Searcharxiv⌕ Search

arXiv subjects

Wen Wang

Publications and source records attributed to Wen Wang.

At least 145 records · Page 8Linked to original sources
arXiv2024

FedCache: A Knowledge Cache-driven Federated Learning Architecture for Personalized Edge Intelligence

Edge Intelligence (EI) allows Artificial Intelligence (AI) applications to run at the edge, where data analysis and decision-making can be performed in real-time and close to data sources. To protect data privacy and unify data silos among end devices in EI, Federated Learning (FL) is proposed for collaborative training of shared AI models across devices without compromising data privacy. However, the prevailing FL approaches cannot guarantee model generalization and adaptation on heterogeneous clients. Recently, Personalized Federated Learning (PFL) has drawn growing awareness in EI, as it enables a productive balance between local-specific training requirements inherent in devices and global-generalized optimization objectives for satisfactory performance. However, most existing PFL methods are based on the Parameters Interaction-based Architecture (PIA) represented by FedAvg, which causes unaffordable communication burdens due to large-scale parameters transmission between devices and the edge server. In contrast, Logits Interaction-based Architecture (LIA) allows to update model parameters with logits transfer and gains the advantages of communication lightweight and heterogeneous on-device model allowance compared to PIA. Nevertheless, previous LIA methods attempt to achieve satisfactory performance either relying on unrealistic public datasets or increasing communication overhead for additional information transmission other than logits. To tackle this dilemma, we propose a knowledge cache-driven PFL architecture, named FedCache, which reserves a knowledge cache on the server for fetching personalized knowledge from the samples with similar hashes to each given on-device sample. During the training phase, ensemble distillation is applied to on-device models for constructive optimization with personalized knowledge transferred from the server-side knowledge cache.

cs.DC↗
arXiv2024

Zero-Shot Video Editing Using Off-The-Shelf Image Diffusion Models

Large-scale text-to-image diffusion models achieve unprecedented success in image generation and editing. However, how to extend such success to video editing is unclear. Recent initial attempts at video editing require significant text-to-video data and computation resources for training, which is often not accessible. In this work, we propose vid2vid-zero, a simple yet effective method for zero-shot video editing. Our vid2vid-zero leverages off-the-shelf image diffusion models, and doesn't require training on any video. At the core of our method is a null-text inversion module for text-to-video alignment, a cross-frame modeling module for temporal consistency, and a spatial regularization module for fidelity to the original video. Without any training, we leverage the dynamic nature of the attention mechanism to enable bi-directional temporal modeling at test time. Experiments and analyses show promising results in editing attributes, subjects, places, etc., in real-world videos. Code is made available at \url{https://github.com/baaivision/vid2vid-zero}.

cs.CV↗
arXiv2023

Improving Communication Efficiency of Federated Distillation via Accumulating Local Updates

As an emerging federated learning paradigm, federated distillation enables communication-efficient model training by transmitting only small-scale knowledge during the learning process. To further improve the communication efficiency of federated distillation, we propose a novel technique, ALU, which accumulates multiple rounds of local updates before transferring the knowledge to the central server. ALU drastically decreases the frequency of communication in federated distillation, thereby significantly reducing the communication overhead during the training process. Empirical experiments demonstrate the substantial effect of ALU in improving the communication efficiency of federated distillation.

cs.LG↗
arXiv2023

GenDeF: Learning Generative Deformation Field for Video Generation

We offer a new perspective on approaching the task of video generation. Instead of directly synthesizing a sequence of frames, we propose to render a video by warping one static image with a generative deformation field (GenDeF). Such a pipeline enjoys three appealing advantages. First, we can sufficiently reuse a well-trained image generator to synthesize the static image (also called canonical image), alleviating the difficulty in producing a video and thereby resulting in better visual quality. Second, we can easily convert a deformation field to optical flows, making it possible to apply explicit structural regularizations for motion modeling, leading to temporally consistent results. Third, the disentanglement between content and motion allows users to process a synthesized video through processing its corresponding static image without any tuning, facilitating many applications like video editing, keypoint tracking, and video segmentation. Both qualitative and quantitative results on three common video generation benchmarks demonstrate the superiority of our GenDeF method.

