Searcharxiv⌕ Search

arXiv subjects

Xiaodong Wang

Publications and source records attributed to Xiaodong Wang.

At least 109 records · Page 6Linked to original sources
arXiv2023

BNS-Net: A Dual-channel Sarcasm Detection Method Considering Behavior-level and Sentence-level Conflicts

Sarcasm detection is a binary classification task that aims to determine whether a given utterance is sarcastic. Over the past decade, sarcasm detection has evolved from classical pattern recognition to deep learning approaches, where features such as user profile, punctuation and sentiment words have been commonly employed for sarcasm detection. In real-life sarcastic expressions, behaviors without explicit sentimental cues often serve as carriers of implicit sentimental meanings. Motivated by this observation, we proposed a dual-channel sarcasm detection model named BNS-Net. The model considers behavior and sentence conflicts in two channels. Channel 1: Behavior-level Conflict Channel reconstructs the text based on core verbs while leveraging the modified attention mechanism to highlight conflict information. Channel 2: Sentence-level Conflict Channel introduces external sentiment knowledge to segment the text into explicit and implicit sentences, capturing conflicts between them. To validate the effectiveness of BNS-Net, several comparative and ablation experiments are conducted on three public sarcasm datasets. The analysis and evaluation of experimental results demonstrate that the BNS-Net effectively identifies sarcasm in text and achieves the state-of-the-art performance.

cs.CL↗
arXiv2023

$C^r$-Chain closing lemma for certain partially hyperbolic diffeomorphisms

For every $r\in\mathbb{N}_{\geq 2}\cup\{\infty\}$, we prove a $C^r$-orbit connecting lemma for dynamically coherent and plaque expansive partially hyperbolic diffeomorphisms with 1-dimensional orientation preserving center bundle. To be precise, for such a diffeomorphism $f$, if a point $y$ is chain attainable from $x$ through pseudo-orbits, then for any neighborhood $U$ of $x$ and any neighborhood $V$ of $y$, there exist true orbits from $U$ to $V$ by arbitrarily $C^r$-small perturbations. As a consequence, we prove that for $C^r$-generic diffeomorphisms in this class, periodic points are dense in the chain recurrent set, and chain transitivity implies transitivity.

math.DS↗
arXiv2023

ORES: Open-vocabulary Responsible Visual Synthesis

Avoiding synthesizing specific visual concepts is an essential challenge in responsible visual synthesis. However, the visual concept that needs to be avoided for responsible visual synthesis tends to be diverse, depending on the region, context, and usage scenarios. In this work, we formalize a new task, Open-vocabulary Responsible Visual Synthesis (ORES), where the synthesis model is able to avoid forbidden visual concepts while allowing users to input any desired content. To address this problem, we present a Two-stage Intervention (TIN) framework. By introducing 1) rewriting with learnable instruction through a large-scale language model (LLM) and 2) synthesizing with prompt intervention on a diffusion synthesis model, it can effectively synthesize images avoiding any concepts but following the user's query as much as possible. To evaluate on ORES, we provide a publicly available dataset, baseline models, and benchmark. Experimental results demonstrate the effectiveness of our method in reducing risks of image generation. Our work highlights the potential of LLMs in responsible visual synthesis. Our code and dataset is public available.

cs.CV↗
arXiv2023

Generation of subcycle isolated attosecond pulses by pumping ionizing gating

We present a novel approach named as pumping ionizing gating (PIG) for the generation of isolated attosecond pulses (IAPs). In this regime, a short laser is used to ionize a pre-existing gas grating, creating a fast-extending plasma grating(FEPG) having an ionization front propagating with the velocity of light. A low-intensity long counterpropagating pump pulse is then reflected by a very narrow region of the ionization front, only where the Bragg conditions for resonant reflection is satisfied. Consequently, the pump reflection is confined within a sub-cycle region called PIG, and forms a wide-band coherent IAP in combination with the frequency up-conversion effect due to the plasma gradient. This approach results in a new scheme to generate IAPs fromlong picosecond pump pulses. Three-dimensional (3D) simulations show that a 1.6-ps, 1-μm pump pulse can be used to generate a 330 as laser pulse with a peak intensity approximately 33 times that of the pump and a conversion efficiency of around 0.1%.These results highlight the potential of the PIG method for generating IAPs with high conversion efficiency and peak intensity.

