arXiv subjects

Lei Wang

Publications and source records attributed to Lei Wang.

At least 811 records · Page 45Linked to original sources

arXiv2022

Primitive Graph Learning for Unified Vector Mapping

Large-scale vector mapping is important for transportation, city planning, and survey and census. We propose GraphMapper, a unified framework for end-to-end vector map extraction from satellite images. Our key idea is a novel unified representation of shapes of different topologies named "primitive graph", which is a set of shape primitives and their pairwise relationship matrix. Then, we convert vector shape prediction, regularization, and topology reconstruction into a unique primitive graph learning problem. Specifically, GraphMapper is a generic primitive graph learning network based on global shape context modelling through multi-head-attention. An embedding space sorting method is developed for accurate primitive relationship modelling. We empirically demonstrate the effectiveness of GraphMapper on two challenging mapping tasks, building footprint regularization and road network topology reconstruction. Our model outperforms state-of-the-art methods in both tasks on public benchmarks. All code will be publicly available.

cs.CV↗

arXiv2022

Water-stable MOFs and Hydrophobically Encapsulated MOFs for CO2 Capture from Ambient Air and Wet Flue Gas

The extra CO2 that has already been released into the atmosphere has to be removed in order to create a world that is carbon neutral. Technologies have been created to remove carbon dioxide from wet flue gas or even directly from ambient air, however these technologies are not widely deployed yet. New generations of creative CO2 capture sorbents have been produced as a consequence of recent improvements in material assembly and surface chemistry. We summarize recent progress on water-stable and encapsulated metal-organic frameworks (MOFs) for CO2 capture under a wide range of environmental and operating conditions. In particular, newly developed water-stable MOFs and hydrophobic coating technologies are discussed with insights into their materials discovery and the synergistic effects between different components of these hybrid sorbent systems. The future perspectives and directions of water-stable and encapsulated MOFs are also given for Direct Air Capture of CO2 and CO2 capture from wet flue gas.

cond-mat.mtrl-sci↗

arXiv2022

A Case Study of Chinese Sentiment Analysis on Social Media Reviews Based on LSTM

Network public opinion analysis is obtained by a combination of natural language processing (NLP) and public opinion supervision, and is crucial for monitoring public mood and trends. Therefore, network public opinion analysis can identify and solve potential and budding social problems. This study aims to realize an analysis of Chinese sentiment in social media reviews using a long short-term memory network (LSTM) model. The dataset was obtained from Sina Weibo using a web crawler and was cleaned with Pandas. First, Chinese comments regarding the legal sentencing in of Tangshan attack and Jiang Ge Case were segmented and vectorized. Then, a binary LSTM model was trained and tested. Finally, sentiment analysis results were obtained by analyzing the comments with the LSTM model. The accuracy of the proposed model has reached approximately 92%.

cs.CL↗

arXiv2022

Mitigating Popularity Bias in Recommendation with Unbalanced Interactions: A Gradient Perspective

Recommender systems learn from historical user-item interactions to identify preferred items for target users. These observed interactions are usually unbalanced following a long-tailed distribution. Such long-tailed data lead to popularity bias to recommend popular but not personalized items to users. We present a gradient perspective to understand two negative impacts of popularity bias in recommendation model optimization: (i) the gradient direction of popular item embeddings is closer to that of positive interactions, and (ii) the magnitude of positive gradient for popular items are much greater than that of unpopular items. To address these issues, we propose a simple yet efficient framework to mitigate popularity bias from a gradient perspective. Specifically, we first normalize each user embedding and record accumulated gradients of users and items via popularity bias measures in model training. To address the popularity bias issues, we develop a gradient-based embedding adjustment approach used in model testing. This strategy is generic, model-agnostic, and can be seamlessly integrated into most existing recommender systems. Our extensive experiments on two classic recommendation models and four real-world datasets demonstrate the effectiveness of our method over state-of-the-art debiasing baselines.

cs.IR↗

arXiv2022

Droplet impact dynamics on Janus-textured heated substrates

In this paper, the droplet impact dynamic behavior of droplet on Janus-textured heated substrates is numerically investigate by using the thermal Lattice Boltzmann method. The effect of several factors like the wettability, the Jakob number and the Weber number on the droplet impact dynamic behavior on Janus-textured heated surface are studied in detail. The simulation results show three types of boiling states, i.e., contact boiling, transition boiling and film boiling. The droplets in contact boiling state directionally migrate toward the denser region, while the droplets move toward the sparser region for film boiling state at low Weber number, which agrees with the direction of droplet motion in experiments [Zhang et al., Adv. Funct. Mater., 2019, 29, 1904535]. The self-propelled droplets on the Janus-textured heated substrates is owning to unbalanced Young's fore, vapor pressure difference and thermophoretic force. Moreover, The Janus-textured surface with low wettability has superior nucleation behavior resulting in their typically easy transition from efficient transition boiling state to inefficient film boiling state. In addition, the boiling states of the droplet change from the contact boiling state to the transition boiling state and then to the film boiling state as the Jakob number increases.

