arXiv subjects

Zhe Li

Publications and source records attributed to Zhe Li.

At least 271 records · Page 15Linked to original sources

arXiv2021

Implicit Feature Alignment: Learn to Convert Text Recognizer to Text Spotter

Text recognition is a popular research subject with many associated challenges. Despite the considerable progress made in recent years, the text recognition task itself is still constrained to solve the problem of reading cropped line text images and serves as a subtask of optical character recognition (OCR) systems. As a result, the final text recognition result is limited by the performance of the text detector. In this paper, we propose a simple, elegant and effective paradigm called Implicit Feature Alignment (IFA), which can be easily integrated into current text recognizers, resulting in a novel inference mechanism called IFAinference. This enables an ordinary text recognizer to process multi-line text such that text detection can be completely freed. Specifically, we integrate IFA into the two most prevailing text recognition streams (attention-based and CTC-based) and propose attention-guided dense prediction (ADP) and Extended CTC (ExCTC). Furthermore, the Wasserstein-based Hollow Aggregation Cross-Entropy (WH-ACE) is proposed to suppress negative predictions to assist in training ADP and ExCTC. We experimentally demonstrate that IFA achieves state-of-the-art performance on end-to-end document recognition tasks while maintaining the fastest speed, and ADP and ExCTC complement each other on the perspective of different application scenarios. Code will be available at https://github.com/WangTianwei/Implicit-feature-alignment.

cs.CV↗

arXiv2021

Salient Positions based Attention Network for Image Classification

The self-attention mechanism has attracted wide publicity for its most important advantage of modeling long dependency, and its variations in computer vision tasks, the non-local block tries to model the global dependency of the input feature maps. Gathering global contextual information will inevitably need a tremendous amount of memory and computing resources, which has been extensively studied in the past several years. However, there is a further problem with the self-attention scheme: is all information gathered from the global scope helpful for the contextual modelling? To our knowledge, few studies have focused on the problem. Aimed at both questions this paper proposes the salient positions-based attention scheme SPANet, which is inspired by some interesting observations on the attention maps and affinity matrices generated in self-attention scheme. We believe these observations are beneficial for better understanding of the self-attention. SPANet uses the salient positions selection algorithm to select only a limited amount of salient points to attend in the attention map computing. This approach will not only spare a lot of memory and computing resources, but also try to distill the positive information from the transformation of the input feature maps. In the implementation, considering the feature maps with channel high dimensions, which are completely different from the general visual image, we take the squared power of the feature maps along the channel dimension as the saliency metric of the positions. In general, different from the non-local block method, SPANet models the contextual information using only the selected positions instead of all, along the channel dimension instead of space dimension. Our source code is available at https://github.com/likyoo/SPANet.

cs.CV↗

arXiv2021

Distributed Community Detection for Large Scale Networks Using Stochastic Block Model

With rapid developments of information and technology, large scale network data are ubiquitous. In this work we develop a distributed spectral clustering algorithm for community detection in large scale networks. To handle the problem, we distribute l pilot network nodes on the master server and the others on worker servers. A spectral clustering algorithm is first conducted on the master to select pseudo centers. The indexes of the pseudo centers are then broadcasted to workers to complete distributed community detection task using a SVD type algorithm. The proposed distributed algorithm has three merits. First, the communication cost is low since only the indexes of pseudo centers are communicated. Second, no further iteration algorithm is needed on workers and hence it does not suffer from problems as initialization and non-robustness. Third, both the computational complexity and the storage requirements are much lower compared to using the whole adjacency matrix. A Python package DCD (www.github.com/Ikerlz/dcd) is developed to implement the distributed algorithm for a Spark system. Theoretical properties are provided with respect to the estimation accuracy and mis-clustering rates. Lastly, the advantages of the proposed methodology are illustrated by experiments on a variety of synthetic and empirical datasets.

stat.ME↗

arXiv2021

CARLS: Cross-platform Asynchronous Representation Learning System

In this work, we propose CARLS, a novel framework for augmenting the capacity of existing deep learning frameworks by enabling multiple components -- model trainers, knowledge makers and knowledge banks -- to concertedly work together in an asynchronous fashion across hardware platforms. The proposed CARLS is particularly suitable for learning paradigms where model training benefits from additional knowledge inferred or discovered during training, such as node embeddings for graph neural networks or reliable pseudo labels from model predictions. We also describe three learning paradigms -- semi-supervised learning, curriculum learning and multimodal learning -- as examples that can be scaled up efficiently by CARLS. One version of CARLS has been open-sourced and available for download at: https://github.com/tensorflow/neural-structured-learning/tree/master/research/carls

cs.LG↗

arXiv2021

ConTNet: Why not use convolution and transformer at the same time?

