learnable with appropriate data. We further probe the models for relevant linguistic features that can be learned from our tasks, including modifier semantic category and modifier scope. Finally, models trained on RNPC achieve strong zero-shot performance on an extrinsic Harm Detection evaluation task, showing the usefulness of the understanding of recursive NPs in downstream applications.

cs.CL↗

arXiv2022

PR-PL: A Novel Transfer Learning Framework with Prototypical Representation based Pairwise Learning for EEG-Based Emotion Recognition

Affective brain-computer interfaces based on electroencephalography (EEG) is an important branch in the field of affective computing. However, individual differences and noisy labels seriously limit the effectiveness and generalizability of EEG-based emotion recognition models. In this paper, we propose a novel transfer learning framework with Prototypical Representation based Pairwise Learning (PR-PL) to learn discriminative and generalized prototypical representations for emotion revealing across individuals and formulate emotion recognition as pairwise learning for alleviating the reliance on precise label information. Extensive experiments are conducted on two benchmark databases under four cross-validation evaluation protocols (cross-subject cross-session, cross-subject within-session, within-subject cross-session, and within-subject within-session). The experimental results demonstrate the superiority of the proposed PR-PL against the state-of-the-arts under all four evaluation protocols, which shows the effectiveness and generalizability of PR-PL in dealing with the ambiguity of EEG responses in affective studies. The source code is available at https://github.com/KAZABANA/PR-PL.

cs.HC↗

arXiv2022

Elastic and inelastic $J/ψ$ photoproduction in $p$-$p$ collisions at LHC energies: the feature of Weizsäcker-Williams approximation

The $J/ψ$ production originating from elastic and inelastic photoproduction processes in p-p collisions at LHC energies is investigated, where the fragmentation processes are involved. An exact treatment is performed, which adopts the Martin-Ryskin method to weight the contribution from different channels, and can return to Weizsäcker-Williams approximation (WWA) when $Q^{2}\rightarrow0$. The relevant kinematical relations are also achieved. We present a comprehensive analysis for the feature of WWA by comparing with the exact treatment. The results are expressed in $Q^{2}$, $y$, $z$, $p_{T}$, and $y_{r}$ (rapidity) distributions, and the total cross sections are also estimated. The numerical results indicate that, the incoherent-photon emission can provide the meaningful contribution to elastic photoproduction, and starts to play a very important role in the inelastic processes. The photoproduction and fragmentation processes can improve the contribution of $J/ψ$ production in $p$-$p$ collisions at LHC energies. Moreover, the WWA is only effective in very restricted domains, and the exact treatment is needed to deal accurately with the $J/ψ$ photoproduction, which can naturally avoid double counting and WWA errors.

hep-ph↗

arXiv2022

SOFT: Softmax-free Transformer with Linear Complexity

Vision transformers (ViTs) have pushed the state-of-the-art for various visual recognition tasks by patch-wise image tokenization followed by self-attention. However, the employment of self-attention modules results in a quadratic complexity in both computation and memory usage. Various attempts on approximating the self-attention computation with linear complexity have been made in Natural Language Processing. However, an in-depth analysis in this work shows that they are either theoretically flawed or empirically ineffective for visual recognition. We further identify that their limitations are rooted in keeping the softmax self-attention during approximations. Specifically, conventional self-attention is computed by normalizing the scaled dot-product between token feature vectors. Keeping this softmax operation challenges any subsequent linearization efforts. Based on this insight, for the first time, a softmax-free transformer or SOFT is proposed. To remove softmax in self-attention, Gaussian kernel function is used to replace the dot-product similarity without further normalization. This enables a full self-attention matrix to be approximated via a low-rank matrix decomposition. The robustness of the approximation is achieved by calculating its Moore-Penrose inverse using a Newton-Raphson method. Extensive experiments on ImageNet show that our SOFT significantly improves the computational efficiency of existing ViT variants. Crucially, with a linear complexity, much longer token sequences are permitted in SOFT, resulting in superior trade-off between accuracy and complexity.

cs.CV↗

arXiv2022

Origami-controlled strain engineering of tunable flat bands and correlated states in folded graphene

