arXiv subjects

Xiaoyu Wang

Publications and source records attributed to Xiaoyu Wang.

At least 181 records · Page 10Linked to original sources

arXiv2023

Fcaformer: Forward Cross Attention in Hybrid Vision Transformer

Currently, one main research line in designing a more efficient vision transformer is reducing the computational cost of self attention modules by adopting sparse attention or using local attention windows. In contrast, we propose a different approach that aims to improve the performance of transformer-based architectures by densifying the attention pattern. Specifically, we proposed forward cross attention for hybrid vision transformer (FcaFormer), where tokens from previous blocks in the same stage are secondary used. To achieve this, the FcaFormer leverages two innovative components: learnable scale factors (LSFs) and a token merge and enhancement module (TME). The LSFs enable efficient processing of cross tokens, while the TME generates representative cross tokens. By integrating these components, the proposed FcaFormer enhances the interactions of tokens across blocks with potentially different semantics, and encourages more information flows to the lower levels. Based on the forward cross attention (Fca), we have designed a series of FcaFormer models that achieve the best trade-off between model size, computational cost, memory cost, and accuracy. For example, without the need for knowledge distillation to strengthen training, our FcaFormer achieves 83.1% top-1 accuracy on Imagenet with only 16.3 million parameters and about 3.6 billion MACs. This saves almost half of the parameters and a few computational costs while achieving 0.7% higher accuracy compared to distilled EfficientFormer.

cs.CV↗

arXiv2023

ParCNetV2: Oversized Kernel with Enhanced Attention

Transformers have shown great potential in various computer vision tasks. By borrowing design concepts from transformers, many studies revolutionized CNNs and showed remarkable results. This paper falls in this line of studies. Specifically, we propose a new convolutional neural network, ParCNetV2, that extends position-aware circular convolution (ParCNet) with oversized convolutions and bifurcate gate units to enhance attention. The oversized convolution employs a kernel with twice the input size to model long-range dependencies through a global receptive field. Simultaneously, it achieves implicit positional encoding by removing the shift-invariant property from convolution kernels, i.e., the effective kernels at different spatial locations are different when the kernel size is twice as large as the input size. The bifurcate gate unit implements an attention mechanism similar to self-attention in transformers. It is applied through element-wise multiplication of the two branches, one serves as feature transformation while the other serves as attention weights. Additionally, we introduce a uniform local-global convolution block to unify the design of the early and late stage convolution blocks. Extensive experiments demonstrate the superiority of our method over other convolutional neural networks and hybrid models that combine CNNs and transformers. Code will be released.

cs.CV↗

arXiv2023

The Sivers asymmetry in charged Kaon and $Λ$ hyperon produced SIDIS process at electron ion colliders

We investigate the single transverse-spin asymmetry with a $\sin (ϕ_h-ϕ_S)$ modulation in the charged Kaon produced and in $Λ$ hyperon produced SIDIS process within the theoretical framework of transverse momentum dependent (TMD) factorization at the next-to-leading-logarithmic order. The asymmetry is contributed by the convolution of Sivers function and the unpolarized fragmentation function $D_1$ for the produced hadron. The parametrization for the proton Qiu-Sterman function, which is closely related to the Sivers function, is adopted to numerically estimate the Sivers asymmetry at the kinematical region of Electron Ion Collider (EIC) and Electron Ion Collider in China (EicC). The TMD evolution of the TMD parton distribution functions are considered by employing the nonperturbative Sudakov form factor. It is found that the predicted Sivers asymmetries $A_{UT}^{\sin (ϕ_h-ϕ_S)}$ as functions of $x$, $z$ and $P_{hT}$ are sizable at the kinematical configurations of both EIC and EicC. The strange constituent of the produced charged Kaon and $Λ$ hyperon in the final state can be a promising probe of the sea quark Sivers function as well as the flavor dependence in the proton target. Therefore, it is important to utilize the future EIC facilities to constrain the sea quark distribution functions as well as the validity of the generalized universality of the Sivers function.

