Search arXiv⌕ Search

arXiv subjects

Lei Wang

Publications and source records attributed to Lei Wang.

At least 829 records · Page 46Linked to original sources
arXiv2022

LibFewShot: A Comprehensive Library for Few-shot Learning

Few-shot learning, especially few-shot image classification, has received increasing attention and witnessed significant advances in recent years. Some recent studies implicitly show that many generic techniques or ``tricks'', such as data augmentation, pre-training, knowledge distillation, and self-supervision, may greatly boost the performance of a few-shot learning method. Moreover, different works may employ different software platforms, backbone architectures and input image sizes, making fair comparisons difficult and practitioners struggle with reproducibility. To address these situations, we propose a comprehensive library for few-shot learning (LibFewShot) by re-implementing eighteen state-of-the-art few-shot learning methods in a unified framework with the same single codebase in PyTorch. Furthermore, based on LibFewShot, we provide comprehensive evaluations on multiple benchmarks with various backbone architectures to evaluate common pitfalls and effects of different training tricks. In addition, with respect to the recent doubts on the necessity of meta- or episodic-training mechanism, our evaluation results confirm that such a mechanism is still necessary especially when combined with pre-training. We hope our work can not only lower the barriers for beginners to enter the area of few-shot learning but also elucidate the effects of nontrivial tricks to facilitate intrinsic research on few-shot learning. The source code is available from https://github.com/RL-VIG/LibFewShot.

cs.CV↗
arXiv2022

Explanation Guided Contrastive Learning for Sequential Recommendation

Recently, contrastive learning has been applied to the sequential recommendation task to address data sparsity caused by users with few item interactions and items with few user adoptions. Nevertheless, the existing contrastive learning-based methods fail to ensure that the positive (or negative) sequence obtained by some random augmentation (or sequence sampling) on a given anchor user sequence remains to be semantically similar (or different). When the positive and negative sequences turn out to be false positive and false negative respectively, it may lead to degraded recommendation performance. In this work, we address the above problem by proposing Explanation Guided Augmentations (EGA) and Explanation Guided Contrastive Learning for Sequential Recommendation (EC4SRec) model framework. The key idea behind EGA is to utilize explanation method(s) to determine items' importance in a user sequence and derive the positive and negative sequences accordingly. EC4SRec then combines both self-supervised and supervised contrastive learning over the positive and negative sequences generated by EGA operations to improve sequence representation learning for more accurate recommendation results. Extensive experiments on four real-world benchmark datasets demonstrate that EC4SRec outperforms the state-of-the-art sequential recommendation methods and two recent contrastive learning-based sequential recommendation methods, CL4SRec and DuoRec. Our experiments also show that EC4SRec can be easily adapted for different sequence encoder backbones (e.g., GRU4Rec and Caser), and improve their recommendation performance.

cs.IR↗
arXiv2022

Improving Compositional Generalization in Math Word Problem Solving

Compositional generalization refers to a model's capability to generalize to newly composed input data based on the data components observed during training. It has triggered a series of compositional generalization analysis on different tasks as generalization is an important aspect of language and problem solving skills. However, the similar discussion on math word problems (MWPs) is limited. In this manuscript, we study compositional generalization in MWP solving. Specifically, we first introduce a data splitting method to create compositional splits from existing MWP datasets. Meanwhile, we synthesize data to isolate the effect of compositions. To improve the compositional generalization in MWP solving, we propose an iterative data augmentation method that includes diverse compositional variation into training data and could collaborate with MWP methods. During the evaluation, we examine a set of methods and find all of them encounter severe performance loss on the evaluated datasets. We also find our data augmentation method could significantly improve the compositional generalization of general MWP methods. Code is available at https://github.com/demoleiwang/CGMWP.

cs.CL↗
arXiv2022

A Variance-Reduced Stochastic Gradient Tracking Algorithm for Decentralized Optimization with Orthogonality Constraints

Decentralized optimization with orthogonality constraints is found widely in scientific computing and data science. Since the orthogonality constraints are nonconvex, it is quite challenging to design efficient algorithms. Existing approaches leverage the geometric tools from Riemannian optimization to solve this problem at the cost of high sample and communication complexities. To relieve this difficulty, based on two novel techniques that can waive the orthogonality constraints, we propose a variance-reduced stochastic gradient tracking (VRSGT) algorithm with the convergence rate of $O(1 / k)$ to a stationary point. To the best of our knowledge, VRSGT is the first algorithm for decentralized optimization with orthogonality constraints that reduces both sampling and communication complexities simultaneously. In the numerical experiments, VRSGT has a promising performance in a real-world autonomous driving application.