physics.optics↗
arXiv2023

On multifractal analysis and large deviations of singular-hyperbolic attractors

In this paper we study the multifractal analysis and large derivations for singular hyperbolic attractors, including the geometric Lorenz attractors. For each singular hyperbolic homoclinic class whose periodic orbits are all homoclinically related and such that the space of ergodic probability measures is connected, we prove that: (i) level sets associated to continuous observables are dense in the homoclinic class and satisfy a variational principle; (ii) irregular sets are either empty or are Baire generic and carry full topological entropy. The assumptions are satisfied by $C^1$-generic singular hyperbolic attractors and $C^r$-generic geometric Lorenz attractors $(r\ge 2)$. Finally we prove level-2 large deviations bounds for weak Gibbs measures, which provide a large deviations principle in the special case of Gibbs measures. The main technique we apply is the horseshoe approximation property.

math.DS↗
arXiv2023

Rateless Coded Blockchain for Dynamic IoT Networks

A key constraint that limits the implementation of blockchain in Internet of Things (IoT) is its large storage requirement resulting from the fact that each blockchain node has to store the entire blockchain. This increases the burden on blockchain nodes, and increases the communication overhead for new nodes joining the network since they have to copy the entire blockchain. In order to reduce storage requirements without compromising on system security and integrity, coded blockchains, based on error correcting codes with fixed rates and lengths, have been recently proposed. This approach, however, does not fit well with dynamic IoT networks in which nodes actively leave and join. In such dynamic blockchains, the existing coded blockchain approaches lead to high communication overheads for new joining nodes and may have high decoding failure probability. This paper proposes a rateless coded blockchain with coding parameters adjusted to network conditions. Our goals are to minimize both the storage requirement at each blockchain node and the communication overhead for each new joining node, subject to a target decoding failure probability. We evaluate the proposed scheme in the context of real-world Bitcoin blockchain and show that both storage and communication overhead are reduced by 99.6\% with a maximum $10^{-12}$ decoding failure probability.

cs.DC↗
arXiv2023

Radar-enabled ambient backscatter communications

In this work, we exploit the radar clutter (i.e., the ensemble of echoes generated by the terrain and/or the surrounding objects in response to the signal emitted by a radar transmitter) as a carrier signal to enable an ambient backscatter communication from a source (tag) to a destination (reader). The proposed idea relies on the fact that, since the radar excitation is periodic, the radar clutter is itself periodic over time scales shorter than the coherence time of the environment. Upon deriving a convenient signal model, we propose two encoding/decoding schemes that do not require any coordination with the radar transmitter or knowledge of the radar waveform. Different tradeoffs in terms of transmission rate and error probability can be obtained upon changing the control signal driving the tag switch or the adopted encoding rule; also, multiple tags can be accommodated with either a sourced or an unsourced multiple access strategy. Some illustrative examples are provided.

eess.SP↗
arXiv2023

NUWA-XL: Diffusion over Diffusion for eXtremely Long Video Generation

In this paper, we propose NUWA-XL, a novel Diffusion over Diffusion architecture for eXtremely Long video generation. Most current work generates long videos segment by segment sequentially, which normally leads to the gap between training on short videos and inferring long videos, and the sequential generation is inefficient. Instead, our approach adopts a ``coarse-to-fine'' process, in which the video can be generated in parallel at the same granularity. A global diffusion model is applied to generate the keyframes across the entire time range, and then local diffusion models recursively fill in the content between nearby frames. This simple yet effective strategy allows us to directly train on long videos (3376 frames) to reduce the training-inference gap, and makes it possible to generate all segments in parallel. To evaluate our model, we build FlintstonesHD dataset, a new benchmark for long video generation. Experiments show that our model not only generates high-quality long videos with both global and local coherence, but also decreases the average inference time from 7.55min to 26s (by 94.26\%) at the same hardware setting when generating 1024 frames. The homepage link is \url{https://msra-nuwa.azurewebsites.net/}

cs.CV↗
arXiv2023

Visual ChatGPT: Talking, Drawing and Editing with Visual Foundation Models

ChatGPT is attracting a cross-field interest as it provides a language interface with remarkable conversational competency and reasoning capabilities across many domains. However, since ChatGPT is trained with languages, it is currently not capable of processing or generating images from the visual world. At the same time, Visual Foundation Models, such as Visual Transformers or Stable Diffusion, although showing great visual understanding and generation capabilities, they are only experts on specific tasks with one-round fixed inputs and outputs. To this end, We build a system called \textbf{Visual ChatGPT}, incorporating different Visual Foundation Models, to enable the user to interact with ChatGPT by 1) sending and receiving not only languages but also images 2) providing complex visual questions or visual editing instructions that require the collaboration of multiple AI models with multi-steps. 3) providing feedback and asking for corrected results. We design a series of prompts to inject the visual model information into ChatGPT, considering models of multiple inputs/outputs and models that require visual feedback. Experiments show that Visual ChatGPT opens the door to investigating the visual roles of ChatGPT with the help of Visual Foundation Models. Our system is publicly available at \url{https://github.com/microsoft/visual-chatgpt}.