hep-ph↗

arXiv2023

Conservative Bayesian Model-Based Value Expansion for Offline Policy Optimization

Offline reinforcement learning (RL) addresses the problem of learning a performant policy from a fixed batch of data collected by following some behavior policy. Model-based approaches are particularly appealing in the offline setting since they can extract more learning signals from the logged dataset by learning a model of the environment. However, the performance of existing model-based approaches falls short of model-free counterparts, due to the compounding of estimation errors in the learned model. Driven by this observation, we argue that it is critical for a model-based method to understand when to trust the model and when to rely on model-free estimates, and how to act conservatively w.r.t. both. To this end, we derive an elegant and simple methodology called conservative Bayesian model-based value expansion for offline policy optimization (CBOP), that trades off model-free and model-based estimates during the policy evaluation step according to their epistemic uncertainties, and facilitates conservatism by taking a lower bound on the Bayesian posterior value estimate. On the standard D4RL continuous control tasks, we find that our method significantly outperforms previous model-based approaches: e.g., MOPO by $116.4$%, MOReL by $23.2$% and COMBO by $23.7$%. Further, CBOP achieves state-of-the-art performance on $11$ out of $18$ benchmark datasets while doing on par on the remaining datasets.

cs.LG↗

arXiv2023

A Lifted Bregman Formulation for the Inversion of Deep Neural Networks

We propose a novel framework for the regularised inversion of deep neural networks. The framework is based on the authors' recent work on training feed-forward neural networks without the differentiation of activation functions. The framework lifts the parameter space into a higher dimensional space by introducing auxiliary variables, and penalises these variables with tailored Bregman distances. We propose a family of variational regularisations based on these Bregman distances, present theoretical results and support their practical application with numerical examples. In particular, we present the first convergence result (to the best of our knowledge) for the regularised inversion of a single-layer perceptron that only assumes that the solution of the inverse problem is in the range of the regularisation operator, and that shows that the regularised inverse provably converges to the true inverse if measurement errors converge to zero.

math.NA↗

arXiv2023

A Multi-modal Deformable Land-air Robot for Complex Environments

Single locomotion robots often struggle to adapt in highly variable or uncertain environments, especially in emergencies. In this paper, a multi-modal deformable robot is introduced that can both fly and drive. Compatibility issues with multi-modal locomotive fusion for this hybrid land-air robot are solved using proposed design conceptions, including power settings, energy selection, and designs of deformable structure. The robot can also automatically transform between land and air modes during 3D planning and tracking. Meanwhile, we proposed a algorithms for evaluation the performance of land-air robots. A series of comparisons and experiments were conducted to demonstrate the robustness and reliability of the proposed structure in complex field environments.

cs.RO↗

arXiv2023

Few-shot Face Image Translation via GAN Prior Distillation

Face image translation has made notable progress in recent years. However, when training on limited data, the performance of existing approaches significantly declines. Although some studies have attempted to tackle this problem, they either failed to achieve the few-shot setting (less than 10) or can only get suboptimal results. In this paper, we propose GAN Prior Distillation (GPD) to enable effective few-shot face image translation. GPD contains two models: a teacher network with GAN Prior and a student network that fulfills end-to-end translation. Specifically, we adapt the teacher network trained on large-scale data in the source domain to the target domain with only a few samples, where it can learn the target domain's knowledge. Then, we can achieve few-shot augmentation by generating source domain and target domain images simultaneously with the same latent codes. We propose an anchor-based knowledge distillation module that can fully use the difference between the training and the augmented data to distill the knowledge of the teacher network into the student network. The trained student network achieves excellent generalization performance with the absorption of additional knowledge. Qualitative and quantitative experiments demonstrate that our method achieves superior results than state-of-the-art approaches in a few-shot setting.