math.OC↗
arXiv2022

Fusing Higher-order Features in Graph Neural Networks for Skeleton-based Action Recognition

Skeleton sequences are lightweight and compact, and thus are ideal candidates for action recognition on edge devices. Recent skeleton-based action recognition methods extract features from 3D joint coordinates as spatial-temporal cues, using these representations in a graph neural network for feature fusion to boost recognition performance. The use of first- and second-order features, i.e., joint and bone representations, has led to high accuracy. Nonetheless, many models are still confused by actions that have similar motion trajectories. To address these issues, we propose fusing higher-order features in the form of angular encoding into modern architectures to robustly capture the relationships between joints and body parts. This simple fusion with popular spatial-temporal graph neural networks achieves new state-of-the-art accuracy in two large benchmarks, including NTU60 and NTU120, while employing fewer parameters and reduced run time. Our source code is publicly available at: https://github.com/ZhenyueQin/Angular-Skeleton-Encoding.

cs.CV↗
arXiv2022

A Medical Semantic-Assisted Transformer for Radiographic Report Generation

Automated radiographic report generation is a challenging cross-domain task that aims to automatically generate accurate and semantic-coherence reports to describe medical images. Despite the recent progress in this field, there are still many challenges at least in the following aspects. First, radiographic images are very similar to each other, and thus it is difficult to capture the fine-grained visual differences using CNN as the visual feature extractor like many existing methods. Further, semantic information has been widely applied to boost the performance of generation tasks (e.g. image captioning), but existing methods often fail to provide effective medical semantic features. Toward solving those problems, in this paper, we propose a memory-augmented sparse attention block utilizing bilinear pooling to capture the higher-order interactions between the input fine-grained image features while producing sparse attention. Moreover, we introduce a novel Medical Concepts Generation Network (MCGN) to predict fine-grained semantic concepts and incorporate them into the report generation process as guidance. Our proposed method shows promising performance on the recently released largest benchmark MIMIC-CXR. It outperforms multiple state-of-the-art methods in image captioning and medical report generation.

cs.CV↗
arXiv2022

Private, Efficient, and Accurate: Protecting Models Trained by Multi-party Learning with Differential Privacy

Secure multi-party computation-based machine learning, referred to as MPL, has become an important technology to utilize data from multiple parties with privacy preservation. While MPL provides rigorous security guarantees for the computation process, the models trained by MPL are still vulnerable to attacks that solely depend on access to the models. Differential privacy could help to defend against such attacks. However, the accuracy loss brought by differential privacy and the huge communication overhead of secure multi-party computation protocols make it highly challenging to balance the 3-way trade-off between privacy, efficiency, and accuracy. In this paper, we are motivated to resolve the above issue by proposing a solution, referred to as PEA (Private, Efficient, Accurate), which consists of a secure DPSGD protocol and two optimization methods. First, we propose a secure DPSGD protocol to enforce DPSGD in secret sharing-based MPL frameworks. Second, to reduce the accuracy loss led by differential privacy noise and the huge communication overhead of MPL, we propose two optimization methods for the training process of MPL: (1) the data-independent feature extraction method, which aims to simplify the trained model structure; (2) the local data-based global model initialization method, which aims to speed up the convergence of the model training. We implement PEA in two open-source MPL frameworks: TF-Encrypted and Queqiao. The experimental results on various datasets demonstrate the efficiency and effectiveness of PEA. E.g. when $ε$ = 2, we can train a differentially private classification model with an accuracy of 88% for CIFAR-10 within 7 minutes under the LAN setting. This result significantly outperforms the one from CryptGPU, one SOTA MPL framework: it costs more than 16 hours to train a non-private deep neural network model on CIFAR-10 with the same accuracy.

cs.CR↗
arXiv2022

Scalable and Sparsity-Aware Privacy-Preserving K-means Clustering with Application to Fraud Detection