cs.CV↗
arXiv2023

AutoStory: Generating Diverse Storytelling Images with Minimal Human Effort

Story visualization aims to generate a series of images that match the story described in texts, and it requires the generated images to satisfy high quality, alignment with the text description, and consistency in character identities. Given the complexity of story visualization, existing methods drastically simplify the problem by considering only a few specific characters and scenarios, or requiring the users to provide per-image control conditions such as sketches. However, these simplifications render these methods incompetent for real applications. To this end, we propose an automated story visualization system that can effectively generate diverse, high-quality, and consistent sets of story images, with minimal human interactions. Specifically, we utilize the comprehension and planning capabilities of large language models for layout planning, and then leverage large-scale text-to-image models to generate sophisticated story images based on the layout. We empirically find that sparse control conditions, such as bounding boxes, are suitable for layout planning, while dense control conditions, e.g., sketches and keypoints, are suitable for generating high-quality image content. To obtain the best of both worlds, we devise a dense condition generation module to transform simple bounding box layouts into sketch or keypoint control conditions for final image generation, which not only improves the image quality but also allows easy and intuitive user interactions. In addition, we propose a simple yet effective method to generate multi-view consistent character images, eliminating the reliance on human labor to collect or draw character images.

cs.CV↗
arXiv2023

CodeTransOcean: A Comprehensive Multilingual Benchmark for Code Translation

Recent code translation techniques exploit neural machine translation models to translate source code from one programming language to another to satisfy production compatibility or to improve efficiency of codebase maintenance. Most existing code translation datasets only focus on a single pair of popular programming languages. To advance research on code translation and meet diverse requirements of real-world applications, we construct CodeTransOcean, a large-scale comprehensive benchmark that supports the largest variety of programming languages for code translation. CodeTransOcean consists of three novel multilingual datasets, namely, MultilingualTrans supporting translations between multiple popular programming languages, NicheTrans for translating between niche programming languages and popular ones, and LLMTrans for evaluating executability of translated code by large language models (LLMs). CodeTransOcean also includes a novel cross-framework dataset, DLTrans, for translating deep learning code across different frameworks. We develop multilingual modeling approaches for code translation and demonstrate their great potential in improving the translation quality of both low-resource and high-resource language pairs and boosting the training efficiency. We also propose a novel evaluation metric Debugging Success Rate@K for program-level code translation. Last but not least, we evaluate LLM ChatGPT on our datasets and investigate its potential for fuzzy execution predictions. We build baselines for CodeTransOcean and analyze challenges of code translation for guiding future research. The CodeTransOcean datasets and code are publicly available at https://github.com/WeixiangYAN/CodeTransOcean.

cs.AI↗
arXiv2023

Ditto: A Simple and Efficient Approach to Improve Sentence Embeddings

Prior studies diagnose the anisotropy problem in sentence representations from pre-trained language models, e.g., BERT, without fine-tuning. Our analysis reveals that the sentence embeddings from BERT suffer from a bias towards uninformative words, limiting the performance in semantic textual similarity (STS) tasks. To address this bias, we propose a simple and efficient unsupervised approach, Diagonal Attention Pooling (Ditto), which weights words with model-based importance estimations and computes the weighted average of word representations from pre-trained models as sentence embeddings. Ditto can be easily applied to any pre-trained language model as a postprocessing operation. Compared to prior sentence embedding approaches, Ditto does not add parameters nor requires any learning. Empirical evaluations demonstrate that our proposed Ditto can alleviate the anisotropy problem and improve various pre-trained models on STS tasks.

cs.CL↗
arXiv2023

Improving Long Document Topic Segmentation Models With Enhanced Coherence Modeling