cs.CV↗

arXiv2023

Being Bayesian in the 2020s: opportunities and challenges in the practice of modern applied Bayesian statistics

Building on a strong foundation of philosophy, theory, methods and computation over the past three decades, Bayesian approaches are now an integral part of the toolkit for most statisticians and data scientists. Whether they are dedicated Bayesians or opportunistic users, applied professionals can now reap many of the benefits afforded by the Bayesian paradigm. In this paper, we touch on six modern opportunities and challenges in applied Bayesian statistics: intelligent data collection, new data sources, federated analysis, inference for implicit models, model transfer and purposeful software products.

stat.AP↗

arXiv2022

Deep Active Learning for Computer Vision: Past and Future

As an important data selection schema, active learning emerges as the essential component when iterating an Artificial Intelligence (AI) model. It becomes even more critical given the dominance of deep neural network based models, which are composed of a large number of parameters and data hungry, in application. Despite its indispensable role for developing AI models, research on active learning is not as intensive as other research directions. In this paper, we present a review of active learning through deep active learning approaches from the following perspectives: 1) technical advancements in active learning, 2) applications of active learning in computer vision, 3) industrial systems leveraging or with potential to leverage active learning for data iteration, 4) current limitations and future research directions. We expect this paper to clarify the significance of active learning in a modern AI model manufacturing process and to bring additional research attention to active learning. By addressing data automation challenges and coping with automated machine learning systems, active learning will facilitate democratization of AI technologies by boosting model production at scale.

cs.LG↗

arXiv2022

Conventional high-temperature superconductivity in metallic, covalently bonded, binary-guest C-B clathrates

Inspired by the synthesis of XB3C3 (X= Sr, La) compounds in the bipartite sodalite clathrate structure, density functional theory (DFT) calculations are performed on members of this family containing up to two different metal atoms. A DFT-chemical pressure analysis on systems with X= Mg, Ca, Sr, Ba reveals that the size of the metal cation, which can be tuned to stabilize the B-C framework, is key for their ambient-pressure dynamic stability. High-throughput density functional theory calculations on 105 Pm-3 symmetry XYB6C6 binary-guest compounds (where X, Y are electropositive metal atoms) find 22 that are dynamically stable at 1 atmosphere, expanding the number of potentially synthesizable phases by 19 (18 metals and 1 insulator). The density of states at the Fermi level and superconducting critical temperature, Tc, can be tuned by changing the average oxidation state of the metal atoms, with Tc being highest for an average valence of +1.5. KPbB6C6, with an ambient-pressure Eliashberg Tc of 88 K, is predicted to possess the highest-Tc among the studied Pm-3n XB3C3 or Pm-3 XY B6C6 phases, and calculations suggest it may be synthesized using high-pressure high-temperature techniques then quenched to ambient conditions.

cond-mat.supr-con↗

arXiv2022

A Critical Review of Traffic Signal Control and A Novel Unified View of Reinforcement Learning and Model Predictive Control Approaches for Adaptive Traffic Signal Control

Recent years have witnessed substantial growth in adaptive traffic signal control (ATSC) methodologies that improve transportation network efficiency, especially in branches leveraging artificial intelligence based optimization and control algorithms such as reinforcement learning as well as conventional model predictive control. However, lack of cross-domain analysis and comparison of the effectiveness of applied methods in ATSC research limits our understanding of existing challenges and research directions. This chapter proposes a novel unified view of modern ATSCs to identify common ground as well as differences and shortcomings of existing methodologies with the ultimate goal to facilitate cross-fertilization and advance the state-of-the-art. The unified view applies the mathematical language of the Markov decision process, describes the process of controller design from both the world (problem) and solution modeling perspectives. The unified view also analyses systematic issues commonly ignored in existing studies and suggests future potential directions to resolve these issues.