Flat electronic bands with tunable structures offer opportunities for the exploitation and manipulation of exotic interacting quantum states. Here, we present a controllable route to construct easily tunable flat bands in folded graphene, by nano origami-controlled strain engineering, and discover correlated states in this system. Via tearing and folding graphene monolayer at arbitrary step edges with scanning tunneling microscope manipulation, we create strain-induced pseudo-magnetic fields as well as resulting flat electronic bands in the curved edges of folded graphene. We show that the intensity of the pseudo-magnetic field can be readily tuned by changing the width of the folding edge due to the edge-width-dependent lattice deformation, leading to the well adjustability of the geometry of flat bands in folded graphene. Furthermore, by creating expected dispersionless flat bands using this technique, the correlation-induced splits of flat bands are successfully observed in the density of states when these bands are partially filled. Our experiment provides a feasible and effective pathway to engineer the system with tunable flat band structures, and establishes a new platform that can be used to realize devisable strain and interaction induced quantum phases.

cond-mat.mes-hall↗

arXiv2022

Characterizing the $γ$-Ray Variability of Active Galactic Nuclei with Stochastic Process Method

The $γ$-ray astronomy in time domain has been by now progressed further as the variabilities of Active Galactic Nuclei (AGNs) on different timescales have been reported a lot. We study the $γ$-ray variabilities of 23 jetted AGNs through applying a stochastic process method to the ~12.7 yr long-term light curve (LC) obtained by Fermi-Large Area Telescope (Fermi-LAT). In this method, the stochastically driven damped simple harmonic oscillator (SHO) and the damped random walk (DRW) models are used to model the long-term LCs. Our results show that the long-term variabilities of 23 AGNs can be characterized well by both SHO and DRW models. However, the SHO model is restricted in the over-damped mode and the parameters are poorly constrained. The SHO power spectral densities (PSDs) are same as the typical DRW PSD. In the plot of the rest-frame timescale that corresponds to the broken frequency in the PSD versus black hole mass, the intrinsic $γ$-ray characteristic timescales of 23 AGNs occupy almost the same space with the optical variability timescales obtained from the accretion disk emission. This suggests a connection between the jet and the accretion disk. Same as the optical variability of AGN accretion disk, the $γ$-ray timescale is also consistent with the thermal timescale caused by the thermal instability in the standard accretion disk of AGN.

astro-ph.HE↗

arXiv2022

In Defense of Subspace Tracker: Orthogonal Embedding for Visual Tracking

The paper focuses on a classical tracking model, subspace learning, grounded on the fact that the targets in successive frames are considered to reside in a low-dimensional subspace or manifold due to the similarity in their appearances. In recent years, a number of subspace trackers have been proposed and obtained impressive results. Inspired by the most recent results that the tracking performance is boosted by the subspace with discrimination capability learned over the recently localized targets and their immediately surrounding background, this work aims at solving such a problem: how to learn a robust low-dimensional subspace to accurately and discriminatively represent these target and background samples. To this end, a discriminative approach, which reliably separates the target from its surrounding background, is injected into the subspace learning by means of joint learning, achieving a dimension-adaptive subspace with superior discrimination capability. The proposed approach is extensively evaluated and compared with the state-of-the-art trackers on four popular tracking benchmarks. The experimental results demonstrate that the proposed tracker performs competitively against its counterparts. In particular, it achieves more than 9% performance increase compared with the state-of-the-art subspace trackers.

cs.CV↗

arXiv2022

UIGR: Unified Interactive Garment Retrieval

Interactive garment retrieval (IGR) aims to retrieve a target garment image based on a reference garment image along with user feedback on what to change on the reference garment. Two IGR tasks have been studied extensively: text-guided garment retrieval (TGR) and visually compatible garment retrieval (VCR). The user feedback for the former indicates what semantic attributes to change with the garment category preserved, while the category is the only thing to be changed explicitly for the latter, with an implicit requirement on style preservation. Despite the similarity between these two tasks and the practical need for an efficient system tackling both, they have never been unified and modeled jointly. In this paper, we propose a Unified Interactive Garment Retrieval (UIGR) framework to unify TGR and VCR. To this end, we first contribute a large-scale benchmark suited for both problems. We further propose a strong baseline architecture to integrate TGR and VCR in one model. Extensive experiments suggest that unifying two tasks in one framework is not only more efficient by requiring a single model only, it also leads to better performance. Code and datasets are available at https://github.com/BrandonHanx/CompFashion.