Topic segmentation is critical for obtaining structured documents and improving downstream tasks such as information retrieval. Due to its ability of automatically exploring clues of topic shift from abundant labeled data, recent supervised neural models have greatly promoted the development of long document topic segmentation, but leaving the deeper relationship between coherence and topic segmentation underexplored. Therefore, this paper enhances the ability of supervised models to capture coherence from both logical structure and semantic similarity perspectives to further improve the topic segmentation performance, proposing Topic-aware Sentence Structure Prediction (TSSP) and Contrastive Semantic Similarity Learning (CSSL). Specifically, the TSSP task is proposed to force the model to comprehend structural information by learning the original relations between adjacent sentences in a disarrayed document, which is constructed by jointly disrupting the original document at topic and sentence levels. Moreover, we utilize inter- and intra-topic information to construct contrastive samples and design the CSSL objective to ensure that the sentences representations in the same topic have higher similarity, while those in different topics are less similar. Extensive experiments show that the Longformer with our approach significantly outperforms old state-of-the-art (SOTA) methods. Our approach improve $F_1$ of old SOTA by 3.42 (73.74 -> 77.16) and reduces $P_k$ by 1.11 points (15.0 -> 13.89) on WIKI-727K and achieves an average relative reduction of 4.3% on $P_k$ on WikiSection. The average relative $P_k$ drop of 8.38% on two out-of-domain datasets also demonstrates the robustness of our approach.

cs.CL↗
arXiv2023

Performance Analysis and Optimization of Reconfigurable Multi-Functional Surface Assisted Wireless Communications

Although reconfigurable intelligent surfaces (RISs) can improve the performance of wireless networks by smartly reconfiguring the radio environment, existing passive RISs face two key challenges, i.e., double-fading attenuation and dependence on grid/battery. To address these challenges, this paper proposes a new RIS architecture, called multi-functional RIS (MF-RIS). Different from conventional reflecting-only RIS, the proposed MF-RIS is capable of supporting multiple functions with one surface, including signal reflection, amplification, and energy harvesting. As such, our MF-RIS is able to overcome the double-fading attenuation by harvesting energy from incident signals. Through theoretical analysis, we derive the achievable capacity of an MF-RIS-aided communication network. Compared to the capacity achieved by the existing self-sustainable RIS, we derive the number of reflective elements required for MF-RIS to outperform self-sustainable RIS. To realize a self-sustainable communication system, we investigate the use of MF-RIS in improving the sum-rate of multi-user wireless networks. Specifically, we solve a non-convex optimization problem by jointly designing the transmit beamforming and MF-RIS coefficients. As an extension, we investigate a resource allocation problem in a practical scenario with imperfect channel state information. By approximating the semi-infinite constraints with the S-procedure and the general sign-definiteness, we propose a robust beamforming scheme to combat the inevitable channel estimation errors. Finally, numerical results show that: 1) compared to the self-sustainable RIS, MF-RIS can strike a better balance between energy self-sustainability and throughput improvement; and 2) unlike reflecting-only RIS which can be deployed near the transmitter or receiver, MF-RIS should be deployed closer to the transmitter for higher spectrum efficiency.

eess.SP↗
arXiv2023

Multi-Functional Reconfigurable Intelligent Surface: System Modeling and Performance Optimization

In this paper, we propose and study a multi-functional reconfigurable intelligent surface (MF-RIS) architecture. In contrast to conventional single-functional RIS (SF-RIS) that only reflects signals, the proposed MF-RIS simultaneously supports multiple functions with one surface, including reflection, refraction, amplification, and energy harvesting of wireless signals. As such, the proposed MF-RIS is capable of significantly enhancing RIS signal coverage by amplifying the signal reflected/refracted by the RIS with the energy harvested. We present the signal model of the proposed MF-RIS, and formulate an optimization problem to maximize the sum-rate of multiple users in an MF-RIS-aided non-orthogonal multiple access network. We jointly optimize the transmit beamforming, power allocations as well as the operating modes and parameters for different elements of the MF-RIS and its deployment location, via an efficient iterative algorithm. Simulation results are provided which show significant performance gains of the MF-RIS over SF-RISs with only some of its functions available. Moreover, we demonstrate that there exists a fundamental trade-off between sum-rate maximization and harvested energy maximization. In contrast to SF-RISs which can be deployed near either the transmitter or receiver, the proposed MF-RIS should be deployed closer to the transmitter for maximizing its communication throughput with more energy harvested.

eess.SP↗
arXiv2023

Large Nonreciprocity of Shear-Horizontal Surface Acoustic Waves induced by Magnetoelastic Bilayers