eess.SY↗

arXiv2022

ALBench: A Framework for Evaluating Active Learning in Object Detection

Active learning is an important technology for automated machine learning systems. In contrast to Neural Architecture Search (NAS) which aims at automating neural network architecture design, active learning aims at automating training data selection. It is especially critical for training a long-tailed task, in which positive samples are sparsely distributed. Active learning alleviates the expensive data annotation issue through incrementally training models powered with efficient data selection. Instead of annotating all unlabeled samples, it iteratively selects and annotates the most valuable samples. Active learning has been popular in image classification, but has not been fully explored in object detection. Most of current approaches on object detection are evaluated with different settings, making it difficult to fairly compare their performance. To facilitate the research in this field, this paper contributes an active learning benchmark framework named as ALBench for evaluating active learning in object detection. Developed on an automatic deep model training system, this ALBench framework is easy-to-use, compatible with different active learning algorithms, and ensures the same training and testing protocols. We hope this automated benchmark system help researchers to easily reproduce literature's performance and have objective comparisons with prior arts. The code will be release through Github.

cs.CV↗

arXiv2022

DYNAFED: Tackling Client Data Heterogeneity with Global Dynamics

The Federated Learning (FL) paradigm is known to face challenges under heterogeneous client data. Local training on non-iid distributed data results in deflected local optimum, which causes the client models drift further away from each other and degrades the aggregated global model's performance. A natural solution is to gather all client data onto the server, such that the server has a global view of the entire data distribution. Unfortunately, this reduces to regular training, which compromises clients' privacy and conflicts with the purpose of FL. In this paper, we put forth an idea to collect and leverage global knowledge on the server without hindering data privacy. We unearth such knowledge from the dynamics of the global model's trajectory. Specifically, we first reserve a short trajectory of global model snapshots on the server. Then, we synthesize a small pseudo dataset such that the model trained on it mimics the dynamics of the reserved global model trajectory. Afterward, the synthesized data is used to help aggregate the deflected clients into the global model. We name our method Dynafed, which enjoys the following advantages: 1) we do not rely on any external on-server dataset, which requires no additional cost for data collection; 2) the pseudo data can be synthesized in early communication rounds, which enables Dynafed to take effect early for boosting the convergence and stabilizing training; 3) the pseudo data only needs to be synthesized once and can be directly utilized on the server to help aggregation in subsequent rounds. Experiments across extensive benchmarks are conducted to showcase the effectiveness of Dynafed. We also provide insights and understanding of the underlying mechanism of our method.

cs.LG↗

arXiv2022

Dynamic Causal Effects Evaluation in A/B Testing with a Reinforcement Learning Framework

A/B testing, or online experiment is a standard business strategy to compare a new product with an old one in pharmaceutical, technological, and traditional industries. Major challenges arise in online experiments of two-sided marketplace platforms (e.g., Uber) where there is only one unit that receives a sequence of treatments over time. In those experiments, the treatment at a given time impacts current outcome as well as future outcomes. The aim of this paper is to introduce a reinforcement learning framework for carrying A/B testing in these experiments, while characterizing the long-term treatment effects. Our proposed testing procedure allows for sequential monitoring and online updating. It is generally applicable to a variety of treatment designs in different industries. In addition, we systematically investigate the theoretical properties (e.g., size and power) of our testing procedure. Finally, we apply our framework to both simulated data and a real-world data example obtained from a technological company to illustrate its advantage over the current practice. A Python implementation of our test is available at https://github.com/callmespring/CausalRL.