K-means is one of the most widely used clustering models in practice. Due to the problem of data isolation and the requirement for high model performance, how to jointly build practical and secure K-means for multiple parties has become an important topic for many applications in the industry. Existing work on this is mainly of two types. The first type has efficiency advantages, but information leakage raises potential privacy risks. The second type is provable secure but is inefficient and even helpless for the large-scale data sparsity scenario. In this paper, we propose a new framework for efficient sparsity-aware K-means with three characteristics. First, our framework is divided into a data-independent offline phase and a much faster online phase, and the offline phase allows to pre-compute almost all cryptographic operations. Second, we take advantage of the vectorization techniques in both online and offline phases. Third, we adopt a sparse matrix multiplication for the data sparsity scenario to improve efficiency further. We conduct comprehensive experiments on three synthetic datasets and deploy our model in a real-world fraud detection task. Our experimental results show that, compared with the state-of-the-art solution, our model achieves competitive performance in terms of both running time and communication size, especially on sparse datasets.

cs.LG↗
arXiv2022

Instance Image Retrieval by Learning Purely From Within the Dataset

Quality feature representation is key to instance image retrieval. To attain it, existing methods usually resort to a deep model pre-trained on benchmark datasets or even fine-tune the model with a task-dependent labelled auxiliary dataset. Although achieving promising results, this approach is restricted by two issues: 1) the domain gap between benchmark datasets and the dataset of a given retrieval task; 2) the required auxiliary dataset cannot be readily obtained. In light of this situation, this work looks into a different approach which has not been well investigated for instance image retrieval previously: {can we learn feature representation \textit{specific to} a given retrieval task in order to achieve excellent retrieval?} Our finding is encouraging. By adding an object proposal generator to generate image regions for self-supervised learning, the investigated approach can successfully learn feature representation specific to a given dataset for retrieval. This representation can be made even more effective by boosting it with image similarity information mined from the dataset. As experimentally validated, such a simple ``self-supervised learning + self-boosting'' approach can well compete with the relevant state-of-the-art retrieval methods. Ablation study is conducted to show the appealing properties of this approach and its limitation on generalisation across datasets.

cs.CV↗
arXiv2022

RDA: Reciprocal Distribution Alignment for Robust Semi-supervised Learning

In this work, we propose Reciprocal Distribution Alignment (RDA) to address semi-supervised learning (SSL), which is a hyperparameter-free framework that is independent of confidence threshold and works with both the matched (conventionally) and the mismatched class distributions. Distribution mismatch is an often overlooked but more general SSL scenario where the labeled and the unlabeled data do not fall into the identical class distribution. This may lead to the model not exploiting the labeled data reliably and drastically degrade the performance of SSL methods, which could not be rescued by the traditional distribution alignment. In RDA, we enforce a reciprocal alignment on the distributions of the predictions from two classifiers predicting pseudo-labels and complementary labels on the unlabeled data. These two distributions, carrying complementary information, could be utilized to regularize each other without any prior of class distribution. Moreover, we theoretically show that RDA maximizes the input-output mutual information. Our approach achieves promising performance in SSL under a variety of scenarios of mismatched distributions, as well as the conventional matched SSL setting. Our code is available at: https://github.com/NJUyued/RDA4RobustSSL.

cs.LG↗
arXiv2022

Projective-truncation-approximation study of the one-dimensional $ϕ^4$ lattice model

In this paper, we first develop the projective truncation approximation (PTA) in the Green's function equation of motion (EOM) formalism for classical statistical models. To implement PTA for a given Hamiltonian, we choose a set of basis variables and projectively truncate the hierarchical EOM. We apply PTA to the one-dimensional $ϕ^4$ lattice model. Phonon dispersion and static correlation functions are studied in detail. Using one- and two-dimensional bases, we obtain results identical to and beyond the quadratic variational approximation, respectively. In particular, we analyze the power-law temperature dependence of the static averages in the low- and high-temperature limits, and we give exact exponents.

cond-mat.stat-mech↗
arXiv2022

Machine Learning assisted excess noise suppression for continuous-variable quantum key distribution

Excess noise is a major obstacle to high-performance continuous-variable quantum key distribution (CVQKD), which is mainly derived from the amplitude attenuation and phase fluctuation of quantum signals caused by channel instability. Here, an excess noise suppression scheme based on equalization is proposed. In this scheme, the distorted signals can be corrected through equalization assisted by a neural network and pilot tone, relieving the pressure on the post-processing and eliminating the hardware cost. For a free-space channel with more intense fluctuation, a classification algorithm is added to classify the received variables, and then the distinctive equalization correction for different classes is carried out. The experimental results show that the scheme can suppress the excess noise to a lower level, and has a significant performance improvement. Moreover, the scheme also enables the system to cope with strong turbulence. It breaks the bottleneck of long-distance quantum communication and lays a foundation for the large-scale application of CVQKD.