physics.flu-dyn↗

arXiv2022

Temporal-Viewpoint Transportation Plan for Skeletal Few-shot Action Recognition

We propose a Few-shot Learning pipeline for 3D skeleton-based action recognition by Joint tEmporal and cAmera viewpoiNt alIgnmEnt (JEANIE). To factor out misalignment between query and support sequences of 3D body joints, we propose an advanced variant of Dynamic Time Warping which jointly models each smooth path between the query and support frames to achieve simultaneously the best alignment in the temporal and simulated camera viewpoint spaces for end-to-end learning under the limited few-shot training data. Sequences are encoded with a temporal block encoder based on Simple Spectral Graph Convolution, a lightweight linear Graph Neural Network backbone. We also include a setting with a transformer. Finally, we propose a similarity-based loss which encourages the alignment of sequences of the same class while preventing the alignment of unrelated sequences. We show state-of-the-art results on NTU-60, NTU-120, Kinetics-skeleton and UWA3D Multiview Activity II.

cs.CV↗

arXiv2022

Time-rEversed diffusioN tEnsor Transformer: A new TENET of Few-Shot Object Detection

In this paper, we tackle the challenging problem of Few-shot Object Detection. Existing FSOD pipelines (i) use average-pooled representations that result in information loss; and/or (ii) discard position information that can help detect object instances. Consequently, such pipelines are sensitive to large intra-class appearance and geometric variations between support and query images. To address these drawbacks, we propose a Time-rEversed diffusioN tEnsor Transformer (TENET), which i) forms high-order tensor representations that capture multi-way feature occurrences that are highly discriminative, and ii) uses a transformer that dynamically extracts correlations between the query image and the entire support set, instead of a single average-pooled support embedding. We also propose a Transformer Relation Head (TRH), equipped with higher-order representations, which encodes correlations between query regions and the entire support set, while being sensitive to the positional variability of object instances. Our model achieves state-of-the-art results on PASCAL VOC, FSOD, and COCO.

cs.CV↗

arXiv2022

Uncertainty-DTW for Time Series and Sequences

Dynamic Time Warping (DTW) is used for matching pairs of sequences and celebrated in applications such as forecasting the evolution of time series, clustering time series or even matching sequence pairs in few-shot action recognition. The transportation plan of DTW contains a set of paths; each path matches frames between two sequences under a varying degree of time warping, to account for varying temporal intra-class dynamics of actions. However, as DTW is the smallest distance among all paths, it may be affected by the feature uncertainty which varies across time steps/frames. Thus, in this paper, we propose to model the so-called aleatoric uncertainty of a differentiable (soft) version of DTW. To this end, we model the heteroscedastic aleatoric uncertainty of each path by the product of likelihoods from Normal distributions, each capturing variance of pair of frames. (The path distance is the sum of base distances between features of pairs of frames of the path.) The Maximum Likelihood Estimation (MLE) applied to a path yields two terms: (i) a sum of Euclidean distances weighted by the variance inverse, and (ii) a sum of log-variance regularization terms. Thus, our uncertainty-DTW is the smallest weighted path distance among all paths, and the regularization term (penalty for the high uncertainty) is the aggregate of log-variances along the path. The distance and the regularization term can be used in various objectives. We showcase forecasting the evolution of time series, estimating the Fréchet mean of time series, and supervised/unsupervised few-shot action recognition of the articulated human 3D body joints.

cs.CV↗

arXiv2022

COST-EFF: Collaborative Optimization of Spatial and Temporal Efficiency with Slenderized Multi-exit Language Models

Transformer-based pre-trained language models (PLMs) mostly suffer from excessive overhead despite their advanced capacity. For resource-constrained devices, there is an urgent need for a spatially and temporally efficient model which retains the major capacity of PLMs. However, existing statically compressed models are unaware of the diverse complexities between input instances, potentially resulting in redundancy and inadequacy for simple and complex inputs. Also, miniature models with early exiting encounter challenges in the trade-off between making predictions and serving the deeper layers. Motivated by such considerations, we propose a collaborative optimization for PLMs that integrates static model compression and dynamic inference acceleration. Specifically, the PLM is slenderized in width while the depth remains intact, complementing layer-wise early exiting to speed up inference dynamically. To address the trade-off of early exiting, we propose a joint training approach that calibrates slenderization and preserves contributive structures to each exit instead of only the final layer. Experiments are conducted on GLUE benchmark and the results verify the Pareto optimality of our approach at high compression and acceleration rate with 1/8 parameters and 1/19 FLOPs of BERT.

