arXiv subjects

Lei Li

Publications and source records attributed to Lei Li.

At least 649 records · Page 36Linked to original sources

arXiv2021

Autocorrect in the Process of Translation -- Multi-task Learning Improves Dialogue Machine Translation

Automatic translation of dialogue texts is a much needed demand in many real life scenarios. However, the currently existing neural machine translation delivers unsatisfying results. In this paper, we conduct a deep analysis of a dialogue corpus and summarize three major issues on dialogue translation, including pronoun dropping (\droppro), punctuation dropping (\droppun), and typos (\typo). In response to these challenges, we propose a joint learning method to identify omission and typo, and utilize context to translate dialogue utterances. To properly evaluate the performance, we propose a manually annotated dataset with 1,931 Chinese-English parallel utterances from 300 dialogues as a benchmark testbed for dialogue translation. Our experiments show that the proposed method improves translation quality by 3.2 BLEU over the baselines. It also elevates the recovery rate of omitted pronouns from 26.09% to 47.16%. We will publish the code and dataset publicly at https://github.com/rgwt123/DialogueMT.

cs.CL↗

arXiv2021

On link deletion and point deletion in games on graphs

We discuss link and point deletion operators on graph games and provide a comparative logic-algorithmic study of the same. In particular, we focus on a popular notion of invariance in transition systems, namely, bisimulation, between the respective games on graphs. We present both logical and algorithmic analyses of the concepts so as to provide a more formal analysis of the natural connection between these two operators.

cs.LO↗

arXiv2021

Numerical stability of Grünwald-Letnikov method for time fractional delay differential equations

This paper is concerned with the numerical stability of time fractional delay differential equations (F-DDEs) based on Grünwald-Letnikov (GL) approximation (also called fraction backward Euler scheme) for the Caputo fractional derivative, in particular, the numerical stability region and the Mittag-Leffler stability. Using the boundary locus technique, we first derive the exact expression of the numerically stability region in the parameter plane, and show that the fractional backward Euler scheme based on GL scheme is not $τ(0)$-stable, which is different from the backward Euler scheme for integer DDE models. Secondly, we also prove the numerical Mittag-Leffler stability for the numerical solutions provided that the parameters fall into the numerical stability region, by employing the singularity analysis of generating function. Our results show that the numerical solutions of F-DDEs are completely different from the classical integer order DDEs, both in terms of $τ(0)$-stabililty and the long-time decay rate.

math.NA↗

arXiv2021

Generative Imagination Elevates Machine Translation

There are common semantics shared across text and images. Given a sentence in a source language, whether depicting the visual scene helps translation into a target language? Existing multimodal neural machine translation methods (MNMT) require triplets of bilingual sentence - image for training and tuples of source sentence - image for inference. In this paper, we propose ImagiT, a novel machine translation method via visual imagination. ImagiT first learns to generate visual representation from the source sentence, and then utilizes both source sentence and the "imagined representation" to produce a target translation. Unlike previous methods, it only needs the source sentence at the inference time. Experiments demonstrate that ImagiT benefits from visual imagination and significantly outperforms the text-only neural machine translation baselines. Further analysis reveals that the imagination process in ImagiT helps fill in missing information when performing the degradation strategy.

cs.CL↗

arXiv2021

Transformations preserving the norm of means between positive cones of general and commutative $C^*$-algebras

In this paper, we consider a (nonlinear) transformation $Φ$ of invertible positive elements in $C^*$-algebras which preserves the norm of any of the three fundamental means of positive elements; namely, $\|Φ(A)\mm Φ(B)\| = \|A\mm B\|$, where $\mm$ stands for the arithmetic mean $A\nabla B=(A+B)/2$, the geometric mean $A\#B=A^{1/2}(A^{-1/2}BA^{-1/2})^{1/2}A^{1/2}$, or the harmonic mean $A!B=2(A^{-1} + B^{-1})^{-1}$. Assuming that $Φ$ is surjective and preserves either the norm of the arithmetic mean or the norm of the geometric mean, we show that $Φ$ extends to a Jordan $*$-isomorphism between the underlying full algebras. If $Φ$ is surjective and preserves the norm of the harmonic mean, then we obtain the same conclusion in the special cases where the underlying algebras are $AW^*$-algebras or commutative $C^*$-algebras. In the commutative case, for a transformation $T: F(\mathrm{X})\subset C_0(\mathrm{X})_+\rightarrow C_0(\mathrm{Y})_+$, we can relax the surjectivity assumption and show that $T$ is a generalized composition operator if $T$ preserves the norm of the (arithmetic, geometric, harmonic, or in general any power) mean of any finite collection of positive functions, provided that the domain $F(\mathrm{X})$ contains sufficiently many elements to peak on compact $G_δ$ sets. When the image $T(F(\mathrm{X}))$ also contains sufficiently many elements to peak on compact $G_δ$ sets, $T$ extends to an algebra $*$-isomorphism between the underlying full function algebras.