cs.CV↗

arXiv2022

Trajectories and Radiation of Charged Particles in the Pulsar Magnetosphere

Trajectories and radiation of the accelerating electrons are studied in the pulsar magnetosphere approximated as the electromagnetic field of the Deutsch's solutions. Because the electrons are accelerated rapidly to ultra-relativistic velocity near the neutron star surface, the electron velocity vector (and then its trajectory) is derived from the balance between Lorentz force and radiation reaction force, which makes the pitch angle between electron trajectories and magnetic field lines nonzero in most part of the magnetosphere. In such a case, the spectral energy distributions (SEDs) of synchro-curvature radiation for the accelerating electrons with a mono-energetic form are calculated. Our results indicate that: (i) the pitch angle is the function of electron position ($r, θ, ϕ$) in the open field line regions, and increases with increasing $r$ and $θ$ as well as increasing the inclination angle; (ii) the radius of curvature becomes large along the particle trajectory, and (iii) the SED appears a double peak structure depending on the emission position, where the synchrotron radiation plays an important role in X-ray band and curvature radiation mainly works in GeV band, which is only determined by parameters $α$ and $ζ$

hep-ph↗

arXiv2022

Graph Node-Feature Convolution for Representation Learning

Graph convolutional network (GCN) is an emerging neural network approach. It learns new representation of a node by aggregating feature vectors of all neighbors in the aggregation process without considering whether the neighbors or features are useful or not. Recent methods have improved solutions by sampling a fixed size set of neighbors, or assigning different weights to different neighbors in the aggregation process, but features within a feature vector are still treated equally in the aggregation process. In this paper, we introduce a new convolution operation on regular size feature maps constructed from features of a fixed node bandwidth via sampling to get the first-level node representation, which is then passed to a standard GCN to learn the second-level node representation. Experiments show that our method outperforms competing methods in semi-supervised node classification tasks. Furthermore, our method opens new doors for exploring new GCN architectures, particularly deeper GCN models.

cs.LG↗

arXiv2022

ImpDet: Exploring Implicit Fields for 3D Object Detection

Conventional 3D object detection approaches concentrate on bounding boxes representation learning with several parameters, i.e., localization, dimension, and orientation. Despite its popularity and universality, such a straightforward paradigm is sensitive to slight numerical deviations, especially in localization. By exploiting the property that point clouds are naturally captured on the surface of objects along with accurate location and intensity information, we introduce a new perspective that views bounding box regression as an implicit function. This leads to our proposed framework, termed Implicit Detection or ImpDet, which leverages implicit field learning for 3D object detection. Our ImpDet assigns specific values to points in different local 3D spaces, thereby high-quality boundaries can be generated by classifying points inside or outside the boundary. To solve the problem of sparsity on the object surface, we further present a simple yet efficient virtual sampling strategy to not only fill the empty region, but also learn rich semantic features to help refine the boundaries. Extensive experimental results on KITTI and Waymo benchmarks demonstrate the effectiveness and robustness of unifying implicit fields into object detection.

cs.CV↗

arXiv2022

ALX: Large Scale Matrix Factorization on TPUs

We present ALX, an open-source library for distributed matrix factorization using Alternating Least Squares, written in JAX. Our design allows for efficient use of the TPU architecture and scales well to matrix factorization problems of O(B) rows/columns by scaling the number of available TPU cores. In order to spur future research on large scale matrix factorization methods and to illustrate the scalability properties of our own implementation, we also built a real world web link prediction dataset called WebGraph. This dataset can be easily modeled as a matrix factorization problem. We created several variants of this dataset based on locality and sparsity properties of sub-graphs. The largest variant of WebGraph has around 365M nodes and training a single epoch finishes in about 20 minutes with 256 TPU cores. We include speed and performance numbers of ALX on all variants of WebGraph. Both the framework code and the dataset is open-sourced.

cs.LG↗

arXiv2022

BigDetection: A Large-scale Benchmark for Improved Object Detector Pre-training

Multiple datasets and open challenges for object detection have been introduced in recent years. To build more general and powerful object detection systems, in this paper, we construct a new large-scale benchmark termed BigDetection. Our goal is to simply leverage the training data from existing datasets (LVIS, OpenImages and Object365) with carefully designed principles, and curate a larger dataset for improved detector pre-training. Specifically, we generate a new taxonomy which unifies the heterogeneous label spaces from different sources. Our BigDetection dataset has 600 object categories and contains over 3.4M training images with 36M bounding boxes. It is much larger in multiple dimensions than previous benchmarks, which offers both opportunities and challenges. Extensive experiments demonstrate its validity as a new benchmark for evaluating different object detection methods, and its effectiveness as a pre-training dataset.