cs.CL↗

arXiv2022

The Championship Simulator: Architectural Simulation for Education and Competition

Recent years have seen a dramatic increase in the microarchitectural complexity of processors. This increase in complexity presents a twofold challenge for the field of computer architecture. First, no individual architect can fully comprehend the complexity of the entire microarchitecture of the core. This leads to increasingly specialized architects, who treat parts of the core outside their particular expertise as black boxes. Second, with increasing complexity, the field becomes decreasingly accessible to new students of the field. When learning core microarchitecture, new students must first learn the big picture of how the system works in order to understand how the pieces all fit together. The tools used to study microarchitecture experience a similar struggle. As with the microarchitectures they simulate, an increase in complexity reduces accessibility to new users. In this work, we present ChampSim. ChampSim uses a modular design and configurable structure to achieve a low barrier to entry into the field of microarchitecural simulation. ChampSim has shown itself to be useful in multiple areas of research, competition, and education. In this way, we seek to promote access and inclusion despite the increasing complexity of the field of computer architecture.

cs.AR↗

arXiv2022

Recommendation with User Active Disclosing Willingness

Recommender system has been deployed in a large amount of real-world applications, profoundly influencing people's daily life and production.Traditional recommender models mostly collect as comprehensive as possible user behaviors for accurate preference estimation. However, considering the privacy, preference shaping and other issues, the users may not want to disclose all their behaviors for training the model. In this paper, we study a novel recommendation paradigm, where the users are allowed to indicate their "willingness" on disclosing different behaviors, and the models are optimized by trading-off the recommendation quality as well as the violation of the user "willingness". More specifically, we formulate the recommendation problem as a multiplayer game, where the action is a selection vector representing whether the items are involved into the model training. For efficiently solving this game, we design a tailored algorithm based on influence function to lower the time cost for recommendation quality exploration, and also extend it with multiple anchor selection vectors.We conduct extensive experiments to demonstrate the effectiveness of our model on balancing the recommendation quality and user disclosing willingness.

cs.IR↗

arXiv2022

Minimum Age of Information in Internet of Things with Opportunistic Channel Access

This paper investigates an Internet of Things (IoT) system in which multiple devices are observing some object's physical parameters and then offloading their observations back to the BS in time with opportunistic channel access. Specifically, each device accesses the common channel through contention with a certain probability firstly and then the winner evaluates the instant channel condition and decides to accept the right of channel access or not. We analyze this system through the perspective of Age of Information (AoI), which describes the freshness of observed information. The target is to minimize average AoI by optimizing the probability of device participation in contention and the transmission rate threshold. The problem is hard to solve since the AoI expression in fractional form is complex. We first decompose the original problem into two single-variable optimization sub-problems through Dinkelbach method and Block Coordinate Descent (BCD) method. And then we transform them to Monotonic optimization problems by proving the monotonicity of the objective functions, whose global optimal solution is able to be found through Polyblock algorithm. Numerical results verify the validity of our proposed method.

cs.IT↗

arXiv2022

Energy Efficient Train-Ground mmWave Mobile Relay System for High Speed Railways

The rapid development of high-speed railways (HSRs) puts forward high requirements on the corresponding communication system. Millimeter wave (mmWave) can be a promising solution due to its wide bandwidth, narrow beams, and rich spectrum resources. However, with the large number of antenna elements employed, energy-efficient solutions at mmWave frequencies are in great demand. Based on a mmWave HSR communication system with multiple mobile relays (MRs) on top of the train, a dynamic power-control scheme for train-ground communications is proposed. The scheme follows the regular movement characteristics of high-speed trains and considers three phases of train movement: the train enters the cell, all MRs are covered in the cell, and the train leaves the cell. The transmit power is further refined according to the number of MRs in the cell and the distance between the train and the remote radio head. By minimizing energy consumption under the constraints of the transmitted data and transmit power budget, the transmit power is allocated to multiple MRs through the multiplier punitive function-based algorithm. Comprehensive simulation results, where the velocity estimation error is taken into account, are provided to demonstrate the effectiveness of the proposed scheme over several baseline schemes.

cs.IT↗

arXiv2022

Cloud-based Automatic Speech Recognition Systems for Southeast Asian Languages

This paper provides an overall introduction of our Automatic Speech Recognition (ASR) systems for Southeast Asian languages. As not much existing work has been carried out on such regional languages, a few difficulties should be addressed before building the systems: limitation on speech and text resources, lack of linguistic knowledge, etc. This work takes Bahasa Indonesia and Thai as examples to illustrate the strategies of collecting various resources required for building ASR systems.