quant-ph↗
arXiv2022

The Shigesada-Kawasaki-Teramoto cross-diffusion system beyond detailed balance

The existence of global weak solutions to the cross-diffusion model of Shigesada, Kawasaki, and Teramoto for an arbitrary number of species is proved. The model consists of strongly coupled parabolic equations for the population densities in a bounded domain with no-flux boundary conditions, and it describes the dynamics of the segregation of the population species. The diffusion matrix is neither symmetric nor positive semidefinite. A new logarithmic entropy allows for an improved condition on the coefficients of heavily nonsymmetric diffusion matrices, without imposing the detailed-balance condition that is often assumed in the literature. Furthermore, the large-time convergence of the solutions to the constant steady state is proved by using the relative entropy associated to the logarithmic entropy.

math.AP↗
arXiv2022

Revealing the CO2 emission reduction of ridesplitting and its determinants based on real-world data

Ridesplitting, which is a form of pooled ridesourcing service, has great potential to alleviate the negative impacts of ridesourcing on the environment. However, most existing studies only explored its theoretical environmental benefits based on optimization models and simulations. By contrast, this study aims to reveal the real-world emission reduction of ridesplitting and its determinants based on the observed data of ridesourcing in Chengdu, China. Integrating the trip data with the COPERT model, this study calculates the CO2 emissions of shared rides (ridesplitting) and their substituted single rides (regular ridesourcing) to estimate the CO2 emission reduction of each ridesplitting trip. The results show that not all ridesplitting trips reduce emissions from ridesourcing in the real world. The CO2 emission reduction rate of ridesplitting varies from trip to trip, averaging at 43.15g/km. Then, interpretable machine learning models, gradient boosting machines, are applied to explore the relationship between the CO2 emission reduction rate of ridesplitting and its determinants. Based on the SHapley Additive exPlanations (SHAP) method, the overlap rate and detour rate of shared rides are identified to be the most important factors that determine the CO2 emission reduction rate of ridesplitting. Increasing the overlap rate, the number of shared rides, average speed, and ride distance ratio while decreasing the detour rate, actual trip distance, and ride distance gap can increase the CO2 emission reduction rate of ridesplitting. In addition, nonlinear effects and interactions of the determinants are examined through the partial dependence plots. To sum up, this study provides a scientific method for the government and ridesourcing companies to better assess and optimize the environmental benefits of ridesplitting.

cs.LG↗
arXiv2022

A joint explanation of W-mass and muon g-2 in 2HDM

Since both $W$-mass and muon $g-2$ can be affected by the mass splittings among extra Higgs bosons $(H,~A,~H^\pm)$ in a 2HDM, we take a model with $μ$-$τ$ LFV interactions to examine the two anomalies reported respectively by CDF II and FNAL. We obtain the following observations: (i) Combined with theoretical constraints, the CDF $W$-mass measurement disfavors $H$ or $A$ to degenerate in mass with $H^\pm$, but allows $H$ and $A$ to degenerate. The mass splitting between $H^\pm$ and $H/A$ is required to be larger than 10 GeV. The $m_{H^\pm}$ and $m_{A}$ are favored to be smaller than 650 GeV for $m_H<120$ GeV, and allowed to have more large values with increasing of $m_H$. (ii) After imposing other relevant experimental constraints, there are parameter spaces that simultaneously satisfy (at $2σ$ level) the CDF $W$-mass, the FNAL muon $g-2$ and the data of lepton universality in $τ$ decays, but the mass splittings among extra Higgs bosons are strictly constrained.

hep-ph↗
arXiv2022

Attitude estimation from vector measurements: Necessary and sufficient conditions and convergent observer design

The paper addresses the problem of attitude estimation for rigid bodies using (possibly time-varying) vector measurements, for which we provide a necessary and sufficient condition of distinguishability. Such a condition is shown to be strictly weaker than those previously used for attitude observer design. Thereafter, we show that even for the single vector case the resulting condition is sufficient to design almost globally convergent attitude observers, and two explicit designs are obtained. To overcome the weak excitation issue, the first design employs to make full use of historical information, whereas the second scheme dynamically generates a virtual reference vector, which remains non-collinear to the given vector measurement. Simulation results illustrate the accurate estimation despite noisy measurements.

eess.SY↗