cs.CV↗

arXiv2022

Discovery of a maximally charged Weyl point

The hypothetical Weyl particles in high-energy physics have been discovered in three-dimensional crystals as collective quasiparticle excitations near two-fold degenerate Weyl points. Such momentum-space Weyl particles carry quantized chiral charges, which can be measured by counting the number of Fermi arcs emanating from the corresponding Weyl points. It is known that merging unit-charged Weyl particles can create new ones with more charges. However, only very recently has it been realised that there is an upper limit - the maximal charge number that a two-fold Weyl point can host is four - achievable only in crystals without spin-orbit coupling. Here, we report the experimental realisation of such a maximally charged Weyl point in a three-dimensional photonic crystal. The four charges support quadruple-helicoid Fermi arcs, forming an unprecedented topology of two non-contractible loops in the surface Brillouin zone. The helicoid Fermi arcs also exhibit the long-pursued type-II van Hove singularities that can reside at arbitrary momenta. This discovery reveals a type of maximally charged Weyl particles beyond conventional topological particles in crystals.

cond-mat.mes-hall↗

arXiv2022

Show Me More Details: Discovering Hierarchies of Procedures from Semi-structured Web Data

Procedures are inherently hierarchical. To "make videos", one may need to "purchase a camera", which in turn may require one to "set a budget". While such hierarchical knowledge is critical for reasoning about complex procedures, most existing work has treated procedures as shallow structures without modeling the parent-child relation. In this work, we attempt to construct an open-domain hierarchical knowledge-base (KB) of procedures based on wikiHow, a website containing more than 110k instructional articles, each documenting the steps to carry out a complex procedure. To this end, we develop a simple and efficient method that links steps (e.g., "purchase a camera") in an article to other articles with similar goals (e.g., "how to choose a camera"), recursively constructing the KB. Our method significantly outperforms several strong baselines according to automatic evaluation, human judgment, and application to downstream tasks such as instructional video retrieval. A demo with partial data can be found at https://wikihow-hierarchy.github.io. The code and the data are at https://github.com/shuyanzhou/wikihow_hierarchy.

cs.CL↗

arXiv2022

Multi-Scale Context-Guided Lumbar Spine Disease Identification with Coarse-to-fine Localization and Classification

Accurate and efficient lumbar spine disease identification is crucial for clinical diagnosis. However, existing deep learning models with millions of parameters often fail to learn with only hundreds or dozens of medical images. These models also ignore the contextual relationship between adjacent objects, such as between vertebras and intervertebral discs. This work introduces a multi-scale context-guided network with coarse-to-fine localization and classification, named CCF-Net, for lumbar spine disease identification. Specifically, in learning, we divide the localization objective into two parallel tasks, coarse and fine, which are more straightforward and effectively reduce the number of parameters and computational cost. The experimental results show that the coarse-to-fine design presents the potential to achieve high performance with fewer parameters and data requirements. Moreover, the multi-scale context-guided module can significantly improve the performance by 6.45% and 5.51% with ResNet18 and ResNet50, respectively. Our code is available at https://github.com/czifan/CCFNet.pytorch.

cs.CV↗

arXiv2022

A general framework for adaptive two-index fusion attribute weighted naive Bayes

Naive Bayes(NB) is one of the essential algorithms in data mining. However, it is rarely used in reality because of the attribute independent assumption. Researchers have proposed many improved NB methods to alleviate this assumption. Among these methods, due to high efficiency and easy implementation, the filter attribute weighted NB methods receive great attentions. However, there still exists several challenges, such as the poor representation ability for single index and the fusion problem of two indexes. To overcome above challenges, we propose a general framework for Adaptive Two-index Fusion attribute weighted NB(ATFNB). Two types of data description category are used to represent the correlation between classes and attributes, intercorrelation between attributes and attributes, respectively. ATFNB can select any one index from each category. Then, we introduce a switching factor \{beta} to fuse two indexes, which can adaptively adjust the optimal ratio of the two index on various datasets. And a quick algorithm is proposed to infer the optimal interval of switching factor \{beta}. Finally, the weight of each attribute is calculated using the optimal value \{beta} and is integrated into NB classifier to improve the accuracy. The experimental results on 50 benchmark datasets and a Flavia dataset show that ATFNB outperforms the basic NB and state-of-the-art filter weighted NB models. In addition, the ATFNB framework can improve the existing two-index NB model by introducing the adaptive switching factor \{beta}. Auxiliary experimental results demonstrate the improved model significantly increases the accuracy compared to the original model without the adaptive switching factor \{beta}.

cs.LG↗