math.OA↗

arXiv2021

Few-Shot Meta-Learning on Point Cloud for Semantic Segmentation

The promotion of construction robots can solve the problem of human resource shortage and improve the quality of decoration. To help the construction robots obtain environmental information, we need to use 3D point cloud, which is widely used in robotics, autonomous driving, and so on. With a good understanding of environmental information, construction robots can work better. However, the dynamic changes of 3D point cloud data may bring difficulties for construction robots to understand environmental information, such as when construction robots renovate houses. The paper proposes a semantic segmentation method for point cloud based on meta-learning. The method includes a basic learning module and a meta-learning module. The basic learning module is responsible for learning data features and evaluating the model, while the meta-learning module is responsible for updating the parameters of the model and improving the model generalization ability. In our work, we pioneered the method of producing datasets for meta-learning in 3D scenes, as well as demonstrated that the Model-Agnostic Meta-Learning (MAML) algorithm can be applied to process 3D point cloud data. At the same time, experiments show that our method can allow the model to be quickly applied to new environments with a few samples. Our method has important applications.

cs.RO↗

arXiv2021

Random Batch Methods for classical and quantum interacting particle systems and statistical samplings

We review the Random Batch Methods (RBM) for interacting particle systems consisting of $N$-particles, with $N$ being large. The computational cost of such systems is of $O(N^2)$, which is prohibitively expensive. The RBM methods use small but random batches so the computational cost is reduced, per time step, to $O(N)$. In this article we discuss these methods for both classical and quantum systems, the corresponding theory, and applications from molecular dynamics, statistical samplings, to agent-based models for collective behavior, and quantum Monte-Carlo methods.

math.NA↗

arXiv2021

"Listen, Understand and Translate": Triple Supervision Decouples End-to-end Speech-to-text Translation

An end-to-end speech-to-text translation (ST) takes audio in a source language and outputs the text in a target language. Existing methods are limited by the amount of parallel corpus. Can we build a system to fully utilize signals in a parallel ST corpus? We are inspired by human understanding system which is composed of auditory perception and cognitive processing. In this paper, we propose Listen-Understand-Translate, (LUT), a unified framework with triple supervision signals to decouple the end-to-end speech-to-text translation task. LUT is able to guide the acoustic encoder to extract as much information from the auditory input. In addition, LUT utilizes a pre-trained BERT model to enforce the upper encoder to produce as much semantic information as possible, without extra data. We perform experiments on a diverse set of speech translation benchmarks, including Librispeech English-French, IWSLT English-German and TED English-Chinese. Our results demonstrate LUT achieves the state-of-the-art performance, outperforming previous methods. The code is available at https://github.com/dqqcasia/st.

cs.CL↗

arXiv2021

Dense Contrastive Learning for Self-Supervised Visual Pre-Training

To date, most existing self-supervised learning methods are designed and optimized for image classification. These pre-trained models can be sub-optimal for dense prediction tasks due to the discrepancy between image-level prediction and pixel-level prediction. To fill this gap, we aim to design an effective, dense self-supervised learning method that directly works at the level of pixels (or local features) by taking into account the correspondence between local features. We present dense contrastive learning, which implements self-supervised learning by optimizing a pairwise contrastive (dis)similarity loss at the pixel level between two views of input images. Compared to the baseline method MoCo-v2, our method introduces negligible computation overhead (only <1% slower), but demonstrates consistently superior performance when transferring to downstream dense prediction tasks including object detection, semantic segmentation and instance segmentation; and outperforms the state-of-the-art methods by a large margin. Specifically, over the strong MoCo-v2 baseline, our method achieves significant improvements of 2.0% AP on PASCAL VOC object detection, 1.1% AP on COCO object detection, 0.9% AP on COCO instance segmentation, 3.0% mIoU on PASCAL VOC semantic segmentation and 1.8% mIoU on Cityscapes semantic segmentation. Code is available at: https://git.io/AdelaiDet

cs.CV↗

arXiv2021

Cross-lingual Supervision Improves Unsupervised Neural Machine Translation

Neural machine translation~(NMT) is ineffective for zero-resource languages. Recent works exploring the possibility of unsupervised neural machine translation (UNMT) with only monolingual data can achieve promising results. However, there are still big gaps between UNMT and NMT with parallel supervision. In this work, we introduce a multilingual unsupervised NMT (\method) framework to leverage weakly supervised signals from high-resource language pairs to zero-resource translation directions. More specifically, for unsupervised language pairs \texttt{En-De}, we can make full use of the information from parallel dataset \texttt{En-Fr} to jointly train the unsupervised translation directions all in one model. \method is based on multilingual models which require no changes to the standard unsupervised NMT. Empirical results demonstrate that \method significantly improves the translation quality by more than 3 BLEU score on six benchmark unsupervised translation directions.