cs.CV↗
arXiv2023

Software-Hardware Co-design for Fast and Scalable Training of Deep Learning Recommendation Models

Deep learning recommendation models (DLRMs) are used across many business-critical services at Facebook and are the single largest AI application in terms of infrastructure demand in its data-centers. In this paper we discuss the SW/HW co-designed solution for high-performance distributed training of large-scale DLRMs. We introduce a high-performance scalable software stack based on PyTorch and pair it with the new evolution of Zion platform, namely ZionEX. We demonstrate the capability to train very large DLRMs with up to 12 Trillion parameters and show that we can attain 40X speedup in terms of time to solution over previous systems. We achieve this by (i) designing the ZionEX platform with dedicated scale-out network, provisioned with high bandwidth, optimal topology and efficient transport (ii) implementing an optimized PyTorch-based training stack supporting both model and data parallelism (iii) developing sharding algorithms capable of hierarchical partitioning of the embedding tables along row, column dimensions and load balancing them across multiple workers; (iv) adding high-performance core operators while retaining flexibility to support optimizers with fully deterministic updates (v) leveraging reduced precision communications, multi-level memory hierarchy (HBM+DDR+SSD) and pipelining. Furthermore, we develop and briefly comment on distributed data ingestion and other supporting services that are required for the robust and efficient end-to-end training in production environments.

cs.DC↗
arXiv2023

Learning 3D Photography Videos via Self-supervised Diffusion on Single Images

3D photography renders a static image into a video with appealing 3D visual effects. Existing approaches typically first conduct monocular depth estimation, then render the input frame to subsequent frames with various viewpoints, and finally use an inpainting model to fill those missing/occluded regions. The inpainting model plays a crucial role in rendering quality, but it is normally trained on out-of-domain data. To reduce the training and inference gap, we propose a novel self-supervised diffusion model as the inpainting module. Given a single input image, we automatically construct a training pair of the masked occluded image and the ground-truth image with random cycle-rendering. The constructed training samples are closely aligned to the testing instances, without the need of data annotation. To make full use of the masked images, we design a Masked Enhanced Block (MEB), which can be easily plugged into the UNet and enhance the semantic conditions. Towards real-world animation, we present a novel task: out-animation, which extends the space and time of input objects. Extensive experiments on real datasets show that our method achieves competitive results with existing SOTA methods.

cs.CV↗
arXiv2023

A Spectral Method for Depth-Separated Solution of a Wavenumber Integration Model in Horizontally Stratified Fluid Acoustic Waveguides

The wavenumber integration model is considered to be the most accurate algorithm for arbitrary horizontally stratified media in computational ocean acoustics. In contrast to the normal mode approach, it considers not only the discrete wavenumber spectrum but also the continuous spectrum components, eliminating errors in the model approximation for horizontally stratified media. Traditionally, analytical and semianalytical methods have been used to solve the depth-separated wave equation in the wavenumber integration method, and numerical solutions have generally focused on the finite difference method and the finite element method. In this paper, an algorithm for solving the depth equation using the Chebyshev--Tau spectral method combined with a domain decomposition strategy is proposed, and a numerical program named WISpec is developed accordingly. The proposed algorithm can simulate not only the sound field excited by a point source but also the sound field excited by an infinite line source. The key idea of the algorithm is to first discretize the depth equations for each layer via the Chebyshev--Tau spectral method and then solve the equations for each layer simultaneously by incorporating boundary and interface conditions. Several representative numerical experiments are devised to test the accuracy and speed of WISpec. The high consistency of the results of different software programs running under the same configuration proves that the numerical algorithm proposed in this paper is accurate, reliable and numerically stable.

math.NA↗
arXiv2023

RD-NAS: Enhancing One-shot Supernet Ranking Ability via Ranking Distillation from Zero-cost Proxies

Neural architecture search (NAS) has made tremendous progress in the automatic design of effective neural network structures but suffers from a heavy computational burden. One-shot NAS significantly alleviates the burden through weight sharing and improves computational efficiency. Zero-shot NAS further reduces the cost by predicting the performance of the network from its initial state, which conducts no training. Both methods aim to distinguish between "good" and "bad" architectures, i.e., ranking consistency of predicted and true performance. In this paper, we propose Ranking Distillation one-shot NAS (RD-NAS) to enhance ranking consistency, which utilizes zero-cost proxies as the cheap teacher and adopts the margin ranking loss to distill the ranking knowledge. Specifically, we propose a margin subnet sampler to distill the ranking knowledge from zero-shot NAS to one-shot NAS by introducing Group distance as margin. Our evaluation of the NAS-Bench-201 and ResNet-based search space demonstrates that RD-NAS achieve 10.7\% and 9.65\% improvements in ranking ability, respectively. Our codes are available at https://github.com/pprp/CVPR2022-NAS-competition-Track1-3th-solution