Although convolutional networks (ConvNets) have enjoyed great success in computer vision (CV), it suffers from capturing global information crucial to dense prediction tasks such as object detection and segmentation. In this work, we innovatively propose ConTNet (ConvolutionTransformer Network), combining transformer with ConvNet architectures to provide large receptive fields. Unlike the recently-proposed transformer-based models (e.g., ViT, DeiT) that are sensitive to hyper-parameters and extremely dependent on a pile of data augmentations when trained from scratch on a midsize dataset (e.g., ImageNet1k), ConTNet can be optimized like normal ConvNets (e.g., ResNet) and preserve an outstanding robustness. It is also worth pointing that, given identical strong data augmentations, the performance improvement of ConTNet is more remarkable than that of ResNet. We present its superiority and effectiveness on image classification and downstream tasks. For example, our ConTNet achieves 81.8% top-1 accuracy on ImageNet which is the same as DeiT-B with less than 40% computational complexity. ConTNet-M also outperforms ResNet50 as the backbone of both Faster-RCNN (by 2.6%) and Mask-RCNN (by 3.2%) on COCO2017 dataset. We hope that ConTNet could serve as a useful backbone for CV tasks and bring new ideas for model design

cs.CV↗

arXiv2021

Class-Incremental Learning with Generative Classifiers

Incrementally training deep neural networks to recognize new classes is a challenging problem. Most existing class-incremental learning methods store data or use generative replay, both of which have drawbacks, while 'rehearsal-free' alternatives such as parameter regularization or bias-correction methods do not consistently achieve high performance. Here, we put forward a new strategy for class-incremental learning: generative classification. Rather than directly learning the conditional distribution p(y|x), our proposal is to learn the joint distribution p(x,y), factorized as p(x|y)p(y), and to perform classification using Bayes' rule. As a proof-of-principle, here we implement this strategy by training a variational autoencoder for each class to be learned and by using importance sampling to estimate the likelihoods p(x|y). This simple approach performs very well on a diverse set of continual learning benchmarks, outperforming generative replay and other existing baselines that do not store data.

cs.LG↗

arXiv2021

POSEFusion: Pose-guided Selective Fusion for Single-view Human Volumetric Capture

We propose POse-guided SElective Fusion (POSEFusion), a single-view human volumetric capture method that leverages tracking-based methods and tracking-free inference to achieve high-fidelity and dynamic 3D reconstruction. By contributing a novel reconstruction framework which contains pose-guided keyframe selection and robust implicit surface fusion, our method fully utilizes the advantages of both tracking-based methods and tracking-free inference methods, and finally enables the high-fidelity reconstruction of dynamic surface details even in the invisible regions. We formulate the keyframe selection as a dynamic programming problem to guarantee the temporal continuity of the reconstructed sequence. Moreover, the novel robust implicit surface fusion involves an adaptive blending weight to preserve high-fidelity surface details and an automatic collision handling method to deal with the potential self-collisions. Overall, our method enables high-fidelity and dynamic capture in both visible and invisible regions from a single RGBD camera, and the results and experiments show that our method outperforms state-of-the-art methods.

cs.CV↗

arXiv2021

Temporal Action Segmentation from Timestamp Supervision

Temporal action segmentation approaches have been very successful recently. However, annotating videos with frame-wise labels to train such models is very expensive and time consuming. While weakly supervised methods trained using only ordered action lists require less annotation effort, the performance is still worse than fully supervised approaches. In this paper, we propose to use timestamp supervision for the temporal action segmentation task. Timestamps require a comparable annotation effort to weakly supervised approaches, and yet provide a more supervisory signal. To demonstrate the effectiveness of timestamp supervision, we propose an approach to train a segmentation model using only timestamps annotations. Our approach uses the model output and the annotated timestamps to generate frame-wise labels by detecting the action changes. We further introduce a confidence loss that forces the predicted probabilities to monotonically decrease as the distance to the timestamps increases. This ensures that all and not only the most distinctive frames of an action are learned during training. The evaluation on four datasets shows that models trained with timestamps annotations achieve comparable performance to the fully supervised approaches.

cs.CV↗

arXiv2021

PolyFit: Polynomial-based Indexing Approach for Fast Approximate Range Aggregate Queries

Range aggregate queries find frequent application in data analytics. In some use cases, approximate results are preferred over accurate results if they can be computed rapidly and satisfy approximation guarantees. Inspired by a recent indexing approach, we provide means of representing a discrete point data set by continuous functions that can then serve as compact index structures. More specifically, we develop a polynomial-based indexing approach, called PolyFit, for processing approximate range aggregate queries. PolyFit is capable of supporting multiple types of range aggregate queries, including COUNT, SUM, MIN and MAX aggregates, with guaranteed absolute and relative error bounds. Experiment results show that PolyFit is faster and more accurate and compact than existing learned index structures.