cs.CL↗

arXiv2021

Scale-aware Automatic Augmentation for Object Detection

We propose Scale-aware AutoAug to learn data augmentation policies for object detection. We define a new scale-aware search space, where both image- and box-level augmentations are designed for maintaining scale invariance. Upon this search space, we propose a new search metric, termed Pareto Scale Balance, to facilitate search with high efficiency. In experiments, Scale-aware AutoAug yields significant and consistent improvement on various object detectors (e.g., RetinaNet, Faster R-CNN, Mask R-CNN, and FCOS), even compared with strong multi-scale training baselines. Our searched augmentation policies are transferable to other datasets and box-level tasks beyond object detection (e.g., instance segmentation and keypoint estimation) to improve performance. The search cost is much less than previous automated augmentation approaches for object detection. It is notable that our searched policies have meaningful patterns, which intuitively provide valuable insight for human data augmentation design. Code and models will be available at https://github.com/Jia-Research-Lab/SA-AutoAug.

cs.CV↗

arXiv2021

Progressive Domain Expansion Network for Single Domain Generalization

Single domain generalization is a challenging case of model generalization, where the models are trained on a single domain and tested on other unseen domains. A promising solution is to learn cross-domain invariant representations by expanding the coverage of the training domain. These methods have limited generalization performance gains in practical applications due to the lack of appropriate safety and effectiveness constraints. In this paper, we propose a novel learning framework called progressive domain expansion network (PDEN) for single domain generalization. The domain expansion subnetwork and representation learning subnetwork in PDEN mutually benefit from each other by joint learning. For the domain expansion subnetwork, multiple domains are progressively generated in order to simulate various photometric and geometric transforms in unseen domains. A series of strategies are introduced to guarantee the safety and effectiveness of the expanded domains. For the domain invariant representation learning subnetwork, contrastive learning is introduced to learn the domain invariant representation in which each class is well clustered so that a better decision boundary can be learned to improve it's generalization. Extensive experiments on classification and segmentation have shown that PDEN can achieve up to 15.28% improvement compared with the state-of-the-art single-domain generalization methods.

cs.CV↗

arXiv2021

Locate then Segment: A Strong Pipeline for Referring Image Segmentation

Referring image segmentation aims to segment the objects referred by a natural language expression. Previous methods usually focus on designing an implicit and recurrent feature interaction mechanism to fuse the visual-linguistic features to directly generate the final segmentation mask without explicitly modeling the localization information of the referent instances. To tackle these problems, we view this task from another perspective by decoupling it into a "Locate-Then-Segment" (LTS) scheme. Given a language expression, people generally first perform attention to the corresponding target image regions, then generate a fine segmentation mask about the object based on its context. The LTS first extracts and fuses both visual and textual features to get a cross-modal representation, then applies a cross-model interaction on the visual-textual features to locate the referred object with position prior, and finally generates the segmentation result with a light-weight segmentation network. Our LTS is simple but surprisingly effective. On three popular benchmark datasets, the LTS outperforms all the previous state-of-the-art methods by a large margin (e.g., +3.2% on RefCOCO+ and +3.4% on RefCOCOg). In addition, our model is more interpretable with explicitly locating the object, which is also proved by visualization experiments. We believe this framework is promising to serve as a strong baseline for referring image segmentation.

cs.CV↗

arXiv2021

Be Careful about Poisoned Word Embeddings: Exploring the Vulnerability of the Embedding Layers in NLP Models

Recent studies have revealed a security threat to natural language processing (NLP) models, called the Backdoor Attack. Victim models can maintain competitive performance on clean samples while behaving abnormally on samples with a specific trigger word inserted. Previous backdoor attacking methods usually assume that attackers have a certain degree of data knowledge, either the dataset which users would use or proxy datasets for a similar task, for implementing the data poisoning procedure. However, in this paper, we find that it is possible to hack the model in a data-free way by modifying one single word embedding vector, with almost no accuracy sacrificed on clean samples. Experimental results on sentiment analysis and sentence-pair classification tasks show that our method is more efficient and stealthier. We hope this work can raise the awareness of such a critical security risk hidden in the embedding layers of NLP models. Our code is available at https://github.com/lancopku/Embedding-Poisoning.