We report large nonreciprocity in the transmission of shear-horizontal surface acoustic waves (SAWs) on LiTaO3 substrate coated with a FeCoSiB/NiFeCu magnetoelastic bilayer. The large difference in saturation magnetization of the two layers not only brings nonreciprocal spin waves (SWs), but also ensures the phonon-magnon (SAWs-SWs) coupling at relatively low wavenumbers. It is found that the angle between the magnetization and the wavevector play important roles in determining the strength of magnetoelastic coupling and nonreciprocity, simultaneously. A large nonreciprocal transmission of SAWs about 30 dB (i.e. 60 dB/mm) is demonstrated at 2.33 GHz. In addition, the dispersion relation between coupled SH-SAWs and nonreciprocal SWs is developed, which provide a good insight into the observed phenomena. Our results offer a convenient approach to implement nonreciprocal SAW isolators or circulators.

physics.app-ph↗
arXiv2023

NADiffuSE: Noise-aware Diffusion-based Model for Speech Enhancement

The goal of speech enhancement (SE) is to eliminate the background interference from the noisy speech signal. Generative models such as diffusion models (DM) have been applied to the task of SE because of better generalization in unseen noisy scenes. Technical routes for the DM-based SE methods can be summarized into three types: task-adapted diffusion process formulation, generator-plus-conditioner (GPC) structures and the multi-stage frameworks. We focus on the first two approaches, which are constructed under the GPC architecture and use the task-adapted diffusion process to better deal with the real noise. However, the performance of these SE models is limited by the following issues: (a) Non-Gaussian noise estimation in the task-adapted diffusion process. (b) Conditional domain bias caused by the weak conditioner design in the GPC structure. (c) Large amount of residual noise caused by unreasonable interpolation operations during inference. To solve the above problems, we propose a noise-aware diffusion-based SE model (NADiffuSE) to boost the SE performance, where the noise representation is extracted from the noisy speech signal and introduced as a global conditional information for estimating the non-Gaussian components. Furthermore, the anchor-based inference algorithm is employed to achieve a compromise between the speech distortion and noise residual. In order to mitigate the performance degradation caused by the conditional domain bias in the GPC framework, we investigate three model variants, all of which can be viewed as multi-stage SE based on the preprocessing networks for Mel spectrograms. Experimental results show that NADiffuSE outperforms other DM-based SE models under the GPC infrastructure. Audio samples are available at: https://square-of-w.github.io/NADiffuSE-demo/.

cs.SD↗
arXiv2023

Enhancing Multi-modal and Multi-hop Question Answering via Structured Knowledge and Unified Retrieval-Generation

Multi-modal multi-hop question answering involves answering a question by reasoning over multiple input sources from different modalities. Existing methods often retrieve evidences separately and then use a language model to generate an answer based on the retrieved evidences, and thus do not adequately connect candidates and are unable to model the interdependent relations during retrieval. Moreover, the pipelined approaches of retrieval and generation might result in poor generation performance when retrieval performance is low. To address these issues, we propose a Structured Knowledge and Unified Retrieval-Generation (SKURG) approach. SKURG employs an Entity-centered Fusion Encoder to align sources from different modalities using shared entities. It then uses a unified Retrieval-Generation Decoder to integrate intermediate retrieval results for answer generation and also adaptively determine the number of retrieval steps. Extensive experiments on two representative multi-modal multi-hop QA datasets MultimodalQA and WebQA demonstrate that SKURG outperforms the state-of-the-art models in both source retrieval and answer generation performance with fewer parameters. Our code is available at https://github.com/HITsz-TMG/SKURG.

cs.CL↗
arXiv2023

DePA: Improving Non-autoregressive Machine Translation with Dependency-Aware Decoder

Non-autoregressive machine translation (NAT) models have lower translation quality than autoregressive translation (AT) models because NAT decoders do not depend on previous target tokens in the decoder input. We propose a novel and general Dependency-Aware Decoder (DePA) to enhance target dependency modeling in the decoder of fully NAT models from two perspectives: decoder self-attention and decoder input. First, we propose an autoregressive forward-backward pre-training phase before NAT training, which enables the NAT decoder to gradually learn bidirectional target dependencies for the final NAT training. Second, we transform the decoder input from the source language representation space to the target language representation space through a novel attentive transformation process, which enables the decoder to better capture target dependencies. DePA can be applied to any fully NAT models. Extensive experiments show that DePA consistently improves highly competitive and state-of-the-art fully NAT models on widely used WMT and IWSLT benchmarks by up to 1.88 BLEU gain, while maintaining the inference latency comparable to other fully NAT models.