cs.DB↗

arXiv2020

Improving Attention-Based Handwritten Mathematical Expression Recognition with Scale Augmentation and Drop Attention

Handwritten mathematical expression recognition (HMER) is an important research direction in handwriting recognition. The performance of HMER suffers from the two-dimensional structure of mathematical expressions (MEs). To address this issue, in this paper, we propose a high-performance HMER model with scale augmentation and drop attention. Specifically, tackling ME with unstable scale in both horizontal and vertical directions, scale augmentation improves the performance of the model on MEs of various scales. An attention-based encoder-decoder network is used for extracting features and generating predictions. In addition, drop attention is proposed to further improve performance when the attention distribution of the decoder is not precise. Compared with previous methods, our method achieves state-of-the-art performance on two public datasets of CROHME 2014 and CROHME 2016.

cs.CV↗

arXiv2020

The LBPM software package for simulating multiphase flow on digital images of porous rocks

Direct pore scale simulations of two-fluid flow on digital rock images provide a promising tool to understand the role of surface wetting phenomena on flow and transport in geologic reservoirs. We present computational protocols that mimic conventional special core analysis laboratory (SCAL) experiments, which are implemented within the open source LBPM software package. Protocols are described to simulate unsteady displacement, steady-state flow at fixed saturation, and to mimic centrifuge experiments. These methods can be used to infer relative permeability and capillary curves, and otherwise understand two-fluid flow behavior based on first principles. Morphological tools are applied to assess image resolution, establish initial conditions, and instantiate surface wetting maps based on the distribution of fluids. Internal analysis tools are described that measure essential aspects of two-fluid flow, including fluid connectivity and surface measures, which are used to track transient aspects of the flow behavior as they occur during simulation. Computationally efficient workflows are developed by combining these components with a two-fluid lattice Boltzmann model to define hybrid methods that can accelerate computations by using morphological tools to incrementally evolve the pore-scale fluid distribution. We show that the described methods can be applied to recover expected trends due to the surface wetting properties based on flow simulation in Benntheimer sandstone.

physics.geo-ph↗

arXiv2020

Robust 3D Self-portraits in Seconds

In this paper, we propose an efficient method for robust 3D self-portraits using a single RGBD camera. Benefiting from the proposed PIFusion and lightweight bundle adjustment algorithm, our method can generate detailed 3D self-portraits in seconds and shows the ability to handle subjects wearing extremely loose clothes. To achieve highly efficient and robust reconstruction, we propose PIFusion, which combines learning-based 3D recovery with volumetric non-rigid fusion to generate accurate sparse partial scans of the subject. Moreover, a non-rigid volumetric deformation method is proposed to continuously refine the learned shape prior. Finally, a lightweight bundle adjustment algorithm is proposed to guarantee that all the partial scans can not only "loop" with each other but also remain consistent with the selected live key observations. The results and experiments show that the proposed method achieves more robust and efficient 3D self-portraits compared with state-of-the-art methods.

cs.CV↗

arXiv2020

Tunable interlayer magnetism and band topology in van der Waals heterostructures of MnBi2Te4-family materials

Manipulating the interlayer magnetic coupling in van der Waals magnetic materials and heterostructures is the key to tailoring their magnetic and electronic properties for various electronic applications and fundamental studies in condensed matter physics. By utilizing the MnBi2Te4-family compounds and their heterostructures as a model system, we systematically studied the dependence of the sign and strength of interlayer magnetic coupling on constituent elements by using first-principles calculations. It was found that the coupling is a long-range superexchange interaction mediated by the chains of p orbitals between the magnetic atoms of neighboring septuple-layers. The interlayer exchange is always antiferromagnetic in the pure compounds, but can be tuned to ferromagnetic in some combinations of heterostructures, dictated by d orbital occupations. Strong interlayer magnetic coupling can be realized if the medial p electrons are delocalized and the d bands of magnetic atoms are near the Fermi level. The knowledge on the interlayer coupling mechanism enables us to engineer magnetic and topological properties of MnBi2Te4-family materials as well as many other insulating van der Waals magnetic materials and heterostructures.

cond-mat.mtrl-sci↗

arXiv2020

RCC-Dual-GAN: An Efficient Approach for Outlier Detection with Few Identified Anomalies