cs.CV↗
arXiv2023

Progressive Meta-Pooling Learning for Lightweight Image Classification Model

Practical networks for edge devices adopt shallow depth and small convolutional kernels to save memory and computational cost, which leads to a restricted receptive field. Conventional efficient learning methods focus on lightweight convolution designs, ignoring the role of the receptive field in neural network design. In this paper, we propose the Meta-Pooling framework to make the receptive field learnable for a lightweight network, which consists of parameterized pooling-based operations. Specifically, we introduce a parameterized spatial enhancer, which is composed of pooling operations to provide versatile receptive fields for each layer of a lightweight model. Then, we present a Progressive Meta-Pooling Learning (PMPL) strategy for the parameterized spatial enhancer to acquire a suitable receptive field size. The results on the ImageNet dataset demonstrate that MobileNetV2 using Meta-Pooling achieves top1 accuracy of 74.6\%, which outperforms MobileNetV2 by 2.3\%.

cs.CV↗
arXiv2022

Application of a Spectral Method to Simulate Quasi-Three-Dimensional Underwater Acoustic Fields

The calculation of a three-dimensional underwater acoustic field has always been a key problem in computational ocean acoustics. Traditionally, this solution is usually obtained by directly solving the acoustic Helmholtz equation using a finite difference or finite element algorithm. Solving the three-dimensional Helmholtz equation directly is computationally expensive. For quasi-three-dimensional problems, the Helmholtz equation can be processed by the integral transformation approach, which can greatly reduce the computational cost. In this paper, a numerical algorithm for a quasi-three-dimensional sound field that combines an integral transformation technique, stepwise coupled modes and a spectral method is designed. The quasi-three-dimensional problem is transformed into a two-dimensional problem using an integral transformation strategy. A stepwise approximation is then used to discretize the range dependence of the two-dimensional problem; this approximation is essentially a physical discretization that further reduces the range-dependent two-dimensional problem to a one-dimensional problem. Finally, the Chebyshev--Tau spectral method is employed to accurately solve the one-dimensional problem. We provide the corresponding numerical program SPEC3D for the proposed algorithm and describe several representative numerical examples. In the numerical experiments, the consistency between SPEC3D and the analytical solution/high-precision finite difference program COACH verifies the reliability and capability of the proposed algorithm. A comparison of running times illustrates that the algorithm proposed in this paper is significantly faster than the full three-dimensional algorithm in terms of computational speed.

physics.comp-ph↗
arXiv2022

Unsupervised Knowledge Graph Construction and Event-centric Knowledge Infusion for Scientific NLI

With the advance of natural language inference (NLI), a rising demand for NLI is to handle scientific texts. Existing methods depend on pre-trained models (PTM) which lack domain-specific knowledge. To tackle this drawback, we introduce a scientific knowledge graph to generalize PTM to scientific domain. However, existing knowledge graph construction approaches suffer from some drawbacks, i.e., expensive labeled data, failure to apply in other domains, long inference time and difficulty extending to large corpora. Therefore, we propose an unsupervised knowledge graph construction method to build a scientific knowledge graph (SKG) without any labeled data. Moreover, to alleviate noise effect from SKG and complement knowledge in sentences better, we propose an event-centric knowledge infusion method to integrate external knowledge into each event that is a fine-grained semantic unit in sentences. Experimental results show that our method achieves state-of-the-art performance and the effectiveness and reliability of SKG.

cs.CL↗
arXiv2022

A Chebyshev-Tau Spectral Method for Coupled Modes of Underwater Sound Propagation in Range-Dependent Ocean Environments

The stepwise coupled-mode model is a classic approach for solving range-dependent sound propagation problems. Existing coupled-mode programs have disadvantages such as high computational cost, weak adaptability to complex ocean environments and numerical instability. In this paper, a new algorithm is designed that uses an improved range normalization and global matrix approach to address range dependence in ocean environments. Due to its high accuracy in solving differential equations, the spectral method has recently been applied to range-independent normal modes and has achieved remarkable results. This algorithm uses the Chebyshev--Tau spectral method to solve for the eigenmodes in the range-independent segments. The main steps of the algorithm are parallelized, so OpenMP multithreading technology is also applied for further acceleration. Based on this algorithm, an efficient program is developed, and numerical simulations verify that this algorithm is reliable, accurate and capable. Compared with the existing coupled-mode programs, the newly developed program is more stable and efficient at comparable accuracies and can solve waveguides in more complex and realistic ocean environments.

physics.comp-ph↗