cs.CL↗

arXiv2021

Finding Sparse Structures for Domain Specific Neural Machine Translation

Neural machine translation often adopts the fine-tuning approach to adapt to specific domains. However, nonrestricted fine-tuning can easily degrade on the general domain and over-fit to the target domain. To mitigate the issue, we propose Prune-Tune, a novel domain adaptation method via gradual pruning. It learns tiny domain-specific sub-networks during fine-tuning on new domains. Prune-Tune alleviates the over-fitting and the degradation problem without model modification. Furthermore, Prune-Tune is able to sequentially learn a single network with multiple disjoint domain-specific sub-networks for multiple domains. Empirical experiment results show that Prune-Tune outperforms several strong competitors in the target domain test set without sacrificing the quality on the general domain in both single and multi-domain settings. The source code and data are available at https://github.com/ohlionel/Prune-Tune.

cs.CL↗

arXiv2021

MARS: Markov Molecular Sampling for Multi-objective Drug Discovery

Searching for novel molecules with desired chemical properties is crucial in drug discovery. Existing work focuses on developing neural models to generate either molecular sequences or chemical graphs. However, it remains a big challenge to find novel and diverse compounds satisfying several properties. In this paper, we propose MARS, a method for multi-objective drug molecule discovery. MARS is based on the idea of generating the chemical candidates by iteratively editing fragments of molecular graphs. To search for high-quality candidates, it employs Markov chain Monte Carlo sampling (MCMC) on molecules with an annealing scheme and an adaptive proposal. To further improve sample efficiency, MARS uses a graph neural network (GNN) to represent and select candidate edits, where the GNN is trained on-the-fly with samples from MCMC. Experiments show that MARS achieves state-of-the-art performance in various multi-objective settings where molecular bio-activity, drug-likeness, and synthesizability are considered. Remarkably, in the most challenging setting where all four objectives are simultaneously optimized, our approach outperforms previous methods significantly in comprehensive evaluations. The code is available at https://github.com/yutxie/mars.

q-bio.BM↗

arXiv2021

A random batch Ewald method for particle systems with Coulomb interactions

We develop a random batch Ewald (RBE) method for molecular dynamics simulations of particle systems with long-range Coulomb interactions, which achieves an $O(N)$ complexity in each step of simulating the $N$-body systems. The RBE method is based on the Ewald splitting for the Coulomb kernel with a random "mini-batch" type technique introduced to speed up the summation of the Fourier series for the long-range part of the splitting. Importance sampling is employed to reduce the induced force variance by taking advantage of the fast decay property of the Fourier coefficients. The stochastic approximation is unbiased with controlled variance. Analysis for bounded force fields gives some theoretic support of the method. Simulations of two typical problems of charged systems are presented to illustrate the accuracy and efficiency of the RBE method in comparison to the results from the Debye-Hückel theory and the classical Ewald summation, demonstrating that the proposed method has the attractiveness of being easy to implement with the linear scaling and is promising for many practical applications.

physics.comp-ph↗

arXiv2021

Social Group Query Based on Multi-fuzzy-constrained Strong Simulation

Traditional social group analysis mostly uses interaction models, event models, or other methods to identify and distinguish groups. This type of method can divide social participants into different groups based on their geographic location, social relationships, and/or related events. However, in some applications, it is necessary to make more specific restrictions on the members and the interaction between members of the group. Generally, graph pattern matching (GPM) is used to solve this problem. However, the existing GPM methods rarely consider the rich contextual information of nodes and edges to measure the credibility between members. In this paper, a social group query problem that needs to consider the trust between members of the group is proposed. To solve this problem, we propose a Strong Simulation GPM algorithm (NTSS) based on the exploration of pattern Node Topological ordered sequence. Aiming at the inefficiency of the NTSS algorithm when matching pattern graph with multiple nodes with zero in-degree and the problem of repeated calculation of matching edges shared by multiple matching subgraphs, two optimization strategies are proposed. Finally, we conduct verification experiments on the effectiveness and efficiency of the NTSS algorithm and the algorithms with the optimization strategies on four social network datasets in real applications. Experimental results show that the NTSS algorithm is significantly better than the existing multi-constrained GPM algorithm, and the NTSS_Inv_EdgC algorithm, which combines two optimization strategies, greatly improves the efficiency of the NTSS algorithm.

cs.SI↗