cs.CL↗
arXiv2023

Improving BERT with Hybrid Pooling Network and Drop Mask

Transformer-based pre-trained language models, such as BERT, achieve great success in various natural language understanding tasks. Prior research found that BERT captures a rich hierarchy of linguistic information at different layers. However, the vanilla BERT uses the same self-attention mechanism for each layer to model the different contextual features. In this paper, we propose a HybridBERT model which combines self-attention and pooling networks to encode different contextual features in each layer. Additionally, we propose a simple DropMask method to address the mismatch between pre-training and fine-tuning caused by excessive use of special mask tokens during Masked Language Modeling pre-training. Experiments show that HybridBERT outperforms BERT in pre-training with lower loss, faster training speed (8% relative), lower memory cost (13% relative), and also in transfer learning with 1.5% relative higher accuracies on downstream tasks. Additionally, DropMask improves accuracies of BERT on downstream tasks across various masking rates.

cs.CL↗
arXiv2023

CLAMP: Prompt-based Contrastive Learning for Connecting Language and Animal Pose

Animal pose estimation is challenging for existing image-based methods because of limited training data and large intra- and inter-species variances. Motivated by the progress of visual-language research, we propose that pre-trained language models (e.g., CLIP) can facilitate animal pose estimation by providing rich prior knowledge for describing animal keypoints in text. However, we found that building effective connections between pre-trained language models and visual animal keypoints is non-trivial since the gap between text-based descriptions and keypoint-based visual features about animal pose can be significant. To address this issue, we introduce a novel prompt-based Contrastive learning scheme for connecting Language and AniMal Pose (CLAMP) effectively. The CLAMP attempts to bridge the gap by adapting the text prompts to the animal keypoints during network training. The adaptation is decomposed into spatial-aware and feature-aware processes, and two novel contrastive losses are devised correspondingly. In practice, the CLAMP enables the first cross-modal animal pose estimation paradigm. Experimental results show that our method achieves state-of-the-art performance under the supervised, few-shot, and zero-shot settings, outperforming image-based methods by a large margin.

cs.CV↗
arXiv2023

DopplerBAS: Binaural Audio Synthesis Addressing Doppler Effect

Recently, binaural audio synthesis (BAS) has emerged as a promising research field for its applications in augmented and virtual realities. Binaural audio helps users orient themselves and establish immersion by providing the brain with interaural time differences reflecting spatial information. However, existing BAS methods are limited in terms of phase estimation, which is crucial for spatial hearing. In this paper, we propose the \textbf{DopplerBAS} method to explicitly address the Doppler effect of the moving sound source. Specifically, we calculate the radial relative velocity of the moving speaker in spherical coordinates, which further guides the synthesis of binaural audio. This simple method introduces no additional hyper-parameters and does not modify the loss functions, and is plug-and-play: it scales well to different types of backbones. DopperBAS distinctly improves the representative WarpNet and BinauralGrad backbones in the phase error metric and reaches a new state of the art (SOTA): 0.780 (versus the current SOTA 0.807). Experiments and ablation studies demonstrate the effectiveness of our method.

eess.AS↗
arXiv2023

Weighted Sampling for Masked Language Modeling

Masked Language Modeling (MLM) is widely used to pretrain language models. The standard random masking strategy in MLM causes the pre-trained language models (PLMs) to be biased toward high-frequency tokens. Representation learning of rare tokens is poor and PLMs have limited performance on downstream tasks. To alleviate this frequency bias issue, we propose two simple and effective Weighted Sampling strategies for masking tokens based on the token frequency and training loss. We apply these two strategies to BERT and obtain Weighted-Sampled BERT (WSBERT). Experiments on the Semantic Textual Similarity benchmark (STS) show that WSBERT significantly improves sentence embeddings over BERT. Combining WSBERT with calibration methods and prompt learning further improves sentence embeddings. We also investigate fine-tuning WSBERT on the GLUE benchmark and show that Weighted Sampling also improves the transfer learning capability of the backbone PLM. We further analyze and provide insights into how WSBERT improves token embeddings.

cs.CL↗