cs.CL↗

arXiv2022

Pronunciation Modeling of Foreign Words for Mandarin ASR by Considering the Effect of Language Transfer

One of the challenges in automatic speech recognition is foreign words recognition. It is observed that a speaker's pronunciation of a foreign word is influenced by his native language knowledge, and such phenomenon is known as the effect of language transfer. This paper focuses on examining the phonetic effect of language transfer in automatic speech recognition. A set of lexical rules is proposed to convert an English word into Mandarin phonetic representation. In this way, a Mandarin lexicon can be augmented by including English words. Hence, the Mandarin ASR system becomes capable to recognize English words without retraining or re-estimation of the acoustic model parameters. Using the lexicon that derived from the proposed rules, the ASR performance of Mandarin English mixed speech is improved without harming the accuracy of Mandarin only speech. The proposed lexical rules are generalized and they can be directly applied to unseen English words.

cs.CL↗

arXiv2022

On Embeddings and Inverse Embeddings of Input Design for Regularized System Identification

Input design is an important problem for system identification and has been well studied for the classical system identification, i.e., the maximum likelihood/prediction error method. For the emerging regularized system identification, the study on input design has just started, and it is often formulated as a non-convex optimization problem that minimizes a scalar measure of the Bayesian mean squared error matrix subject to certain constraints, and the state-of-art method is the so-called quadratic mapping and inverse embedding (QMIE) method, where a time domain inverse embedding (TDIE) is proposed to find the inverse of the quadratic mapping. In this paper, we report some new results on the embeddings/inverse embeddings of the QMIE method. Firstly, we present a general result on the frequency domain inverse embedding (FDIE) that is to find the inverse of the quadratic mapping described by the discrete-time Fourier transform. Then we show the relation between the TDIE and the FDIE from a graph signal processing perspective. Finally, motivated by this perspective, we further propose a graph induced embedding and its inverse, which include the previously introduced embeddings as special cases. This deepens the understanding of input design from a new viewpoint beyond the real domain and the frequency domain viewpoints.

eess.SY↗

arXiv2022

Deciphering the Maximal Transcendentality Principle via Bootstrap

We prove the principle of maximal transcendentality for a class of form factors, including the general two-loop minimal form factors, the two-loop three-point form factor of ${\rm tr}(F^2)$, and the two-loop four-point form factor of ${\rm tr}(F^3)$. Our proof is based on a recently developed bootstrap method using the representation of master integral expansions, together with some unitarity cuts that are universal in general gauge theories. The maximally transcendental parts of the two-loop four-gluon form factor of $\mathrm{tr}(F^3)$ are obtained for the first time in both planar $\mathcal{N}=4$ SYM and pure YM theories. This form factor can be understood as the Higgs-plus-four-gluon amplitudes involving a dimension-seven operator in the Higgs effective theory. In this case, we find that the maximally transcendental part of the $\mathcal{N}=4$ SYM result is different from that of pure YM, and the discrepancy is due to the gluino-loop contributions in $\mathcal{N}=4$ SYM. In contrast, the scalar-loop contributions have no maximally transcendental parts. Thus, the maximal transcendentality principle still holds for the form factor results in $\mathcal{N}=4$ SYM and QCD, after a proper identification of the fundamental quarks and adjoint gluinos as $n_f \rightarrow 4N_c$. This seems to be the first example of the maximally transcendental principle that involves fermion-loop contributions. As another intriguing observation, we find that the four-point form factor of the half-BPS $\mathrm{tr}(ϕ^3)$ operator is precisely a building block in the form factor of $\mathrm{tr}(F^3)$.

hep-th↗

arXiv2022

Towards Super-resolution via Iterative multi-exposure Coaddition

In this article, we provide an alternative up-sampling and PSF deconvolution method for the iterative multi-exposure coaddition. Different from the previous works, the new method has a ratio-correction term, which allows the iterations to converge more rapidly to an accurate representation of the underlying image than those with difference-correction terms. By employing this method, one can coadd the under-sampled multi-exposures to a super-resolution and obtain a higher peak signal-to-noise ratio. A set of simulations show that we can take many advantages of the new method, e.g. in the signal-to-noise ratio, the average deviation of all source fluxes, super-resolution, and source distortion ratio, etc., which are friendly to astronomical photometry and morphology, and benefits faint source detection and shear measurement of weak gravitational lensing. It provides an improvement in fidelity over the previous works tested in this paper.

astro-ph.IM↗