cs.LG↗

arXiv2022

Commonsense Knowledge Salience Evaluation with a Benchmark Dataset in E-commerce

In e-commerce, the salience of commonsense knowledge (CSK) is beneficial for widespread applications such as product search and recommendation. For example, when users search for ``running'' in e-commerce, they would like to find products highly related to running, such as ``running shoes'' rather than ``shoes''. Nevertheless, many existing CSK collections rank statements solely by confidence scores, and there is no information about which ones are salient from a human perspective. In this work, we define the task of supervised salience evaluation, where given a CSK triple, the model is required to learn whether the triple is salient or not. In addition to formulating the new task, we also release a new Benchmark dataset of Salience Evaluation in E-commerce (BSEE) and hope to promote related research on commonsense knowledge salience evaluation. We conduct experiments in the dataset with several representative baseline models. The experimental results show that salience evaluation is a challenging task where models perform poorly on our evaluation set. We further propose a simple but effective approach, PMI-tuning, which shows promise for solving this novel problem. Code is available in \url{https://github.com/OpenBGBenchmark/OpenBG-CSK.

cs.CL↗

arXiv2022

Predictive Edge Caching through Deep Mining of Sequential Patterns in User Content Retrievals

Edge caching plays an increasingly important role in boosting user content retrieval performance while reducing redundant network traffic. The effectiveness of caching ultimately hinges on the accuracy of predicting content popularity in the near future. However, at the network edge, content popularity can be extremely dynamic due to diverse user content retrieval behaviors and the low-degree of user multiplexing. It's challenging for the traditional reactive caching systems to keep up with the dynamic content popularity patterns. In this paper, we propose a novel Predictive Edge Caching (PEC) system that predicts the future content popularity using fine-grained learning models that mine sequential patterns in user content retrieval behaviors, and opportunistically prefetches contents predicted to be popular in the near future using idle network bandwidth. Through extensive experiments driven by real content retrieval traces, we demonstrate that PEC can adapt to highly dynamic content popularity, and significantly improve cache hit ratio and reduce user content retrieval latency over the state-of-art caching policies. More broadly, our study demonstrates that edge caching performance can be boosted by deep mining of user content retrieval behaviors.

cs.NI↗

arXiv2022

Unusual magnetotransport in twisted bilayer graphene from strain-induced open Fermi surfaces

Anisotropic hopping in a toy Hofstadter model was recently invoked to explain a rich and surprising Landau spectrum measured in twisted bilayer graphene away from the magic angle. Suspecting that such anisotropy could arise from unintended uniaxial strain, we extend the Bistritzer-MacDonald model to include uniaxial heterostrain. We find that such strain strongly influences band structure, shifting the three otherwise-degenerate van Hove points to different energies. Coupled to a Boltzmann magnetotransport calculation, this reproduces previously-unexplained non-saturating $B^2$ magnetoresistance over broad ranges of density near filling $ν=\pm 2$, and predicts subtler features that had not been noticed in the experimental data. In contrast to these distinctive signatures in longitudinal resistivity, the Hall coefficient is barely influenced by strain, to the extent that it still shows a single sign change on each side of the charge neutrality point -- surprisingly, this sign change no longer occurs at a van Hove point. The theory also predicts a marked rotation of the electrical transport principal axes as a function of filling even for fixed strain and for rigid bands. More careful examination of interaction-induced nematic order versus strain effects in twisted bilayer graphene could thus be in order.

cond-mat.mes-hall↗

arXiv2022

Spin-selective tunneling from nanowires of the candidate topological Kondo insulator SmB6

Incorporating relativistic physics into quantum tunneling can lead to exotic behavior such as perfect transmission via Klein tunneling. Here, we probe the tunneling properties of spin-momentum locked relativistic fermions by designing and implementing a tunneling geometry that utilizes nanowires of the topological Kondo insulator candidate, SmB6. The nanowires are attached to the end of scanning tunneling microscope tips, and used to image the bicollinear stripe spin-order in the antiferromagnet Fe1.03Te with a Neel temperature of ~50 K. The antiferromagnetic stripes become invisible above 10 K concomitant with the suppression of the topological surface states. We further demonstrate that the direction of spin-polarization is tied to the tunneling direction. Our technique establishes SmB6 nanowires as ideal conduits for spin-polarized currents.

cond-mat.str-el↗