Outlier detection is an important task in data mining and many technologies have been explored in various applications. However, due to the default assumption that outliers are non-concentrated, unsupervised outlier detection may not correctly detect group anomalies with higher density levels. As for the supervised outlier detection, although high detection rates and optimal parameters can usually be achieved, obtaining sufficient and correct labels is a time-consuming task. To address these issues, we focus on semi-supervised outlier detection with few identified anomalies, in the hope of using limited labels to achieve high detection accuracy. First, we propose a novel detection model Dual-GAN, which can directly utilize the potential information in identified anomalies to detect discrete outliers and partially identified group anomalies simultaneously. And then, considering the instances with similar output values may not all be similar in a complex data structure, we replace the two MO-GAN components in Dual-GAN with the combination of RCC and M-GAN (RCC-Dual-GAN). In addition, to deal with the evaluation of Nash equilibrium and the selection of optimal model, two evaluation indicators are created and introduced into the two models to make the detection process more intelligent. Extensive experiments on both benchmark datasets and two practical tasks demonstrate that our proposed approaches (i.e., Dual-GAN and RCC-Dual-GAN) can significantly improve the accuracy of outlier detection even with only a few identified anomalies. Moreover, compared with the two MO-GAN components in Dual-GAN, the network structure combining RCC and M-GAN has greater stability in various situations.

cs.LG↗

arXiv2020

Long Short-Term Sample Distillation

In the past decade, there has been substantial progress at training increasingly deep neural networks. Recent advances within the teacher--student training paradigm have established that information about past training updates show promise as a source of guidance during subsequent training steps. Based on this notion, in this paper, we propose Long Short-Term Sample Distillation, a novel training policy that simultaneously leverages multiple phases of the previous training process to guide the later training updates to a neural network, while efficiently proceeding in just one single generation pass. With Long Short-Term Sample Distillation, the supervision signal for each sample is decomposed into two parts: a long-term signal and a short-term one. The long-term teacher draws on snapshots from several epochs ago in order to provide steadfast guidance and to guarantee teacher--student differences, while the short-term one yields more up-to-date cues with the goal of enabling higher-quality updates. Moreover, the teachers for each sample are unique, such that, overall, the model learns from a very diverse set of teachers. Comprehensive experimental results across a range of vision and NLP tasks demonstrate the effectiveness of this new training method.

cs.CV↗

arXiv2019

Learning From Brains How to Regularize Machines

Despite impressive performance on numerous visual tasks, Convolutional Neural Networks (CNNs) --- unlike brains --- are often highly sensitive to small perturbations of their input, e.g. adversarial noise leading to erroneous decisions. We propose to regularize CNNs using large-scale neuroscience data to learn more robust neural features in terms of representational similarity. We presented natural images to mice and measured the responses of thousands of neurons from cortical visual areas. Next, we denoised the notoriously variable neural activity using strong predictive models trained on this large corpus of responses from the mouse visual system, and calculated the representational similarity for millions of pairs of images from the model's predictions. We then used the neural representation similarity to regularize CNNs trained on image classification by penalizing intermediate representations that deviated from neural ones. This preserved performance of baseline models when classifying images under standard benchmarks, while maintaining substantially higher performance compared to baseline or control models when classifying noisy images. Moreover, the models regularized with cortical representations also improved model robustness in terms of adversarial attacks. This demonstrates that regularizing with neural data can be an effective tool to create an inductive bias towards more robust inference.

cs.LG↗

arXiv2019

Prior-aware Neural Network for Partially-Supervised Multi-Organ Segmentation

Accurate multi-organ abdominal CT segmentation is essential to many clinical applications such as computer-aided intervention. As data annotation requires massive human labor from experienced radiologists, it is common that training data are partially labeled, e.g., pancreas datasets only have the pancreas labeled while leaving the rest marked as background. However, these background labels can be misleading in multi-organ segmentation since the "background" usually contains some other organs of interest. To address the background ambiguity in these partially-labeled datasets, we propose Prior-aware Neural Network (PaNN) via explicitly incorporating anatomical priors on abdominal organ sizes, guiding the training process with domain-specific knowledge. More specifically, PaNN assumes that the average organ size distributions in the abdomen should approximate their empirical distributions, a prior statistics obtained from the fully-labeled dataset. As our training objective is difficult to be directly optimized using stochastic gradient descent [20], we propose to reformulate it in a min-max form and optimize it via the stochastic primal-dual gradient algorithm. PaNN achieves state-of-the-art performance on the MICCAI2015 challenge "Multi-Atlas Labeling Beyond the Cranial Vault", a competition on organ segmentation in the abdomen. We report an average Dice score of 84.97%, surpassing the prior art by a large margin of 3.27%.

cs.CV↗

arXiv2019

Identifying vital nodes based on reverse greedy method

The identification of vital nodes that maintain the network connectivity is a long-standing challenge in network science. In this paper, we propose a so-called reverse greedy method where the least important nodes are preferentially chosen to make the size of the largest component in the corresponding induced subgraph as small as possible. Accordingly, the nodes being chosen later are more important in maintaining the connectivity. Empirical analyses on ten real networks show that the reverse greedy method performs remarkably better than well-known state-of-the-art methods.

physics.soc-ph↗